QuanterLab produced this study: it wasn’t written up afterwards. Registered hypothesis and search record in Appendix A2.
As seen on Quantocracy
All research
QuanterLab · Research

Dalio’s All Weather, Constructed and Then Tested Against Inverse Volatility, Equal Risk Contribution, Its Levered Version and a Plain 60/40

The retail articles print All Seasons and call it All Weather. We built the levered one Bridgewater describes and charged it the bill it would have paid each January.
By Serhat Girgin
Universe · All Seasons sleeves (fixed list) (membership resolution not recorded)
Method · Comparative: All Seasons static recipe vs Equal risk contribution, unlevered
Manipulated variable ·
The manipulated variable is the WEIGHTING RULE on an identical book: arm A holds the declared All Seasons vector (30 equity, 40 long Treasury, 15 intermediate Treasury, 7.5 gold, 7.5 broad commodity) and arm B holds equal risk contribution weights re-estimated point-in-time from the trailing 252 tra… (full sealed statement)The manipulated variable is the WEIGHTING RULE on an identical book: arm A holds the declared All Seasons vector (30 equity, 40 long Treasury, 15 intermediate Treasury, 7.5 gold, 7.5 broad commodity) and arm B holds equal risk contribution weights re-estimated point-in-time from the trailing 252 trading days of total-return covariance. Everything else is registered identical: the same five sleeves (SPY, TLT, IEF, GLD, DBC), quarterly rebalance with re-estimation at each rebalance, one-year windows, total return with dividends from the payment record, ten basis points per one-way traded dollar, SPY total return as the platform ruler. This walk is one of a DECLARED FAMILY OF FOUR registered books on the same sleeves - the static recipe, inverse volatility, unlevered equal risk contribution, and equal risk contribution levered ex ante to the trailing volatility of a 60/40 SPY/AGG reference and financed at the point-in-time three-month bill rate plus 50 basis points on the borrowed excess - with a 60/40 benchmark carried as the author benchmark, not as a searched rule. The five sleeves are among the most liquid funds in existence: any pricing refusal is a data defect and halts the walk rather than shrinking a book. No result of any arm was known when this text was registered.
Step size · 1 year per forward window
Out-of-sample span · 2008-01-02 → 2025-12-31
Compiled · August 30, 2026
Search family · the All Weather family: four weighting rules on one fixed five-sleeve book across three registered walks - the static All Seasons recipe and unlevered equal risk contribution as the two-arm comparative, inverse volatility and vol-targeted levered equal risk contribution as single-arm walks - with the 60/40 carried as the author benchmark, not a searched rule (N = 4, every member reported)
Abstract

People usually take All Weather for the fixed allocation Dalio gave Tony Robbins, which is the one every retail article prints, but that one is All Seasons and it carries no borrowing at all. All Weather proper is risk parity borrowed up until the book carries a 60/40's volatility, which means the borrowing is the strategy itself. We built that one and charged it the three-month Treasury bill that existed at each January anchor plus fifty basis points.

All Weather cleared a 60/40 by 0.11 points a year before financing and trailed it by 0.82 after, at 11.6 percent volatility against the 60/40's 11.9. The bill came to 0.93 points a year, which is larger than the gap, so a test that charges nothing for the borrowing reaches the opposite answer.

Two windows carry the whole deficit. The 2009 window cost 17.2 points against the 60/40 and the 2013 window cost 21.2, so dropping either one reverses the sign, and dropping both leaves the levered book 22.9 points ahead across the sixteen that remain. Both were years when equities ran double digits and the bond side made nothing. What this record supports is that the levered book did not clear its benchmark here, and not that it reliably loses.

The retail recipe won its own contest against unlevered risk parity, 6.19 percent a year against 5.08, and it holds the highest Sharpe in the study at 0.76 over zero and 0.60 over bills. That win is one long-bond block. The solver put 52.5 percent of the book in Treasuries against the recipe's fixed 55, so the same bet by weight, but 17.0 percent in the long bond against the recipe's fixed 40, and 77 percent of the eighteen-year margin sits in the 2014 and 2020 windows, both long-Treasury rallies.

Every unlevered book in the family set its worst drawdown in October 2022 rather than in 2008. In 2022 long Treasuries fell with equities, so both sides of the book lost together, and the family built to survive 2008 took its worst loss from bonds fourteen years later. Every walk is a registered study on this platform, and the circuits, per-window books and frozen records are linked below.

1  Methodology

The five sleeves and the clock. SPY for equities, TLT for long Treasuries, IEF for intermediate Treasuries, GLD for gold, DBC for broad commodities, which are the standard fund wrappers for the asset classes the recipe names. DBC is the youngest of them and its first bar is February 2006, so with a 252-day estimation window it dates the whole walk: eighteen one-year windows anchored each January from 2008 to 2025, which opens the record on the financial crisis and holds 2022 in window fifteen. All books rebalance quarterly and re-estimate their weights point-in-time at each rebalance from data available that day. Total return throughout, with dividends credited from the payment record, and ten basis points per one-way traded dollar. A 60/40 SPY/AGG rebalanced monthly on the same total-return series is the benchmark, and it is the book this study scores the leverage against. The platform normally also draws its own equity reference in grey, and this study switches it off, because a book that is entirely equities is a different risk class from the five diversified ones here and only invites the question of why every one of them trails it.

The four mechanisms, plainly. The RECIPE holds fixed percentages, 30 equity, 40 long Treasury, 15 intermediate Treasury, 7.5 gold and 7.5 commodity, and trades once a quarter back to them. It uses no data at all, because the weights are the opinion. INVERSE VOLATILITY measures each sleeve's own volatility over the trailing year and gives calmer sleeves more money, one over sigma, normalized, which means it sees each sleeve alone and nothing about how they move together. EQUAL RISK CONTRIBUTION is inverse volatility computed with the covariances included, so it solves for the weights at which every sleeve contributes exactly one fifth of the whole book's variance, and a sleeve that diversifies the others is given more weight for it. This is risk parity in the proper sense. The LEVERED book takes those same equal risk contribution weights and adds one step, which is to compare the book's trailing volatility to a 60/40's at the anchor, ex ante, and scale the whole book by that ratio, borrowing when risk parity runs quieter than the target and clamped between 0.5 and 2.0 times.

Leverage is financed here. The borrowed excess pays the point-in-time three-month Treasury bill plus 50 basis points, with the bill resolved at each window's anchor and held for that window. Bills were near five percent in 2008 and again from 2023, so a backtest that charges a flat spread, or nothing, flatters every levered year in exactly those windows, and anyone who has priced a levered book knows it. The financing shows up as a per-window drag between 0.10 percent of capital in 2014 and 3.48 percent in 2008, and the ledger figure prints all eighteen. A flat spread set at the sample-average bill would charge 17.36 points of capital across the walk against the 17.44 the point-in-time bill charged, so the totals land within 0.08 points of each other. What the point-in-time bill changes is which windows pay it, because a flat charge would take about 0.96 percent in every window while the real one took 0.10 in 2014 and 3.48 in 2008. In a year when bills rise sharply after January, and 2022 above all, this convention undercharges the borrowing, which flatters the levered book rather than the benchmark.

Four conventions the comparisons rest on, stated once. Gross exposure is set once a year at the anchor while the weights re-estimate quarterly, so the leverage decision is annual and the book underneath it is not. The 60/40 reference carries no trading cost while all four books pay ten basis points one-way, and its monthly rebalance turns over about 13 percent of the book a year, which at the same ten basis points is 0.41 points of capital across the whole walk against a gap of 0.82 points a year. Sharpe is measured over zero rather than over bills, uniformly for every book and both rulers, which flatters unlevered risk parity against the levered book by roughly 0.07 and the levered book against the 60/40 by about 0.005. Table 1 computes volatility, Sharpe and drawdown on each book's own stitched daily series, which is the convention every Sharpe in this paper quotes, while the paired comparison in section 2.4 runs on the days both arms share and its per-window returns sit about a tenth of a point above the panel's.

The family. Four weighting rules on one fixed book across three registered walks, which are the recipe against equal risk contribution as this paper's two-arm comparative, then inverse volatility and the levered book as single-arm walks, all declared as one search family of four. The 60/40 is the author benchmark and not a searched rule. The five sleeves are among the most liquid funds in existence, and the walks completed with every sleeve priced in every window and an empty census.

2  Results

2.1  Headline

All Seasons static recipe, pooled Sharpe
0.78
4512 OOS bars
Equal risk contribution, unlevered, pooled Sharpe
0.75
4512 OOS bars
The statistic this paper stands on
Before financing the levered book made 7.95 percent a year against the 60/40’s 7.84, and it carried 11.6 percent volatility against the 60/40’s 11.9, so it cleared the benchmark on the same risk. After financing it made 7.02, which puts it 0.82 behind, and the bill itself came to 0.93 points a year. The financing is bigger than the gap.
60/40 SPY/AGG · study benchmark, Sharpe
0.69
7.84%/yr · worst drawdown -34.8% · computed
show or hide a line:
Out-of-sample equity: normalised growth (1.00x = break even)0.39x2.29x4.18x200820102012201420162018202020222024
Figure 1. Both arms stitched through the identical windows,  All Seasons static recipe (+194.6%),  Equal risk contribution, unlevered (+144.2%),  60/40 SPY/AGG, rebalanced monthly (computed) (+288.8%),  Inverse volatility, unlevered (+134.8%),  ERC levered to 60/40 vol (All Weather) (+239.2%). The benchmark of this study is the 60/40 SPY/AGG line, the unit its own literature measures itself in. Dotted verticals mark the step boundaries; the dashed horizontal is break-even. These figures compound each arm's own stitched daily series; the pooled statistics in the text inner-join both arms' trading days, one session apart, both are printed from the frozen record.
Table 1. Five sleeves, quarterly rebalance with point-in-time re-estimation, one-year windows, total return net of costs. Sharpe is each book’s own daily series, measured over zero rather than over bills.
Book Total return Per year Volatility Sharpe Worst drawdown Mean window
60/40 SPY/AGG, rebalanced monthly (computed) (companion run) +288.8% 7.84% 11.9% 0.69 -34.8% +8.48%
ERC levered to 60/40 vol (All Weather) (companion run) +239.2% 7.02% 11.6% 0.64 -27.7% +7.61%
All Seasons static recipe +194.6% 6.19% 8.3% 0.76 -22.9% +6.71%
Equal risk contribution, unlevered +144.2% 5.08% 7.1% 0.73 -18.4% +5.44%
Inverse volatility, unlevered (companion run) +134.8% 4.86% 7.2% 0.70 -18.3% +5.12%

Volatility, Sharpe and worst drawdown are computed on each book's own stitched daily series over the identical trading days that Figure 1 draws, so the panel and the figure are the same arithmetic. Sharpe carries no cash hurdle. Mean window is the arithmetic average of the one-year window returns and does not compound to the total beside it; the difference is volatility drag.

The same walk, measured five ways0.00x3.31x6.61x
Figure 2. The measurement ladder: All Seasons static recipe's whole walk, chained five ways.  price only (+103.3%),  with dividends (+200.6%),  net of costs (+194.6%), against the benchmark measured both ways:  price only (+335.0%),  total return (+507.9%). The distance between the two green pairs is the dividends collected; the sliver between the last two greens is the cost bill; the distance between the two greys is what a price-only chart hides about the index. Every other figure on this page uses the deepest rung on each side, net of costs against the total-return benchmark.
Out-of-sample equity: normalised growth (1.00x = break even)0.73x1.04x1.35x2022202320242025
Figure 3. The same walk, re-based to 1.00x at the first window starting in 2022, 4 of the 18 windows above.  All Seasons static recipe (+6.8%),  Equal risk contribution, unlevered (+19.1%),  60/40 SPY/AGG, rebalanced monthly (computed) (+30.1%). This is a subset of Figure 1, not a correction to it. The era boundary here is pinned by the author at 2022 rather than left at the platform default, and the era rows below put a number on the two periods it separates. Whether the record actually breaks there is a question those rows answer, not one this caption settles.
The levered book against the benchmark it was built to match. Across the walk the vol target matched the 60/40’s volatility, 11.6 percent against 11.9, and the book still finished 0.82 points a year behind. Two windows carry all of that and more, because 2009 and 2013 together are minus 38.4 points while the other sixteen windows are 22.9 ahead. Both were years when equities ran double digits and the bond side made nothing. Every window is printed, so the total can be checked off the figure.
Figure 4. The levered book against the benchmark it was built to match. Across the walk the vol target matched the 60/40’s volatility, 11.6 percent against 11.9, and the book still finished 0.82 points a year behind. Two windows carry all of that and more, because 2009 and 2013 together are minus 38.4 points while the other sixteen windows are 22.9 ahead. Both were years when equities ran double digits and the bond side made nothing. Every window is printed, so the total can be checked off the figure.
Every book, window by window: eighteen one-year returns per weighting rule, with the 60/40 benchmark as the last row. 2008 is the window the whole family is built for, with four flat books against a 20.6 percent loss on the benchmark, and 2022 is the shared red stripe, the bond year that set every unlevered book’s worst window.
Figure 5. Every book, window by window: eighteen one-year returns per weighting rule, with the 60/40 benchmark as the last row. 2008 is the window the whole family is built for, with four flat books against a 20.6 percent loss on the benchmark, and 2022 is the shared red stripe, the bond year that set every unlevered book’s worst window.
Drawdown from peak, the four books and the 60/40. Every unlevered book’s worst loss of the whole record dates to October 2022, from bonds rather than stocks, while the levered book’s dates to 2008 because leverage amplified the crisis the family was built through. At 27.7 percent against the benchmark’s 34.8 it is the one comparison the levered book wins.
Figure 6. Drawdown from peak, the four books and the 60/40. Every unlevered book’s worst loss of the whole record dates to October 2022, from bonds rather than stocks, while the levered book’s dates to 2008 because leverage amplified the crisis the family was built through. At 27.7 percent against the benchmark’s 34.8 it is the one comparison the levered book wins.
The All Weather mechanism’s ledger. Top: the gross exposure the vol target chose ex ante each January, above 1.0 in all eighteen windows, pinned at the 2.0 cap four times, always after calm years. Bottom: financing paid on the borrowed excess at the point-in-time bill plus 50 bps, near-free through the zero-rate decade, 3.48 percent of capital in 2008 and above 2 percent again by 2025. 17.4 points cumulative, a third of everything the leverage earned.
Figure 7. The All Weather mechanism’s ledger. Top: the gross exposure the vol target chose ex ante each January, above 1.0 in all eighteen windows, pinned at the 2.0 cap four times, always after calm years. Bottom: financing paid on the borrowed excess at the point-in-time bill plus 50 bps, near-free through the zero-rate decade, 3.48 percent of capital in 2008 and above 2 percent again by 2025. 17.4 points cumulative, a third of everything the leverage earned.
Leave-one-out on the second finding. Each bar drops one window and prints what the other seventeen window gaps add up to, so every value can be checked against the comparison table’s own column. Drop 2014 and the 23.1 point margin becomes 13.5; drop 2020 and it becomes 14.9; drop both and 5.3 points are left across sixteen windows, with the recipe leading 10 of them. Two windows carry 77 percent of an eighteen-year result, and both are long-Treasury rallies meeting a fixed 40 percent TLT block. The same test on this paper’s own headline is in the table below it.
Figure 8. Leave-one-out on the second finding. Each bar drops one window and prints what the other seventeen window gaps add up to, so every value can be checked against the comparison table’s own column. Drop 2014 and the 23.1 point margin becomes 13.5; drop 2020 and it becomes 14.9; drop both and 5.3 points are left across sixteen windows, with the recipe leading 10 of them. Two windows carry 77 percent of an eighteen-year result, and both are long-Treasury rallies meeting a fixed 40 percent TLT block. The same test on this paper’s own headline is in the table below it.
Table 2. Table 1 carries every book’s total, rate, volatility, Sharpe and drawdown on its own stitched series. This is the one number it does not carry: the probability each book keeps a positive Sharpe once the declared four-rule search family is priced in. A book at 0.96 clears that bar in 96 samples of 100.
BookDeflated Sharpe probability
All Seasons recipe0.99
Equal risk contribution0.98
Inverse volatility0.98
ERC levered to 60/40 vol0.96
Table 3. The levered book’s vol target, ex ante at each anchor: gross exposure chosen, financing paid on the borrowed excess at the point-in-time 3-month bill plus 50 bps, and what survived to the investor. Financing of 0.93 points a year exceeds the 0.82 point gap against the benchmark.
Ledger lineValue
Windows asking gross above 1.018 of 18
Average gross1.52
Windows pinned at the 2.0 cap4 (2009, 2012, 2019, 2021)
Cheapest financing window0.10% of capital (2014)
Dearest financing window3.48% of capital (2008)
Financing paid, all windows17.4 points of capital
Unlevered risk parity made5.08% a year
Levered, before financing7.95% a year
Levered, after financing7.02% a year
So: leverage earned+2.86 points a year
and paid the lender-0.93, a third of it
The 60/40 benchmark made7.84%, borrowing nothing
Gross of rent, vs benchmark+0.11 points a year
Net of rent, vs benchmark-0.82 points a year
Table 4. The levered book minus the 60/40, for the two windows that carry the whole eighteen-year deficit, then the totals. Both were years when equities ran double digits and the bond side made nothing: 22.7 percent against 0.1 in the 2009 window, 29.0 against a loss of 1.1 in 2013. Every one of the eighteen windows is printed on the figure above.
WindowLevered60/40Difference
2009-1.5%+15.7%-17.2 pp
2013-5.4%+15.8%-21.2 pp
Those two windows together-38.4 pp
The other sixteen windows+22.9 pp
2008-2021, levered led 7 of 14-4.7 pp
2022-2025, levered led 1 of 4-10.8 pp
All eighteen, levered led 8-15.5 pp
Table 5. Leave one out, run cumulatively on this paper’s own headline as well as on the finding it demotes. Each step drops one more window and reports the sum of what survives. A result that changes this much when one window is dropped should be read as a description of this record.
FindingAll 18 windowsDrop its biggest windowAnd the next one too
Recipe minus risk parity+23.1 pp totalno 2014: +13.5 ppand no 2020: +5.3, leads 10/16
Levered book minus 60/40-15.5 pp totalno 2013: +5.7 ppand no 2009: +22.9, leads 8/16

2.2  Per-step results

Table 6. One row per step, raw out-of-sample results. A short window can pair a negative return with a positive annualised Sharpe: at high daily volatility the arithmetic mean of daily returns sits above the compounded window return, and the Sharpe reads the former. Volatility drag, printed rather than smoothed.
#Out-of-sample window All Seasons static recipe SR Equal risk contribution, unlevered SR Inverse volatility, unlevered SRERC levered to 60/40 vol (All Weather) SR
1 2008-01-02 → 2008-12-31 0.10 0.24 0.55 0.09
2 2009-01-02 → 2009-12-31 0.07 0.03 0.07 0.03
3 2010-01-04 → 2010-12-31 1.76 1.85 1.85 1.73
4 2011-01-03 → 2011-12-30 2.40 1.99 1.81 1.97
5 2012-01-03 → 2012-12-31 1.39 1.39 1.10 1.37
6 2013-01-02 → 2013-12-31 -0.11 -0.37 -0.80 -0.53
7 2014-01-02 → 2014-12-31 2.56 0.87 0.63 0.74
8 2015-01-02 → 2015-12-31 -0.40 -0.84 -0.91 -0.80
9 2016-01-04 → 2016-12-30 1.04 1.25 1.05 1.11
10 2017-01-03 → 2017-12-29 2.29 2.49 2.02 2.31
11 2018-01-02 → 2018-12-31 -0.32 -0.28 -0.35 -0.33
12 2019-01-02 → 2019-12-31 3.05 2.77 2.44 2.52
13 2020-01-02 → 2020-12-31 1.54 1.10 1.38 1.25
14 2021-01-04 → 2021-12-31 1.18 1.06 0.88 0.92
15 2022-01-03 → 2022-12-30 -1.47 -1.01 -1.06 -1.02
16 2023-01-03 → 2023-12-29 0.92 0.84 0.81 0.71
17 2024-01-02 → 2024-12-31 0.79 1.31 1.25 1.23
18 2025-01-02 → 2025-12-31 1.45 2.01 2.10 1.88
Out-of-sample equity: normalised growth (1.00x = break even)0.74x0.98x1.22xbars into the window →
Figure 9. All Seasons static recipe: every step's out-of-sample curve overlaid, each rebased to 1× at its own start. Read alongside Table 1: consistent shape across steps is the walk-forward's evidence; a single lucky leg is not.
Out-of-sample equity: normalised growth (1.00x = break even)0.81x1.00x1.19xbars into the window →
Figure 10. Equal risk contribution, unlevered: the same windows, the other arm. Compare shape-for-shape with the previous figure: the two arms trade the identical out-of-sample legs.

2.3  Search accounting

This paper's search is a declared family: the All Weather family: four weighting rules on one fixed five-sleeve book across three registered walks - the static All Seasons recipe and unlevered equal risk contribution as the two-arm comparative, inverse volatility and vol-targeted levered equal risk contribution as single-arm walks - with the 60/40 carried as the author benchmark, not a searched rule, counted at N = 4 evaluated books. Every member is either a registered walk with its own sealed hypothesis and frozen record, or a derived average computed from those frozen records; every member is reported, in the family matrix table and the robustness figure, and none was selected away. The count is declared by the author rather than derived from one project's ledger, because the members are sibling registered studies; the declaration names them and is frozen in this artifact. What the source strategy's author searched before publishing is not knowable from here and is not counted. The registered per-step record below still guarantees each window's hypothesis was hashed and registered before that window was scored.

2.4  The comparison

Both arms trade the same registered windows, so their returns can be PAIRED: inside each window the two return series are inner-joined date by date and the difference rAll Seasons static recipe − rEqual risk contribution, unlevered is the object under test. Because this is ONE pre-declared contrast, frozen at registration before any window was scored, the paired statistic needs no multiple-testing deflation; the arm-level records carry the declared family count of §2.3 as their search accounting, and this contrast, sealed per window before scoring, is not multiplied by it.

In the table: Arm A = All Seasons static recipe · Arm B = Equal risk contribution, unlevered.

Table 7. Window-by-window paired comparison. Δ is the growth gap (Arm A − Arm B) over the window's paired dates. The Inverse volatility, unlevered and ERC levered to 60/40 vol (All Weather) columns carry companion books' growth over the full window; the Δ and Leader columns compare only this study's two registered arms.
#WindowPaired bars Arm AArm B ΔLeader Inverse volatility, unleveredERC levered to 60/40 vol (All Weather)
1 2008-01-03 → 2008-12-31 252 +0.5% +1.9% -1.4 pp Arm B +4.9% -0.4%
2 2009-01-05 → 2009-12-31 251 +0.2% -0.1% +0.3 pp Arm A +0.1% -1.5%
3 2010-01-05 → 2010-12-31 251 +12.9% +12.7% +0.2 pp Arm A +13.1% +20.3%
4 2011-01-04 → 2011-12-30 251 +18.9% +13.8% +5.1 pp Arm A +13.4% +19.8%
5 2012-01-04 → 2012-12-31 249 +7.5% +6.5% +1.1 pp Arm A +5.5% +12.4%
6 2013-01-03 → 2013-12-31 251 -1.0% -2.4% +1.4 pp Arm A -5.4% -5.4%
7 2014-01-03 → 2014-12-31 251 +13.5% +3.9% +9.6 pp Arm A +2.8% +3.8%
8 2015-01-05 → 2015-12-31 251 -3.1% -5.3% +2.2 pp Arm A -5.8% -7.0%
9 2016-01-05 → 2016-12-30 251 +6.6% +7.1% -0.5 pp Arm B +6.1% +10.0%
10 2017-01-04 → 2017-12-29 250 +11.4% +10.4% +1.0 pp Arm A +8.9% +11.3%
11 2018-01-03 → 2018-12-31 250 -2.1% -1.7% -0.4 pp Arm B -1.9% -2.8%
12 2019-01-03 → 2019-12-31 251 +18.3% +14.7% +3.7 pp Arm A +13.8% +26.9%
13 2020-01-03 → 2020-12-31 252 +18.1% +9.9% +8.2 pp Arm A +12.2% +19.5%
14 2021-01-05 → 2021-12-31 251 +8.8% +6.4% +2.5 pp Arm A +5.1% +9.5%
15 2022-01-04 → 2022-12-30 250 -18.4% -11.5% -6.9 pp Arm B -11.6% -16.3%
16 2023-01-04 → 2023-12-29 249 +9.3% +7.0% +2.4 pp Arm A +6.8% +7.3%
17 2024-01-03 → 2024-12-31 251 +6.4% +9.6% -3.2 pp Arm B +8.9% +9.6%
18 2025-01-03 → 2025-12-31 249 +12.9% +15.2% -2.2 pp Arm B +15.2% +19.9%

Paired Sharpe of the difference track: 0.37 · block bootstrap (2000 paths, block 10, seed 1234): P(All Seasons static recipe beats Equal risk contribution, unlevered) = 94.9%.

Window win-rate. All Seasons static recipe led 12 of 18 windows (66.7%), Equal risk contribution, unlevered led 6, and the mean window gap of +1.26 pp points the same way. Widest single window: 2014 at +9.6 pp.

Table 8. The same comparison split at 2022. Pooling the whole walk into one row hides which side of the split the difference came from.
PeriodWindows All Seasons static recipeEqual risk contribution, unlevered 60/40 SPY/AGG Mean gapAll Seasons static recipe led
All windows 18 +6.71% +5.45% +8.46% +1.26 pp 12/18
Before 2022 14 +7.89% +5.56% +8.73% +2.34 pp 11/14
2022 onward 4 +2.55% +5.07% +7.53% -2.52 pp 1/4
All windowsn=18 · All Seasons static recipe led 12 · Equal risk contribution, unlevered led 6 · ties 0+6.7%+5.5%+1.26 ppBefore 2022n=14 · All Seasons static recipe led 11 · Equal risk contribution, unlevered led 3 · ties 0+7.9%+5.6%+2.34 pp2022 onwardn=4 · All Seasons static recipe led 1 · Equal risk contribution, unlevered led 3 · ties 0+2.5%+5.1%-2.52 ppgap
Figure 11. Mean window return per period. All Seasons static recipe above, Equal risk contribution, unlevered below, with the gap at right. The pooled bar and the post-2022 bar are the same comparison over different periods.

The two eras disagree by 4.86 pp. The pooled figure is therefore not a standing property of either method, it is dominated by the later period. Read the two rows, not the average.

3  The circuit

The strategy is a circuit of platform primitives, frozen when the study is registered. Below is the circuit as wired on the canvas, the objective it encodes and how the search runs through it, followed by the mathematics each primitive actually computes, the same formulas the execution engine runs. The complete parameterisation is preserved in the study ledger (Appendix A).

The hypothesis under test

A COMPARATIVE study: All Seasons static recipe vs Equal risk contribution, unlevered, walked on the same registered out-of-sample windows. All Seasons static recipe: All_seasons, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor: anything the design estimates from history, where it estimates at all, is re-estimated at each anchor from pre-anchor data only, and the walk advances through registered out-of-sample windows; its disposition is the realized forward path versus the benchmark. Equal risk contribution, unlevered: All_seasons, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor: anything the design estimates from history, where it estimates at all, is re-estimated at each anchor from pre-anchor data only, and the walk advances through registered out-of-sample windows; its disposition is the realized forward path versus the benchmark. The arms differ in: Portfolio Σ → Portfolio Optimizer, substituted (its parameters change with the swap). The contrast under test: whether All Seasons static recipe generates better risk-adjusted returns than Equal risk contribution, unlevered over the identical out-of-sample windows.

The frozen circuit, data flows left to rightuniverse: click for detailsuniverseprice loader: click for detailsprice loaderportfolio builder: click for detailsportfolio builderportfolio backtest: click for detailsportfolio backtesttransaction cost: click for detailstransaction costuniverse: click for detailsuniverseprice loader: click for detailsprice loaderportfolio optimizer: click for detailsportfolio optimizerportfolio backtest: click for detailsportfolio backtesttransaction cost: click for detailstransaction costAll Seasons static recipeEqual risk contribution, unleveredshared
Figure 12. The frozen circuit, every node a primitive, every wire a typed data-flow; the two arms are colour-coded (All Seasons static recipe green, Equal risk contribution, unlevered blue, shared feeds neutral). Each box is one step of the strategy; data flows along the wires left to right, and no box can see data dated later than the box feeding it. The whole diagram was frozen when the hypothesis was registered. Click any node to open what that step ran with and what it produced.

Envelopes show counts, ratios, dates, and the parameters the author chose. Full price and per-name data series are not republished: the underlying market data is licensed to QuanterLab. Point figures quoted in the prose, a named holding's return over a stated span, are summary facts derived from public market prices, not redistributed series.

What each part does
Universe, The starting set of tickers, the All Seasons sleeves: one fixed list of five exchange-traded funds standing in for the asset classes the recipe names; no membership reconstruction applies.
Price Loader, Bulk OHLCV fetch for the whole universe, point-in-time, no future bars.
Portfolio Builder, From "which names" to "how much of each", the weighting rules.
Portfolio Optimizer, MVO · HRP · IVOL, three rigorous ways to weight a pool.
Transaction Cost, Charge for trading, slippage + commission on every turn.
Portfolio Backtest, Replay the portfolio forward, rebalanced, point-in-time, with costs.

The objective and the search

All Seasons static recipe

Universeall_seasons index constituents.
Validation & out-of-sampleportfolio forward test (buy-and-hold book) (1y horizon from the anchor, quarterly rebalance); overlays: Transaction Cost.
Other componentsCombine: Portfolio Σ.

Equal risk contribution, unlevered

Other componentsCombine: Portfolio Optimizer.

Every other specification row is identical to All Seasons static recipe's table above.

What differs between the arms, one difference; the comparison is clean:

  • substitutedPortfolio Σ → Portfolio Optimizer, substituted (its parameters change with the swap)

Companion book. Inverse volatility, unlevered comes from a companion run of the identical configuration on its own sealed anchor schedule. Its results run through this paper's tables under its own name.

Companion book. ERC levered to 60/40 vol (All Weather) comes from a companion run of the identical configuration on its own sealed anchor schedule. Its results run through this paper's tables under its own name.

Everything else is held identical, so an out-of-sample gap between the arms is attributable to this one change.

Cost elements are wired into the circuit, the realised drag is reported per step in Appendix B.

Show the mathematics, 6 primitives, formulas and parity notes

3.1  Universe

The starting set of tickers, resolved point-in-time from the index change-log, so names delisted or removed later still compete on the dates they traded.

Before any math, you need a list of stocks. An index preset (S&P 500, Nasdaq-100, Dow 30) is reconstructed as it stood ON your anchor date by replaying the historical add/drop change-log backwards, so a 2018 backtest sees the 2018 membership, not today's winners.

Point-in-time membership

Start from today's constituents and un-apply every membership change after the anchor t:

\mathcal{U}(t) = \mathcal{U}_{\text{now}} \;\ominus\; \{\text{adds after } t\} \;\oplus\; \{\text{drops after } t\}
Constituents resolved from the index change-log; the same point-in-time set the factor + screening modules use.

3.2  Price Loader

Bulk OHLCV fetch for the whole universe, point-in-time, no future bars.

Momentum, volatility, trend, every price-based metric needs history. This loads open/high/low/close/volume for all names in parallel, clipped so nothing after the anchor can leak in. The lookback window is derived automatically from the deepest metric you wired.

The window is derived, not guessed

It loads exactly enough history for the hungriest downstream metric plus a warm-up buffer:

W = \max_k(\text{lookback}_k) + \text{buffer}, \qquad \text{bars} \le \text{anchor } t

3.3  Portfolio Builder

From "which names" to "how much of each", the weighting rules.

Turns a basket into capital weights by a simple rule: equal weight, score-weighted, inverse-volatility, or cap-weighted. An optional max-weight cap waterfills the excess onto the under-cap names so nothing is over-concentrated.

Weighting schemes
w_i^{\text{eq}}=\tfrac1N,\quad w_i^{\text{score}}\propto \text{score}_i,\quad w_i^{\text{ivol}}\propto \tfrac{1}{\sigma_i},\quad w_i^{\text{cap}}\propto \text{mktcap}_i
Max-weight water-filling

Clip any weight above the cap, redistribute the excess proportionally to names still under the cap, and repeat until none exceed it (≤ 24 rounds).

w_i \leftarrow \min(w_i, w_{\max}); \;\; \text{redistribute excess} \propto w_j\,[w_j < w_{\max}]

3.4  Portfolio Backtest

Replay the portfolio forward, rebalanced, point-in-time, with costs.

Holds the basket and rebalances on schedule, re-selecting and re-optimizing point-in-time at each rebalance (so it only ever uses information available then), and reports the equity curve, Sharpe, drawdown and trade stats, optionally net of cost and risk overlays.

Compounded equity
E_t = E_{t-1}\big(1 + \mathbf w_{t}^{\top}\mathbf r_t - \text{costs}_t\big)
Drawdown
\text{DD}_t = \frac{E_t}{\max_{\tau\le t}E_\tau} - 1, \qquad \text{MaxDD} = \min_t \text{DD}_t
Financing a levered book
\text{charge}_t \;=\; \text{loan}_t \cdot \frac{\text{spread}}{252}, \qquad \text{loan}_t = \begin{cases}\max(0,\,-\text{cash}_t) & \text{institutional (netted)}\\ \max(0,\,\text{long MV}_t - E_t) & \text{retail (no netting)}\end{cases}
A levered long/short book (the β-neutral Long/Short Select) borrows its excess notional. WHO you are decides the loan: a prime broker nets short-sale proceeds against the margin loan, a fully-netted BAB book carries almost none, while a retail margin account cannot net, so the same book borrows the long leg’s excess over equity. Profiles: institutional = 50 bps spread + 25 bps GC short borrow; retail = 350 bps + 150 bps (the BEST retail tier, so any verdict is conservative); custom = your own knobs, spread on negative cash. Charged daily, reported as financing_drag_pct, never silent.

3.5  Transaction Cost

Charge for trading, slippage + commission on every turn.

Real trading isn't free. This deducts a cost proportional to how much you trade (turnover), in basis points, so the backtest reflects net, not gross, performance.

Cost per rebalance
\text{cost}_t = \frac{\text{bps}}{10{,}000}\;\times\;\text{turnover}_t, \qquad \text{turnover}_t = \tfrac12\sum_i \lvert w_{i,t}-w_{i,t^-}\rvert

3.6  Portfolio Optimizer

MVO · HRP · IVOL, three rigorous ways to weight a pool.

Beyond simple rules, three portfolio-theory optimizers. MVO maximizes return per unit of risk on the efficient frontier; HRP spreads risk across correlation clusters with no return forecast; IVOL is the simplest risk-balanced rule. All re-solve point-in-time at every rebalance.

MVO, maximize the Sharpe ratio
\max_{\mathbf w}\ \frac{\mathbf w^{\top}\boldsymbol\mu - r_f}{\sqrt{\mathbf w^{\top}\boldsymbol\Sigma\,\mathbf w}} \quad \text{s.t.}\ \ \mathbf 1^{\top}\mathbf w = 1,\ \ 0 \le w_i \le w_{\max}
Σ via Ledoit–Wolf shrinkage; long-only; negative-expected-return names excluded (SLSQP).
HRP, cluster, then split risk
d_{ij} = \sqrt{\tfrac12\,(1-\rho_{ij})} \ \to\ \text{hierarchical clusters}\ \to\ \text{recursive inverse-variance bisection}
No return forecast, robust to estimation error.
IVOL, inverse volatility
w_i = \frac{1/\sigma_i}{\sum_j 1/\sigma_j}
1% volatility floor. Fallback chain (IVOL → per-name → equal) is disclosed if a method degenerates.
Byte-identical to FM101FBKT optimizer engines (shared_libs/factor_core).

4  Projection calibration, pooled across the walk

Every rebalance carried a Monte Carlo cone and a 95% VaR estimated before the segment it is scored against. Two questions, pooled over the whole study: did realized outcomes land inside the band as often as the band claims, and were VaR breaches as frequent as 5%?

This section is produced by the forward tester itself: every portfolio backtest fits the cone and the VaR estimate at each rebalance and scores them against the segment that followed. It does not require, and this circuit does not contain, a Monte Carlo primitive; that primitive is a separate, standalone analysis.

Arm A62 of 72 inside the 90% band-13%-0%+13%in band200820092010201120122013201420152016201720182019202020212022202320242025Arm B61 of 72 inside the 90% band-13%-0%+13%in band200820092010201120122013201420152016201720182019202020212022202320242025
Figure 13. Projected range versus what occurred, at each of 144 scored rebalance segments, pooled across both arms. The final rebalance of each step has no following segment to score, the ledger marks those rows “no segment follows this rebalance”, which is why this count sits below the raw rebalance totals in the table beneath. Each vertical bar is that rebalance's P5–P95 Monte Carlo cone with the median ticked; the dot is the realized return of the segment that followed. Filled green = the outcome landed inside its own cone; red = it did not. The strip beneath repeats that as one mark per rebalance, so a run of misses in one period is visible as a run. Every cone was fitted only on data prior to the segment it is scored against.
Arm Steps Rebalances In band Coverage Expected VaR days Breach rate Expected
All Seasons static recipe 18 85 62 / 72 86.1% ±4.08 90.0% 4457 5.59% ±0.344 5.0%
Equal risk contribution, unlevered 18 85 61 / 72 84.7% ±4.24 90.0% 4457 5.63% ±0.345 5.0%

Note. The platform grades its own risk model in the appendix. Across the family the 90 percent Monte Carlo cones contained 84.7 to 86.1 percent of realized rebalance segments against the 90 expected, and the one-day 95 percent VaR was breached on 5.6 to 5.7 percent of days against the 5 expected, which is a milder miss than the concentrated-book studies on this site and is what five sleeves should do to a risk model. The same appendix explains the occasional window that pairs a negative return with a positive annualised Sharpe, which is volatility drag, stated once.

± values are binomial standard errors on the estimate. A coverage figure below the expected band means the projection was over-confident; a breach rate above 5% means the same of the risk model. Both forecasts used only data prior to the segment scored.

5  Discussion

5.1  Findings

Unlevered, equal risk contribution on these five sleeves made 5.08 percent a year, and levered to the 60/40's trailing volatility it made 7.95 before financing, so the leverage added 2.86 points a year over the unlevered version of the same book. It then paid 0.93 points a year to the lender, which is 17.4 points of capital across the eighteen windows and a third of everything the leverage earned, and it arrived at 7.02. The 60/40 it was scaled to match made 7.84 and borrowed nothing. So the mechanism cleared its benchmark by 0.11 points a year gross of the bill and finished 0.82 points behind after it, which makes the bill larger than the gap.

The deficit is dateable. The eighteen window differences add up to minus 15.5 points against the 60/40, and two windows carry more than all of it, because the 2009 window cost 17.2 points and the 2013 window cost 21.2, together minus 38.4, while across the other sixteen windows the levered book finished 22.9 points ahead. Both years have the same cause. In the 2009 window equities returned 22.7 percent while the bond-heavy recipe made 0.1, and in the 2013 window equities returned 29.0 percent while the recipe lost 1.1, which makes them the only two windows in the sample where equities ran double digits and the diversified book made nothing at all. A benchmark that is sixty percent equities gains in those two years and a book carrying about a fifth of its risk in equities does not follow it, so the deficit is what a book holding a fifth of its risk in equities does in the two largest equity-only years of the record.

It did not beat the benchmark, and the record will not carry a stronger verb than that, because the levered book led in only 8 of 18 windows and the shortfall is not a stable estimate: dropping the 2013 window alone reverses its sign, and dropping 2009 as well leaves the levered book 22.9 points ahead over the remaining sixteen. We print that leave-one-out beside the one we run on the recipe. The leverage cap is a design note here rather than a finding, because gross was pinned at 2.0 in four windows, always after calm years, so in those the book ran below the volatility it was targeting, and three of the four net to nothing against the benchmark while the fourth is 2009, which the paragraph above already counts.

Realized volatility came out at 11.6 percent against the benchmark's 11.9, so across the walk the book carried the risk it was built to carry and the return comparison is between two books at the same risk.

Every unlevered book in the family set its worst drawdown of the entire eighteen years in October 2022, which is the recipe at 22.9 percent, inverse volatility at 18.3 and equal risk contribution at 18.4. None of them set it in 2008 or in 2020. In 2022 long Treasuries fell with equities, so both sides of the book lost together. The levered book is the one exception, because its worst loss of 27.7 percent dates to November 2008 and leverage amplified the crisis the family was built through. The four books all drew down less than the 60/40's 34.8. The levered book gave up eight tenths of a point a year against that benchmark and drew down seven points less than it, at the same volatility.

The 2008 window holds the best and the worst of the levered book. Window one, formed January 2008, has the recipe at 0.4 percent, inverse volatility at 4.9, risk parity at 1.8 and the levered book at minus 0.4, against the 60/40 at minus 20.6, which is a twenty point gap in the levered book's favour and the largest margin it posts over the benchmark anywhere in the record. It is also the window in which the levered book set that 27.7 percent drawdown, in November, so the window return and the worst drawdown describe the same twelve months.

The recipe beat the engineering, and the reason is duration. Against equal risk contribution on the same sleeves, the same clock and the one sealed difference of who sets the weights, the static recipe wins by 1.11 points a year and in 94.9 percent of paired daily bootstrap paths.

Risk parity does not hold less in bonds, which is what we assumed before pulling the weights. The solver averaged 17.0 percent in the long bond across the eighteen anchors against the recipe's fixed 40, and never held more than 26.3 or less than 12.7. TLT carries roughly three times the interest-rate sensitivity of IEF, so the recipe holds about the same amount of bonds at roughly twice the duration. From 2008 through 2021 the long bond was the best asset in the set, so a fixed allocation to it beat every data-driven attempt to size it. In January 2022 the solver held 16.2 percent long bonds against the recipe's 40, and that is the window where the recipe lost 18.4 percent, which is the worst any unlevered book printed in the sample, and gave back 6.9 of its 23.1 points in one year.

That margin is two windows. The eighteen window gaps in the comparison table add up to 23.1 points, the 2014 window contributes 9.6 of them and the 2020 window contributes 8.2, so together they are 17.8 points, or 77 percent of the entire eighteen-year margin. Drop 2014 and 13.5 points survive across seventeen windows, drop 2020 as well and 5.3 points are left across sixteen, with the recipe leading ten of them rather than twelve of eighteen. Twelve of eighteen is a sign test of 0.238, which is not significant at any conventional level, and the 94.9 percent is the same difference track restated, because the normal approximation to a paired Sharpe of 0.37 over eighteen windows returns 94.2 percent. The bootstrap resamples days while the strategy trades quarters, so the window count is the one to believe, and the consistency-across-steps line under the walk-forward figure is answered by the leave-one-out figure above it. Inverse volatility, which cannot see the correlations at all, trails everything at 4.86 percent a year against the recipe's 6.19.

5.2  Interpretation

The naming decides what a reader thinks they own. All Seasons, the fixed recipe, is what retail articles, robo portfolios and YouTube mean when they say All Weather, while All Weather proper has always been levered risk parity, and Bridgewater's own literature is explicit that the point is to equalize risk across environments and then lever the result to a chosen volatility. This study built both. The retail one beat its unlevered engineering cousins on a margin two windows carry, and the institutional one, priced at real bills, did not clear a portfolio anyone can buy in two trades. Anyone selling the recipe as All Weather is selling the name without the borrowing that defines it.

The construction itself holds up. The leverage decision uses only trailing data, the clamp held gross at or below 2.0 in every window, and the financing is charged at a rate that actually existed on each anchor date rather than at a flat spread. What that bought was a book which matched its benchmark's volatility, beat it before financing by a tenth of a point and trailed it by eight tenths after. Against the recipe the leverage argument survives by 0.83 points a year, and against the 60/40 it was calibrated to match it does not survive at all.

One thing the Sharpe numbers do not say, because a reader will otherwise read it into them. Absent a binding leverage cap, levering a book cannot lower its risk-adjusted return except through the financing spread, and the fall from 0.73 unlevered to 0.64 levered in Table 1 is almost entirely the zero-cash-hurdle convention dividing a larger volatility. Measured over the 1.35 percent average bill this walk actually faced, unlevered risk parity scores 0.54 against the levered book's 0.52 and the 60/40's 0.58, so the residual cost of the leverage is about 0.02, which is the spread and nothing more. The static recipe leads the whole study on that measure at 0.60, which is the simplest book here and the only one that borrows nothing, and it pays for that with 6.19 percent a year against the levered book's 7.02. The condition matters, because the cap bound in four of eighteen windows and in those the levered book is not a constant multiple of the unlevered one. What the Sharpe comparison does support is the ranking against the benchmark, whose volatility sits within a third of a point of the levered book's, and there the 60/40 is ahead on either convention.

What the ledger adds is the shape of the financing cost across rate regimes, which is near free through the zero-rate decade, 3.48 percent of capital in 2008 and above 2 percent again by 2025. Any levered strategy evaluated with flat or missing financing is being measured over the years when borrowing was cheapest, and the era split shows it, because the levered book led the 60/40 in 7 of 14 windows through 2021 and in only 1 of 4 since.

The claim these constructions actually make is about drawdown, and the drawdown figure answers it. The four unlevered books bottomed between 18.3 and 22.9 percent against the 60/40's 34.8, and the levered book at 27.7 percent is seven points shallower than the benchmark it was scaled to match.

Where this stops. The result holds for an institutional borrower at bills plus fifty and for this one era, and it would go the other way for anyone who could borrow cheaper, or who ran the book through a period when equities and bonds paid more evenly. It says nothing about whether the levered book is worth running today, because that answer depends on a financing rate we do not know yet and on a stock-bond correlation which has already changed once inside this record. What is still open is the retail case, which needs margin rates several times the spread modelled here, and a longer record, which exchange-traded funds cannot supply.

This study is one member of a declared search family: the same design walked at several sealed settings across sibling registered projects, every member either a registered walk with its own frozen record or a derived average of those records, and every member reported. The family size is declared by the author and named in the lineage; it is the search-accounting count for this paper. What was searched before the source strategy was published is not knowable from here and is not counted.

5.3  Limitations

The record holds one long bond bull from 2008 to 2021 and one bond crash in 2022, fourteen windows and four. The recipe's edge over risk parity is concentrated in that bull and, as the leave-one-out shows, in two windows of it, while the levered book's deficit against the 60/40 sits in two windows of the same sample and reverses on dropping either one. A record reaching the 1970s would interrogate all of this far more harshly, but exchange-traded funds do not reach there and this platform does not splice index data onto fund data.

The statistical weight is thin on both findings and is stated as such. The levered book led the 60/40 in 8 of 18 windows, which is a two-sided sign test of 0.815, so the defensible claim is that the mechanism did not clear its benchmark over this record rather than that it is reliably beaten in a given year. The recipe led risk parity in 12 of 18, a sign test of 0.238, which is also not significant, and the 94.9 percent figure elsewhere on this page is a daily-resampled bootstrap of the same difference track that reproduces to 94.2 percent from a normal approximation, so it is not independent evidence.

The financing convention has a direction. The bill is resolved at each window's anchor and held for that window, so a year in which bills rose sharply after January is undercharged, and 2022 is the case that matters, because the anchor bill was 0.06 percent while the actual three-month bill passed 4 percent inside the window. That flatters the levered book rather than the benchmark, so correcting it would widen the shortfall rather than close it. Gross exposure is likewise set once at the anchor while the weights re-estimate quarterly.

Two more conventions, both named where the comparisons are made. The 60/40 reference carries no trading cost while all four books pay ten basis points per one-way traded dollar, and charged the same way it would give back 0.41 points of capital across the walk, or about 0.02 a year against a gap of 0.82. Sharpe carries no cash hurdle, which flatters low-volatility books most, so it flatters unlevered risk parity against the levered book by roughly 0.07 and the levered book against the 60/40 by about 0.005, and correcting it widens the benchmark's lead rather than narrowing it.

The estimation choices were fixed rather than searched, which are a 252-day window for volatilities and covariances, quarterly re-estimation, the 0.5 to 2.0 gross clamp and bills plus 50 basis points. The family is declared at four rules and the deflated statistics use exactly that count, so none of the knobs was tuned and none of the alternatives was run. The cap bound in four windows, so a different ceiling would produce a different number, though three of those four net to nothing against the benchmark.

Financing is modelled as bill plus a 50 basis point institutional spread on the borrowed excess. A retail investor levering through margin pays several times that spread and would take the pain of it every window, so at retail financing the levered book loses its 0.83 point edge over the recipe and trails the 60/40 by well over two points a year. Shorting, taxes and fund expense drift are absent, though the fund fees embedded in these total-return series are a real cost of owning the mechanism and are carried on both sides. The 10 basis point trading cost treats DBC's spread the same as SPY's, which flatters the commodity sleeve slightly.

References

QuanterLab reference architecture
  1. Gelman, A., & Loken, E. (2013). The garden of forking paths: Why multiple comparisons can be a problem, even when there is no “fishing expedition.” Working paper, Columbia University.
  2. Harvey, C. R., Liu, Y., & Zhu, H. (2016). … and the Cross-Section of Expected Returns. Review of Financial Studies, 29(1), 5–68. doi:10.1093/rfs/hhv059
  3. Lo, A. W. (2002). The Statistics of Sharpe Ratios. Financial Analysts Journal, 58(4), 36–52. doi:10.2469/faj.v58.n4.2453
Author’s references?
  1. Dalio, R. et al. (Bridgewater Associates). The All Weather Story; Engineering Targeted Returns and Risks. Bridgewater research notes.
  2. Robbins, T. (2014). Money: Master the Game. Simon and Schuster. (The All Seasons allocation as stated by Dalio in interview.)
  3. Maillard, S., Roncalli, T., Teiletche, J. (2010). The Properties of Equally Weighted Risk Contribution Portfolios. Journal of Portfolio Management 36(4), 60-70.
  4. Asness, C., Frazzini, A., Pedersen, L. H. (2012). Leverage Aversion and Risk Parity. Financial Analysts Journal 68(1), 47-59. (They argued the high-Sharpe diversified book should be levered rather than diluted; this record prices what the borrowing cost.)

Appendix A  Reproducibility in QuanterLab

Each step is backed by a frozen run report. The study is re-derivable from the ledger below.

#CommitReportAnchorOOS window
1 33e9fa855116 2050 2008-01-01 2008-01-02 → 2008-12-31
2 24c369614700 2051 2009-01-01 2009-01-02 → 2009-12-31
3 c239d6eb9fc0 2052 2010-01-01 2010-01-04 → 2010-12-31
4 d68ac5dd69eb 2053 2011-01-01 2011-01-03 → 2011-12-30
5 37b6338c469e 2054 2012-01-01 2012-01-03 → 2012-12-31
6 aeee5387e377 2055 2013-01-01 2013-01-02 → 2013-12-31
7 f00eade18cc4 2056 2014-01-01 2014-01-02 → 2014-12-31
8 81f2def2d89e 2057 2015-01-01 2015-01-02 → 2015-12-31
9 d7a334dbb515 2058 2016-01-01 2016-01-04 → 2016-12-30
10 d370622195e3 2059 2017-01-01 2017-01-03 → 2017-12-29
11 4c1303853a82 2060 2018-01-01 2018-01-02 → 2018-12-31
12 f0338b9560b4 2061 2019-01-01 2019-01-02 → 2019-12-31
13 da11b9b43cce 2062 2020-01-01 2020-01-02 → 2020-12-31
14 41d5a6e83a6c 2063 2021-01-01 2021-01-04 → 2021-12-31
15 7f8b5e7e78e3 2064 2022-01-01 2022-01-03 → 2022-12-30
16 05d8c09f080b 2065 2023-01-01 2023-01-03 → 2023-12-29
17 212334417a0e 2066 2024-01-01 2024-01-02 → 2024-12-31
18 cf7a7ec001bd 2067 2025-01-01 2025-01-02 → 2025-12-31

Appendix A2  Registration record

What this record does and does not establish. Every window in this study is historical: the data existed before the study began, so this is sequential sealing on past windows, not pre-registration in the clinical-trial sense, and no procedure could make it so. What the platform does enforce is order, each step's specification was frozen and hashed before that step was scored, and the walk cannot advance past a step that was never run or close one with a result registered for a different window. The two timestamp columns below are the evidence: read them together and each seal precedes its own run, and each run precedes the next seal. A study whose seals all post-date its runs would show it here. Wall-clock spacing between seals varies with the author's schedule and queue latency; the ordering, not the tempo, is the claim.

“A COMPARATIVE study: All Seasons static recipe vs Equal risk contribution, unlevered, walked on the same registered out-of-sample windows. All Seasons static recipe: All_seasons, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor: anything the design estimates from history, where it estimates at all, is re-estimated at each anchor from pre-anchor data only, and the walk advances through registered out-of-sample windows; its disposition is the realized forward path versus the benchmark. Equal risk contribution, unlevered: All_seasons, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor: anything the design estimates from history, where it estimates at all, is re-estimated at each anchor from pre-anchor data only, and the walk advances through registered out-of-sample windows; its disposition is the realized forward path versus the benchmark. The arms differ in: Portfolio Σ → Portfolio Optimizer, substituted (its parameters change with the swap). The contrast under test: whether All Seasons static recipe generates better risk-adjusted returns than Equal risk contribution, unlevered over the identical out-of-sample windows.”

The same hypothesis was registered independently at every step, hashed before each step's out-of-sample window was scored:

Table 9. Registration audit, one row per registered step, with the time each specification was frozen and the time its window was scored. The hypothesis is identical on every row by design: it was registered once and re-registered unchanged at each anchor. Rows that differ would mean the specification moved mid-walk, which is the thing this record exists to rule out. The timestamps are the separate claim: each seal precedes its own run, and each run precedes the next seal.
#AnchorRegistered at (UTC)Run completed (UTC)
1 2008-01-012026-08-30 14:54:45 2026-08-30 14:55:03
2 2009-01-012026-08-30 14:55:06 2026-08-30 14:55:21
3 2010-01-012026-08-30 14:55:24 2026-08-30 14:55:39
4 2011-01-012026-08-30 14:55:42 2026-08-30 14:55:57
5 2012-01-012026-08-30 14:56:00 2026-08-30 14:56:15
6 2013-01-012026-08-30 14:56:19 2026-08-30 14:56:34
7 2014-01-012026-08-30 14:56:37 2026-08-30 14:56:52
8 2015-01-012026-08-30 14:56:55 2026-08-30 14:57:10
9 2016-01-012026-08-30 14:57:13 2026-08-30 14:57:28
10 2017-01-012026-08-30 14:57:31 2026-08-30 14:57:46
11 2018-01-012026-08-30 14:57:49 2026-08-30 14:58:04
12 2019-01-012026-08-30 14:58:07 2026-08-30 14:58:22
13 2020-01-012026-08-30 14:58:25 2026-08-30 14:58:41
14 2021-01-012026-08-30 14:58:44 2026-08-30 14:58:59
15 2022-01-012026-08-30 14:59:02 2026-08-30 14:59:17
16 2023-01-012026-08-30 14:59:20 2026-08-30 14:59:35
17 2024-01-012026-08-30 14:59:38 2026-08-30 14:59:53
18 2025-01-012026-08-30 14:59:56 2026-08-30 15:00:11

Appendix B  Per-step diagnostics

Realized in the projection tables below is the risk engine scoring its own forecast: the buy-and-hold return of the segment that followed each rebalance, on the same gross basis the cone was projected on. It is deliberately not the charged, calendar-window total return the study’s tables print, so the two will not reconcile line by line; the cone and its outcome share one basis, which is what a calibration test requires. Each row names its segment’s span so a boundary session is visible.

What each step's run actually did beyond its return: capital allocation across lanes and regimes, the portfolio book's rebalancing and cost drag, and how positions were sized. Harvested from the frozen run reports, present where the circuit produced them.

Open the full per-step grid (18 steps: every rebalance, capital routing and sizing, per window)

Step 1 · 2008-01-02 → 2008-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 13.65% of 249 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2008-01-01 → 2008-04-01 -3.6936% 1.4518% 6.3646% 0.0884%yes 0.6055% 6 / 60
2008-04-01 → 2008-07-01 -3.8636% 1.5346% 6.7221% -0.7252%yes 0.6259% 3 / 63
2008-07-01 → 2008-10-01 -3.7521% 1.589% 6.7187% -3.5575%yes 0.6216% 6 / 63
2008-10-01 → window end -4.1177% 1.1065% 6.1193% 1.3865%yes 0.6438% 19 / 63

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.3× · cost drag 0.13%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 13.25% of 249 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2008-01-01 → 2008-04-01 -3.2621% 1.5317% 6.0927% 2.4613%yes 0.5672% 3 / 60
2008-04-01 → 2008-07-01 -3.2341% 1.8969% 6.8138% 0.5257%yes 0.5868% 5 / 63
2008-07-01 → 2008-10-01 -3.1636% 1.8297% 6.6083% -4.1075%no 0.5778% 5 / 63
2008-10-01 → window end -3.5282% 1.2405% 5.7953% 0.1737%yes 0.5746% 20 / 63

Step 2 · 2009-01-02 → 2009-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 6.45% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2009-01-01 → 2009-04-01 -7.1997% 0.7704% 8.6033% -7.8308%no 0.8666% 12 / 60
2009-04-01 → 2009-07-01 -8.6326% -0.2636% 8.9273% -0.5199%yes 1.0156% 3 / 62
2009-07-01 → 2009-10-01 -9.7084% -0.3771% 8.9726% 7.1129%yes 1.0484% 0 / 63
2009-10-01 → 2010-01-01 -9.0156% 0.5349% 10.1174% -0.7138%yes 1.0259% 1 / 63
2010-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 7.66% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2009-01-01 → 2009-04-01 -4.955% 1.5136% 7.7714% -4.6667%yes 0.7001% 8 / 60
2009-04-01 → 2009-07-01 -5.8928% 0.9519% 8.3386% -2.0728%yes 0.8005% 8 / 62
2009-07-01 → 2009-10-01 -7.0722% 0.6944% 8.3432% 4.5343%yes 0.8629% 0 / 63
2009-10-01 → 2010-01-01 -6.7486% 1.3824% 9.4153% -0.3208%yes 0.8929% 3 / 63
2010-01-01 no segment follows this rebalance, not scored

Step 3 · 2010-01-04 → 2010-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 2.02% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2010-01-01 → 2010-04-01 -9.7944% -0.4151% 8.9492% 0.5062%yes 1.0363% 1 / 60
2010-04-01 → 2010-07-01 -9.4939% -0.5599% 9.3157% 2.8751%yes 1.0484% 0 / 62
2010-07-01 → 2010-10-01 -9.5369% -0.0078% 9.5566% 7.3672%yes 1.049% 0 / 63
2010-10-01 → 2011-01-01 -8.4993% 1.1671% 10.8716% -0.5004%yes 1.049% 4 / 63
2011-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 2.02% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2010-01-01 → 2010-04-01 -8.113% 0.2598% 8.5264% -0.1667%yes 0.9426% 1 / 60
2010-04-01 → 2010-07-01 -8.0177% -0.2158% 8.299% 3.5323%yes 0.8937% 0 / 62
2010-07-01 → 2010-10-01 -7.8883% 0.218% 8.2324% 7.0168%yes 0.8547% 0 / 63
2010-10-01 → 2011-01-01 -7.0056% 1.0996% 9.107% 0.151%yes 0.8366% 4 / 63
2011-01-01 no segment follows this rebalance, not scored

Step 4 · 2011-01-03 → 2011-12-30

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 3.23% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2011-01-01 → 2011-04-01 -5.8667% 1.9295% 11.178% 1.6586%yes 0.9594% 0 / 61
2011-04-01 → 2011-07-01 -5.1388% 1.9212% 9.5525% 1.1069%yes 0.814% 1 / 62
2011-07-01 → 2011-10-01 -4.3725% 2.2506% 8.6908% 7.433%yes 0.7374% 6 / 63
2011-10-01 → 2012-01-01 -3.4713% 2.5862% 9.0625% 3.6096%yes 0.7374% 1 / 62
2012-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 3.63% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2011-01-01 → 2011-04-01 -4.8787% 2.1375% 10.3879% 1.959%yes 0.829% 1 / 61
2011-04-01 → 2011-07-01 -4.1819% 2.288% 9.2362% 1.2988%yes 0.7151% 1 / 62
2011-07-01 → 2011-10-01 -3.5486% 2.5915% 8.5321% 4.1724%yes 0.6701% 5 / 63
2011-10-01 → 2012-01-01 -3.1033% 2.4694% 8.3978% 3.0007%yes 0.6627% 2 / 62
2012-01-01 no segment follows this rebalance, not scored

Step 5 · 2012-01-03 → 2012-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 2.03% of 246 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2012-01-01 → 2012-04-01 -2.9858% 2.5883% 9.0388% 1.0011%yes 0.7167% 2 / 61
2012-04-01 → 2012-07-01 -3.0499% 2.8336% 9.1114% 2.6485%yes 0.7087% 2 / 62
2012-07-01 → 2012-10-01 -3.1433% 2.6966% 8.9258% 2.7656%yes 0.7074% 1 / 62
2012-10-01 → window end -2.6504% 2.7554% 8.9993% -2.1515%yes 0.6856% 0 / 61

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 1.63% of 246 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2012-01-01 → 2012-04-01 -2.612% 2.3128% 7.9731% 1.4142%yes 0.6103% 2 / 61
2012-04-01 → 2012-07-01 -2.5677% 2.5653% 8.001% 0.9078%yes 0.6074% 2 / 62
2012-07-01 → 2012-10-01 -2.8085% 2.275% 7.6563% 3.4656%yes 0.6032% 0 / 62
2012-10-01 → window end -2.2789% 2.3803% 7.7199% -1.9602%yes 0.5776% 0 / 61

Step 6 · 2013-01-02 → 2013-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 10.08% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2013-01-01 → 2013-04-01 -3.4282% 2.2614% 7.4075% 1.0705%yes 0.6479% 1 / 59
2013-04-01 → 2013-07-01 -2.8025% 2.6054% 7.8% -4.8996%no 0.5675% 11 / 63
2013-07-01 → 2013-10-01 -4.2416% 1.4878% 7.012% 0.368%yes 0.6561% 7 / 63
2013-10-01 → 2014-01-01 -4.664% 1.2441% 6.9515% 0.2932%yes 0.674% 6 / 63
2014-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 7.66% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2013-01-01 → 2013-04-01 -3.1495% 1.8352% 6.3143% 0.7588%yes 0.5564% 1 / 59
2013-04-01 → 2013-07-01 -2.6153% 2.0566% 6.5129% -5.7017%no 0.4734% 9 / 63
2013-07-01 → 2013-10-01 -4.1294% 0.8571% 5.6311% 0.9458%yes 0.5274% 6 / 63
2013-10-01 → 2014-01-01 -4.6968% 0.6827% 5.8537% 0.0723%yes 0.6352% 3 / 63
2014-01-01 no segment follows this rebalance, not scored

Step 7 · 2014-01-02 → 2014-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 2.82% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2014-01-01 → 2014-04-01 -5.1009% 0.3229% 5.5189% 4.3869%yes 0.676% 1 / 60
2014-04-01 → 2014-07-01 -4.6848% 0.3891% 5.7651% 4.1966%yes 0.6758% 0 / 62
2014-07-01 → 2014-10-01 -4.3542% 0.8979% 5.9396% -0.1444%yes 0.6558% 4 / 63
2014-10-01 → 2015-01-01 -4.7461% 0.5369% 5.6107% 2.7455%yes 0.6758% 2 / 63
2015-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 4.03% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2014-01-01 → 2014-04-01 -4.8553% 0.1444% 4.9141% 3.4526%yes 0.5865% 1 / 60
2014-04-01 → 2014-07-01 -4.4335% 0.1788% 5.0424% 3.509%yes 0.5428% 0 / 62
2014-07-01 → 2014-10-01 -4.2916% 0.6913% 5.462% -2.8499%yes 0.5375% 4 / 63
2014-10-01 → 2015-01-01 -4.5666% 0.1936% 4.7421% -1.3891%yes 0.5083% 5 / 63
2015-01-01 no segment follows this rebalance, not scored

Step 8 · 2015-01-02 → 2015-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 6.85% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2015-01-01 → 2015-04-01 -4.3839% 0.7997% 5.7524% 1.1275%yes 0.7053% 6 / 60
2015-04-01 → 2015-07-01 -4.3386% 1.0029% 6.6764% -4.8767%no 0.7449% 4 / 62
2015-07-01 → 2015-10-01 -5.7094% 0.1783% 5.8686% -0.4921%yes 0.7754% 4 / 63
2015-10-01 → 2016-01-01 -4.7942% 0.8013% 6.1913% -0.7032%yes 0.7128% 3 / 63
2016-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 7.66% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2015-01-01 → 2015-04-01 -4.4423% 0.1525% 4.5182% 0.3749%yes 0.559% 6 / 60
2015-04-01 → 2015-07-01 -4.6242% 0.1086% 5.1059% -2.6752%yes 0.6047% 3 / 62
2015-07-01 → 2015-10-01 -5.3231% -0.2384% 4.6372% -2.655%yes 0.6007% 6 / 63
2015-10-01 → 2016-01-01 -5.007% -0.1115% 4.5735% -1.938%yes 0.6243% 4 / 63
2016-01-01 no segment follows this rebalance, not scored

Step 9 · 2016-01-04 → 2016-12-30

All Seasons static recipe

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 3.63% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2016-01-01 → 2016-04-01 -4.9247% 0.4667% 5.6295% 5.5776%yes 0.7106% 1 / 60
2016-04-01 → 2016-07-01 -4.3428% 1.1998% 6.5348% 5.3086%yes 0.6907% 1 / 63
2016-07-01 → 2016-10-01 -4.2008% 1.4138% 6.8214% -0.4775%yes 0.6896% 3 / 63
2016-10-01 → window end -4.4473% 0.9971% 6.7861% -5.7806%no 0.6988% 4 / 62

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 25.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 4.03% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2016-01-01 → 2016-04-01 -5.3459% -0.6008% 3.9162% 5.3643%no 0.6442% 1 / 60
2016-04-01 → 2016-07-01 -4.6456% 0.2143% 4.8627% 5.7164%no 0.5807% 1 / 63
2016-07-01 → 2016-10-01 -4.1531% 0.6804% 5.3013% -0.768%yes 0.595% 3 / 63
2016-10-01 → window end -4.3055% 0.3352% 5.23% -4.6198%no 0.5985% 5 / 62

Step 10 · 2017-01-03 → 2017-12-29

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 1.21% of 247 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2017-01-01 → 2017-04-01 -5.107% 0.0934% 6.0957% 2.2568%yes 0.7079% 0 / 61
2017-04-01 → 2017-07-01 -5.7323% -0.229% 5.6305% 1.6685%yes 0.6988% 0 / 62
2017-07-01 → 2017-10-01 -4.7603% 0.4334% 5.943% 2.0897%yes 0.6326% 1 / 62
2017-10-01 → 2018-01-01 -4.1193% 0.8405% 6.088% 3.3172%yes 0.5896% 2 / 62
2018-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 1.21% of 247 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2017-01-01 → 2017-04-01 -4.614% -0.1673% 4.9235% 2.1228%yes 0.6002% 3 / 61
2017-04-01 → 2017-07-01 -4.8795% -0.0796% 4.9926% 0.6488%yes 0.6142% 0 / 62
2017-07-01 → 2017-10-01 -4.4016% 0.2019% 5.0558% 2.6668%yes 0.57% 0 / 62
2017-10-01 → 2018-01-01 -3.6921% 0.938% 5.8195% 3.45%yes 0.5473% 0 / 62
2018-01-01 no segment follows this rebalance, not scored

Step 11 · 2018-01-02 → 2018-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 8.5% of 247 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2018-01-01 → 2018-04-01 -3.7349% 1.2348% 5.9719% -1.7522%yes 0.5661% 10 / 60
2018-04-01 → 2018-07-01 -3.7031% 1.3743% 6.2385% 1.4056%yes 0.5897% 3 / 63
2018-07-01 → 2018-10-01 -4.0162% 0.7299% 5.7403% 0.5148%yes 0.6279% 1 / 62
2018-10-01 → 2019-01-01 -4.4318% 0.1567% 4.9941% -3.2289%yes 0.6232% 7 / 62
2019-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.4× · cost drag 0.14%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 8.1% of 247 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2018-01-01 → 2018-04-01 -3.0778% 1.3098% 5.4681% -1.2399%yes 0.4942% 8 / 60
2018-04-01 → 2018-07-01 -3.1% 1.2885% 5.4644% 0.6915%yes 0.4569% 3 / 63
2018-07-01 → 2018-10-01 -3.61% 0.4402% 4.6856% 0.2538%yes 0.4672% 4 / 62
2018-10-01 → 2019-01-01 -3.999% -0.1208% 3.9379% -2.0339%yes 0.4742% 5 / 62
2019-01-01 no segment follows this rebalance, not scored

Step 12 · 2019-01-02 → 2019-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 4.03% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2019-01-01 → 2019-04-01 -4.6967% 0.1557% 4.778% 6.3603%no 0.6326% 0 / 60
2019-04-01 → 2019-07-01 -3.4207% 1.0375% 5.7292% 4.6992%yes 0.5868% 1 / 62
2019-07-01 → 2019-10-01 -3.267% 1.4921% 6.0369% 4.0532%yes 0.6025% 6 / 63
2019-10-01 → 2020-01-01 -3.3264% 1.6583% 6.4287% 0.9552%yes 0.6307% 3 / 63
2020-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 5.24% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2019-01-01 → 2019-04-01 -4.2601% -0.1837% 3.67% 4.3605%no 0.4967% 1 / 60
2019-04-01 → 2019-07-01 -3.0847% 0.643% 4.5369% 4.5155%yes 0.4634% 1 / 62
2019-07-01 → 2019-10-01 -2.8504% 1.0937% 4.8306% 3.0886%yes 0.456% 7 / 63
2019-10-01 → 2020-01-01 -2.9303% 1.2726% 5.2647% 1.1377%yes 0.4806% 4 / 63
2020-01-01 no segment follows this rebalance, not scored

Step 13 · 2020-01-02 → 2020-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.4× · cost drag 0.14%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 8.84% of 249 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2020-01-01 → 2020-04-01 -3.2121% 1.3157% 6.4999% 1.0258%yes 0.6396% 10 / 61
2020-04-01 → 2020-07-01 -5.4329% 1.2449% 8.4372% 8.5332%no 0.694% 4 / 62
2020-07-01 → 2020-10-01 -4.6171% 2.7882% 10.0429% 3.3121%yes 0.6924% 6 / 63
2020-10-01 → window end -4.5863% 2.9903% 10.4246% 2.6412%yes 0.7495% 2 / 63

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.4× · cost drag 0.14%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 50.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 9.24% of 249 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2020-01-01 → 2020-04-01 -2.8272% 1.1076% 5.5846% -3.2788%no 0.5435% 9 / 61
2020-04-01 → 2020-07-01 -4.5071% 0.6433% 6.1038% 7.068%no 0.5832% 4 / 62
2020-07-01 → 2020-10-01 -3.7913% 1.8626% 7.3087% 2.9266%yes 0.5854% 8 / 63
2020-10-01 → window end -3.4406% 2.2352% 7.7026% 2.1236%yes 0.5939% 2 / 63

Step 14 · 2021-01-04 → 2021-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 4.44% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2021-01-01 → 2021-04-01 -4.1609% 3.4114% 10.8085% -4.5391%no 0.7463% 7 / 60
2021-04-01 → 2021-07-01 -4.8128% 2.6059% 10.651% 5.5991%yes 0.8084% 1 / 62
2021-07-01 → 2021-10-01 -4.9336% 3.1302% 11.0802% 0.3159%yes 0.83% 2 / 63
2021-10-01 → 2022-01-01 -5.4829% 2.6448% 10.6662% 4.0634%yes 0.8566% 1 / 63
2022-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 5.24% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2021-01-01 → 2021-04-01 -3.1% 2.4844% 7.8365% -3.819%no 0.5885% 6 / 60
2021-04-01 → 2021-07-01 -3.7041% 1.8441% 7.747% 4.6459%yes 0.643% 2 / 62
2021-07-01 → 2021-10-01 -3.694% 2.484% 8.4639% 0.5292%yes 0.6645% 3 / 63
2021-10-01 → 2022-01-01 -4.2698% 2.1887% 8.4587% 2.5396%yes 0.7094% 2 / 63
2022-01-01 no segment follows this rebalance, not scored

Step 15 · 2022-01-03 → 2022-12-30

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 50.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 14.17% of 247 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2022-01-01 → 2022-04-01 -4.7762% 2.3892% 10.8272% -3.504%yes 0.8402% 9 / 61
2022-04-01 → 2022-07-01 -5.9766% 1.444% 10.2138% -11.1165%no 0.9487% 9 / 61
2022-07-01 → 2022-10-01 -7.3615% 0.1766% 7.5858% -8.7535%no 0.9557% 10 / 63
2022-10-01 → 2023-01-01 -9.542% -1.9869% 6.2487% 0.2292%yes 1.1134% 7 / 62
2023-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.3× · cost drag 0.13%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 50.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 13.77% of 247 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2022-01-01 → 2022-04-01 -3.7924% 1.9838% 8.6852% -0.3784%yes 0.7027% 6 / 61
2022-04-01 → 2022-07-01 -4.7304% 1.7789% 9.3926% -7.9423%no 0.7916% 11 / 61
2022-07-01 → 2022-10-01 -5.526% 1.1639% 7.6781% -8.1009%no 0.8122% 13 / 63
2022-10-01 → 2023-01-01 -7.7487% -1.0023% 6.2809% 1.4507%yes 1.0104% 4 / 62
2023-01-01 no segment follows this rebalance, not scored

Step 16 · 2023-01-03 → 2023-12-29

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 3.66% of 246 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2023-01-01 → 2023-04-01 -9.2937% -1.4479% 7.8926% 5.21%yes 1.1577% 2 / 61
2023-04-01 → 2023-07-01 -9.3273% -1.1976% 8.5104% -0.1265%yes 1.2102% 1 / 61
2023-07-01 → 2023-10-01 -9.728% -1.1201% 8.3661% -6.7158%yes 1.1953% 2 / 62
2023-10-01 → 2024-01-01 -11.2996% -2.6084% 6.9926% 9.9242%no 1.2292% 4 / 62
2024-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 2.03% of 246 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2023-01-01 → 2023-04-01 -7.2185% -0.3275% 7.7796% 4.3006%yes 1.0235% 2 / 61
2023-04-01 → 2023-07-01 -7.3069% -0.1007% 8.406% -1.8342%yes 1.0163% 1 / 61
2023-07-01 → 2023-10-01 -7.9015% -0.285% 8.0112% -3.2131%yes 1.0087% 2 / 62
2023-10-01 → 2024-01-01 -8.7736% -1.0542% 7.369% 5.7864%yes 1.0034% 0 / 62
2024-01-01 no segment follows this rebalance, not scored

Step 17 · 2024-01-02 → 2024-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 1.61% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2024-01-01 → 2024-04-01 -11.2601% -1.6781% 7.9216% 2.3499%yes 1.275% 1 / 60
2024-04-01 → 2024-07-01 -10.7279% -1.5188% 8.7022% 1.3361%yes 1.2787% 2 / 62
2024-07-01 → 2024-10-01 -9.8134% -0.1293% 9.6077% 6.715%yes 1.1981% 0 / 63
2024-10-01 → window end -8.654% 0.6344% 9.9277% -5.0723%yes 1.1267% 1 / 63

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 4 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.2× · cost drag 0.12%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 2.02% of 248 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2024-01-01 → 2024-04-01 -8.8947% -0.4853% 7.8259% 4.1749%yes 0.9845% 1 / 60
2024-04-01 → 2024-07-01 -8.1409% -0.1321% 8.6277% 1.8299%yes 0.9807% 2 / 62
2024-07-01 → 2024-10-01 -7.9135% 0.333% 8.497% 4.8501%yes 0.9682% 0 / 63
2024-10-01 → window end -6.792% 0.9802% 8.6332% -3.9653%yes 0.8862% 2 / 63

Step 18 · 2025-01-02 → 2025-12-31

All Seasons static recipe

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.1× · cost drag 0.11%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 3.25% of 246 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2025-01-01 → 2025-04-01 -6.8292% 1.6741% 9.5828% 2.3559%yes 1.0042% 0 / 59
2025-04-01 → 2025-07-01 -6.5076% 0.497% 8.743% 2.1111%yes 0.9728% 7 / 61
2025-07-01 → 2025-10-01 -7.2102% 1.1558% 9.4425% 4.4663%yes 0.9865% 0 / 63
2025-10-01 → 2026-01-01 -6.2533% 1.9517% 10.0601% 0.4734%yes 0.9142% 1 / 63
2026-01-01 no segment follows this rebalance, not scored

Equal risk contribution, unlevered

Portfolio book, rebalanced quarterly · 5 constructions · 5 names held · selection: reselect · 0.0% in cash · turnover 1.3× · cost drag 0.13%

Projection accuracy, realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (a well-calibrated 90% band contains ~90.0%) · 95% VaR breached on 2.85% of 246 days (expected ~5.0%)

Rebalance P5 Median P95 Realized In band VaR 95 (1d) Breaches
2025-01-01 → 2025-04-01 -5.5255% 1.4926% 7.9257% 3.7954%yes 0.7767% 0 / 59
2025-04-01 → 2025-07-01 -5.0299% 0.7391% 7.4369% 2.0364%yes 0.7429% 6 / 61
2025-07-01 → 2025-10-01 -5.5641% 1.1007% 7.5889% 4.4767%yes 0.7794% 0 / 63
2025-10-01 → 2026-01-01 -4.5659% 1.8982% 8.1752% 1.3747%yes 0.7302% 1 / 63
2026-01-01 no segment follows this rebalance, not scored
QuanterLab · Study 4909a6327d30 · compiled August 30, 2026. Point-in-time constituents and hypothesis-registration timestamps are enforced by the platform. This report is generated from the frozen study artifact and is reproducible from the ledger above. Educational research, not investment advice: every result on this page is simulated, and nothing here is a recommendation to buy or sell any security.

Run a study like this one

Everything above was produced inside QuanterLab, the registration, the walk, the statistics and the paper itself. Build the circuit on a canvas, register the hypothesis before you score it, and the platform enforces the rest.

Tour the lab Follow the research RSS

The lab is in private beta and opens in September 2026. Reading the research needs no account, follow it and we'll tell you when the next study publishes.

As seen on Quantocracy

A note on AI. QuanterLab is a quantitative finance research platform, and every number in this study comes from a run on the platform. The hypothesis, the parameter choices, the validation design and the conclusions belong to the author. Runs execute on point-in-time data with walk-forward validation, and each study ships with its methodology and logs, so a reader can reconstruct the result instead of trusting it. I use AI to edit and structure the prose; it does not generate results, produce numbers, or decide what a study concludes.