QuanterLab · Research

A VIX Circuit Breaker on Momentum: One Crisis Averted, Every Recovery Missed (S&P 500, 2006–2026)

A point-in-time walk-forward investigation with sealed hypotheses and the size of the search left unreconstructed rather than guessed.

By Serhat Girgin

Universe · S&P 500 (point-in-time constituents)

Method · Comparative — Arm A vs Arm B

Manipulated variable · VIX Cut

Step size · 1 year per forward window

In-sample · 2 years before each anchor

Out-of-sample span · 2006-01-03 → 2025-12-31

Compiled · August 02, 2026

Search record · none (size unknown — see §2.3)

Abstract

A volatility circuit breaker is the most intuitive risk control a systematic equity investor can bolt onto a momentum book: when the VIX crosses a panic threshold, go to cash; when it settles, return. This study tests that rule directly. Two arms run the identical combined-momentum strategy on the point-in-time S&P 500 across twenty sealed one-year out-of-sample windows from 2006 to 2026. They differ in exactly one node: Arm A carries a VIX cut that moves the book to cash above VIX 30 and re-enters below 22; Arm B carries nothing.

The cut is decisively harmful. Arm A compounds at 2.5% against Arm B's 10.0% — a gap of 7.5 percentage points a year — and a block bootstrap over 5,011 paired daily observations puts the probability that the protected arm outperforms at 0.8%. Pooled out-of-sample Sharpe is 0.225 with the cut against 0.488 without.

The interesting part is not the average. The rule never fires in nine of the twenty windows, so the entire verdict rests on eleven. Of those, it helps in two and hurts in nine. Its single clear success is 2008, where it turned a 52.4% loss into a 37.5% loss. Its single worst failure is the year immediately following: in 2009 it held the book in cash 83.7% of the time and delivered 9.6% against the unprotected arm's 41.6%. The 14.9 points bought in the crisis were repaid more than three times over in the recovery.

The mechanism is visible in the volatility. In 2022 the cut reduced realised volatility from 37.2% to 22.5% while turning a 10.9% loss into a 41.2% loss. It did exactly what it promised and removed the good days with the bad. A binary threshold on a mean-reverting index cuts after drawdowns and re-enters after rebounds, and momentum's recoveries are where its returns live.

Author’s note

This study set out to test a piece of received wisdom — that cutting exposure when the VIX spikes protects a momentum book — and found that it does the opposite over twenty years, costing 7.5 percentage points a year while genuinely lowering volatility. The one window where the rule works, 2008, is immediately undone by the recovery that follows it, and that pairing is only visible because the walk starts before the crisis rather than after it. I am publishing the negative result because the intuition it contradicts is the one most retail investors act on.

1 Methodology

UNIVERSE AND POINT-IN-TIME CONSTRUCTION. The universe is the S&P 500 as constituted at each window's anchor date, resolved point-in-time from vendor membership history rather than from today's index. Both arms use the same resolution at every step, so no survivorship difference can enter the comparison. Twenty anchors run from 2006-01-01 to 2025-01-01 inclusive, each followed by a full one-year out-of-sample window; the final window closes at 2026-01-01.

SIGNAL. Each name is scored on a combined price-momentum factor built from four horizons — trailing 1, 3, 6 and 12-month returns — each cross-sectionally winsorized and z-scored over the loaded universe at the anchor, then averaged at equal importance. The momentum family carries 100% of the composite weight; no fundamental family contributes. Momentum is computed from prices alone and requires no filing, so a name enters the cross-section on price history rather than on statement coverage.

SELECTION AND HOLDING. The top ten names by composite score are held equal-weighted. The book re-selects quarterly against the full point-in-time universe, so each rebalance re-ranks the index rather than re-ranking the incumbent holdings. Positions are long only. The benchmark is SPY. Each window runs one year forward from its anchor and the twenty windows are chained end to end into a single out-of-sample track.

THE VARIABLE UNDER TEST. Arm A and Arm B are byte-identical circuits with one exception: Arm A wires a VIX cut into the forward tester as a risk overlay. The rule is a two-level hysteresis — when the VIX closes above 30 the book moves entirely to cash, and it re-enters only once the VIX closes below 22. The asymmetric thresholds exist to prevent a single index print from toggling the book daily. Arm B receives no overlay of any kind. No regime classifier is wired in either arm, so the forward tester's regime exposure controls never engage; the cut is the sole source of cash in the study, which the per-window cash percentages confirm exactly.

SEQUENTIAL SEALING, NOT PRE-REGISTRATION. Every window in this study is historical: the data existed before the study began. No procedure applied afterwards can turn that into pre-registration in the clinical-trial sense, and this paper does not claim otherwise. What the platform does enforce is ORDER. Each step's circuit was frozen and hashed, and the resulting commitment recorded against that window's anchor, before that step was scored; the walk refuses to advance past a step whose run never landed, and refuses to close a window with a result registered for a different one. Appendix A2 prints the seal time and the run-completion time for all twenty steps at second resolution, so the interleaving — seal, run, seal, run — is checkable rather than asserted. The whole walk was executed in one session on 2026-08-02; the twenty seals are minutes apart, not months, and the table shows exactly that. What this rules out is a specification that moved mid-walk or a window scored before it was sealed. What it cannot rule out, and nothing could, is that the author had prior exposure to the historical record.

STATISTICS. The arms are compared on paired daily returns over the stitched out-of-sample track — 5,011 paired observations. The reported probability that Arm A outperforms Arm B is a stationary block bootstrap with a block length of ten trading days over 2,000 resampled paths; the block preserves the short-horizon autocorrelation that an i.i.d. bootstrap would destroy and that would otherwise overstate significance. The paired Sharpe of the difference series is reported alongside each arm's own pooled Sharpe.

WHAT IS NOT MODELLED. No transaction-cost overlay is wired, so all results are gross of commissions, spread, slippage and market impact. No deflated Sharpe ratio is claimed: this design was not promoted from a recorded search, so the number of alternatives considered before it is unknown, and an unrecorded search count cannot honestly be deflated against. Returns are price returns and exclude dividends for both the book and the benchmark.

Transaction costs are not modelled in this study; all results are gross of costs.

2 Results

2.1 Headline

Arm A — pooled Sharpe

0.23

5012 OOS bars

Arm B — pooled Sharpe

0.49

5012 OOS bars

P(Arm A beats Arm B)

0.8%

5011 paired bars · CAGR gap (Arm A − Arm B) -7.5 pp

Figure 1. Both arms stitched through the identical windows — Arm A (+64.3%), Arm B (+557.7%), benchmark grey (+402.8%). Dotted verticals mark the step boundaries; the dashed horizontal is break-even.

Figure 2. The same walk, re-based to 1.00x at the first window starting in 2010 — 16 of the 20 windows above. Arm A (+214.7%), Arm B (+785.9%), benchmark grey (+482.7%). This is a subset of Figure 1, not a correction to it. The study is anchored before the 2007–09 crisis on purpose: a method that only works in calm markets should be caught doing it. But one crisis window and its equally singular recovery set the vertical scale for the whole of Figure 1, and everything after 2010 is compressed into the bottom of it. This figure shows the same windows, same method, same data, with that period outside the frame — so the post-crisis years can be read at their own scale. Neither figure is the honest one on its own; the full record is what the study claims, and the era rows below put a number on how much of the gap came from which period.

2.2 Per-step results

**Table 1.** One row per step — raw out-of-sample results.
#	Out-of-sample window	Arm A SR	Arm B SR
1	2006-01-03 → 2006-12-29	0.09	0.09
2	2007-01-03 → 2007-12-31	-0.24	0.55
3	2008-01-02 → 2008-12-31	-1.81	-1.34
4	2009-01-02 → 2009-12-31	-0.75	1.27
5	2010-01-04 → 2010-12-31	0.67	0.67
6	2011-01-03 → 2011-12-30	-0.54	-0.44
7	2012-01-03 → 2012-12-31	0.75	0.75
8	2013-01-02 → 2013-12-31	1.92	1.92
9	2014-01-02 → 2014-12-31	0.84	0.84
10	2015-01-02 → 2015-12-31	-0.06	0.48
11	2016-01-04 → 2016-12-30	0.65	0.65
12	2017-01-03 → 2017-12-29	1.29	1.29
13	2018-01-02 → 2018-12-31	-0.28	0.10
14	2019-01-02 → 2019-12-31	1.34	1.34
15	2020-01-02 → 2020-12-31	-0.68	0.70
16	2021-01-04 → 2021-12-31	0.94	1.23
17	2022-01-03 → 2022-12-30	-2.27	-0.13
18	2023-01-03 → 2023-12-29	1.32	1.32
19	2024-01-02 → 2024-12-31	0.81	1.05
20	2025-01-02 → 2025-12-31	0.75	0.94

Figure 3. Arm A — 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.

Figure 4. Arm B — 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

No search record exists for this design. It was not promoted from a recorded evolving search, so the number of alternatives tried before it — on paper, in another tool, or in the author's head — is unknown. Unknown is a different fact from one: a study with no lineage is not a strategy with one trial, it is a strategy with an unrecorded number of them. Accordingly this paper claims no deflated Sharpe and no trial count; the honest statement is the raw out-of-sample result plus this disclosure. The registered per-step record below (§4) still guarantees each window's hypothesis was sealed before that window was scored.

2.4 The comparison

Both arms trade the same sealed windows, so their returns can be PAIRED: inside each window the two return series are inner-joined date by date and the difference r_{Arm A} − r_{Arm B} is the object under test. Because this is ONE pre-declared contrast — sealed before any window was scored — the paired statistic needs no multiple-testing deflation, and the per-arm pooled numbers above carry no deflated Sharpe either — this design has no recorded search to deflate against (§2.3). The paired contrast is the one statistic here that a missing search record does not weaken: it was declared in advance, and it is scored on the difference rather than on either arm's level.

**Table 2.** Window-by-window paired comparison. Δ is the growth gap (Arm A − Arm B) over the window's paired dates.
#	Window	Paired bars	Arm A	Arm B	Δ	Leader
1	2006-01-04 → 2006-12-29	250	-0.5%	-0.5%	+0.0 pp	tie
2	2007-01-04 → 2007-12-31	250	-7.2%	+10.7%	-17.9 pp	Arm B
3	2008-01-03 → 2008-12-31	252	-37.5%	-52.4%	+14.9 pp	Arm A
4	2009-01-05 → 2009-12-31	251	-9.6%	+41.6%	-51.1 pp	Arm B
5	2010-01-05 → 2010-12-31	251	+15.9%	+15.9%	+0.0 pp	tie
6	2011-01-04 → 2011-12-30	251	-10.9%	-16.9%	+6.0 pp	Arm A
7	2012-01-04 → 2012-12-31	249	+13.5%	+13.5%	+0.0 pp	tie
8	2013-01-03 → 2013-12-31	251	+37.5%	+37.5%	+0.0 pp	tie
9	2014-01-03 → 2014-12-31	251	+15.6%	+15.6%	+0.0 pp	tie
10	2015-01-05 → 2015-12-31	251	-2.4%	+7.5%	-9.9 pp	Arm B
11	2016-01-05 → 2016-12-30	251	+10.3%	+10.3%	+0.0 pp	tie
12	2017-01-04 → 2017-12-29	250	+20.8%	+20.8%	+0.0 pp	tie
13	2018-01-03 → 2018-12-31	250	-7.8%	-0.3%	-7.5 pp	Arm B
14	2019-01-03 → 2019-12-31	251	+19.4%	+19.4%	+0.0 pp	tie
15	2020-01-03 → 2020-12-31	252	-8.5%	+22.0%	-30.6 pp	Arm B
16	2021-01-05 → 2021-12-31	251	+24.8%	+36.1%	-11.3 pp	Arm B
17	2022-01-04 → 2022-12-30	250	-41.2%	-10.9%	-30.3 pp	Arm B
18	2023-01-04 → 2023-12-29	249	+27.7%	+27.7%	+0.0 pp	tie
19	2024-01-03 → 2024-12-31	251	+17.6%	+25.1%	-7.4 pp	Arm B
20	2025-01-03 → 2025-12-31	249	+17.1%	+26.6%	-9.6 pp	Arm B

Paired Sharpe of the difference track: -0.49 · block bootstrap (2000 paths, block 10, seed 1234): P(Arm A beats Arm B) = 0.8%.

Window win-rate. Arm A led 2 of 20 windows (10.0%), Arm B led 9 , and the mean window gap of -7.73 pp points the same way. Widest single window: 2009 at -51.1 pp.

**Table 3.** The same comparison split at 2010. Averaging across a crisis and a decade of calm hides which one the difference came from.
Period	Windows	Arm A	Arm B	Mean gap	Arm A led
All windows	20	+4.73%	+12.46%	-7.73 pp	2/20
Before 2010	4	-13.70%	-0.15%	-13.55 pp	1/4
2010 onward	16	+9.34%	+15.62%	-6.28 pp	1/16

Figure A1 — mean window return per period. Arm A above, Arm B below, with the gap at right. The pooled bar and the post-2010 bar are the same comparison over different periods.

The two eras disagree by 7.27 pp. The pooled figure is therefore not a standing property of either method — it is dominated by the earlier 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 — Arm A vs Arm B, walked on the same sealed out-of-sample windows. Arm A: S&P 500, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor — with no in-sample parameters to optimize, a walk-forward does not apply; its disposition is the realized forward path versus the benchmark. Arm B: S&P 500, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor — with no in-sample parameters to optimize, a walk-forward does not apply; its disposition is the realized forward path versus the benchmark. The arms differ in: VIX Cut — removed in Arm B. The contrast under test: whether Arm A generates better risk-adjusted returns than Arm B over the identical out-of-sample windows.

Figure 5. The frozen circuit — every node a primitive, every wire a typed data-flow; the two arms are colour-coded (Arm A green, Arm B 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. Price series and per-name figures are not published: the underlying market data is licensed to QuanterLab, and redistributing it isn't ours to do.

What each part does

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.

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

Factor Loader — Point-in-time fundamentals — never let the user see a number before the SEC did.

Factor Momentum — Momentum — recent price strength, no filing required.

Factor Composite — The weighting console — blend Value, Quality, Momentum, Growth into one 0–100 score.

Factor Top Tier — The cut out of the factor lane — keep the top-ranked names.

Vix Cut — Risk-off switch — go to cash when fear spikes.

Portfolio Backtest — Replay the portfolio forward — rebalanced, point-in-time. No transaction-cost overlay is wired in this circuit, so these results are gross of costs.

The objective and the search

Arm A — S&P 500, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor — with no in-sample parameters to optimize, a walk-forward does not apply; its disposition is the realized forward path versus the benchmark.

Universe	S&P 500 index constituents.
Validation & out-of-sample	portfolio forward test (buy-and-hold book) (1y horizon from the anchor, quarterly rebalance); overlays: VIX Cut.
Other components	Factor models: Factor Composite, Factor Select, Fundamentals Loader (PIT), Momentum Factor.

Arm B — S&P 500, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor — with no in-sample parameters to optimize, a walk-forward does not apply; its disposition is the realized forward path versus the benchmark.

Universe	S&P 500 index constituents.
Validation & out-of-sample	portfolio forward test (buy-and-hold book) (1y horizon from the anchor, quarterly rebalance).
Other components	Factor models: Factor Composite, Factor Select, Fundamentals Loader (PIT), Momentum Factor.

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

removedVIX Cut — removed in Arm B

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

No transaction-cost elements are wired into this circuit; results are gross of costs.

Show the mathematics — 8 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 Factor Loader

Point-in-time fundamentals — never let the user see a number before the SEC did.

Loads ~23 fundamental metrics (valuation, quality, growth) per name, but with one inviolable rule: a financial statement becomes visible only on or after its SEC acceptedDate. A 2020 backtest sees only what was actually filed by 2020 — no look-ahead, ever.

The PIT gate

\text{visible}(f, t) \iff \text{acceptedDate}(f) \le t

Missing acceptance dates fall back to filingDate, else statement date + 45 days.

Byte-identical to FM101FBKT (shared_libs/factor_core). US indexes only (SEC reliability).

3.4 Factor Momentum

Momentum — recent price strength, no filing required.

Pure price-based: 1/3/6/12-month returns measured at the anchor. Because it needs no financial statement, it is always available — even for names with thin fundamentals.

Multi-horizon return, z-scored

\text{Mom}_i = \frac{\sum_k \omega_k\, z\!\big(R_{i}^{(k)}\big)}{\sum_k \omega_k}, \quad R^{(k)} = \frac{P_{\mathrm{near}}}{P_{t-k}}-1

Horizons k ∈ {1, 3, 6, 12} months, measured point-in-time. The near leg follows the node’s "Most recent month" switch: include (the default) reads the anchor close, P_near = P_t — the classic J–0; skip reads one month back, P_near = P(t−21 bars) — the Fama-French J–1 — and the 1-month horizon is excluded, since its window is exactly the skipped month.

Byte-identical to FM101FBKT (shared_libs/factor_core) at the default include setting; the skip is a Primitives-side option FM101 does not run.

3.5 Factor Composite

The weighting console — blend Value, Quality, Momentum, Growth into one 0–100 score.

Where the four factor families become a single ranking. Each family score is standardized across the universe, blended with your slider weights (or the radar's suggested tilt), and min-max scaled to 0–100. Winsorizing tames outliers; z-score or percentile normalization is your choice.

Cross-sectional standardize + winsorize

z_{i,f} = \frac{x_{i,f} - \bar x_f}{s_f}\quad(\text{clipped at the 1st / 99th percentile})

Weighted blend, scaled to 0–100

C_i = \sum_f W_f\,z_{i,f}, \qquad \text{score}_i = 100\cdot\frac{C_i - \min_j C_j}{\max_j C_j - \min_j C_j}

W = your four slider weights (total 100) OR the Regime Tilt radar's suggestion. Needs ≥ 10 names, ≥ 3 valid metrics each.

Byte-identical to FM101FBKT ranking (shared_libs/factor_core.rank_stocks_at_date).

3.6 Factor Top Tier

The cut out of the factor lane — keep the top-ranked names.

Takes the composite-ranked factor set and keeps the best N, carrying the composite score, the four family scores and the point-in-time market cap for each survivor. Feed 10–20 to a direct portfolio, or 30–100 as an optimizer pool.

Rank cut

\{\, i : \operatorname{rank}(C_i) \le N\,\}, \quad C_i = \text{composite score}

Byte-identical to FM101FBKT ranking (shared_libs/factor_core).

3.7 Portfolio Backtest

Replay the portfolio forward — rebalanced, point-in-time.

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

3.8 Vix Cut

Risk-off switch — go to cash when fear spikes.

Liquidates the book when the VIX exceeds a threshold — a blunt but effective way to step aside in a market panic. An overlay on the forward tester.

Cut condition

\text{liquidate if}\quad \text{VIX}_t > v_{\text{cut}}

5 Discussion

5.1 Findings

THE HEADLINE. Over twenty out-of-sample years the protected arm compounds at 2.5% a year against 10.0% for the unprotected arm, a gap of 7.5 percentage points annualised. Pooled out-of-sample Sharpe is 0.225 with the cut and 0.488 without, on 5,012 daily observations. The paired difference series has a Sharpe of -0.49. A block bootstrap over 5,011 paired daily observations puts the probability that the protected arm outperforms at 0.8%. On this evidence the VIX cut does not merely fail to help; it is the single largest destroyer of return in the strategy.

THE RULE IS INERT MOST OF THE TIME. In nine of the twenty windows the VIX never crosses 30 and the two arms are byte-identical, producing the same return to the decimal. The entire result therefore rests on the eleven windows in which the rule actually fired. Reporting this as "twenty windows of evidence" would overstate the sample by nearly half, and any reader weighing the strength of the conclusion should weigh eleven.

OF THE ELEVEN, IT HELPS IN TWO. The cut improves the outcome in 2008 (-37.5% against -52.4%, a gain of 14.9 points) and marginally in 2011 (-10.9% against -16.9%, a gain of 6.0 points). In the other nine it costs between 7.5 and 51.1 points.

2008 AND 2009 ARE ONE EVENT. The rule's only decisive success is immediately undone by its worst failure. In 2008 it held the book in cash 46.6% of the year and saved 14.9 points against an arm that lost more than half its value. In 2009 it held the book in cash 83.7% of the year and returned -9.6% while the unprotected arm returned +41.6% — a loss of 51.1 points, the largest single-window gap in the study. Treating these as two independent observations flatters the rule badly. They are the two halves of one crisis, and across the pair the cut is 36 points behind.

IT LOWERS VOLATILITY AND DESTROYS RETURN. The 2022 window is the cleanest demonstration. The cut fired four times, held the book in cash 52.6% of the year, and reduced realised volatility from 37.2% to 22.5%. Over the same window it turned a 10.9% loss into a 41.2% loss. The rule delivered precisely the risk reduction it advertises while removing the recovery days that carry momentum's returns. A reader who judges the overlay on volatility alone would conclude it works.

WHIPSAW IS THE FAILURE MODE, NOT LATENCY. In 2022 the VIX crossed 30 on four separate occasions — 25 January at 31.2, 23 February at 31.0, 26 April at 33.5 and 26 September at 32.3 — in a year when the index oscillated around the threshold rather than spiking through it. Each crossing cut the book after a drawdown and each retreat below 22 returned it after a recovery. The rule did not fire late; it fired repeatedly, and each firing was a round trip paid at the worst two points of the cycle.

THE SHAPE OF THE CRISIS DECIDES THE OUTCOME. Comparing the three high-volatility episodes: 2008 was a persistent, grinding high-volatility regime and the cut earned 14.9 points. 2020 was a sharp spike with a near-vertical recovery — the cut fired on 27 February, held cash 69.6% of the year, and lost 30.6 points. 2022 was a sawtooth and cost 30.3 points. The rule pays only when high volatility persists long enough that being absent is genuinely correct, and it is punished whenever volatility spikes and mean-reverts. Nothing in a VIX level distinguishes these cases at the moment of crossing.

5.2 Interpretation

This study lands squarely in an unresolved argument. Barroso and Santa-Clara (2015) show that scaling a momentum book by its own realised volatility roughly halves crash risk and materially raises Sharpe, and Moreira and Muir (2017) extend volatility management across factors and find it pays broadly. Against that, Cederburg, O'Doherty, Wang and Yan (2020) test volatility-managed strategies out of sample across a wide set of factors and find that the gains largely fail to survive. The result here is an unambiguous data point for the sceptical side — but the reason matters more than the direction, because it identifies which part of the volatility management idea is doing the damage.

The successful versions in the literature scale exposure continuously by realised volatility. The rule tested here is binary and triggered by a forward-looking implied index. Those are not the same intervention. A continuous scaler reduces position size as volatility rises and restores it as volatility falls, holding some exposure throughout and never facing a re-entry decision. A binary cut with hysteresis holds either everything or nothing and must decide, at a threshold, when to return. In 2009 that decision cost 51 points. The evidence here is therefore not that volatility carries no information — Arm A's realised volatility was genuinely and substantially lower — but that a discrete on/off rule converts that information into a market-timing bet, and the timing bet loses.

Daniel and Moskowitz (2016) explain why momentum in particular is hostile to this mechanism. Momentum crashes occur in panic states and, critically, during the market's rebound from them, when the loser portfolio's beta is extreme and prior losers rally violently. The dangerous period is not the drawdown but the recovery. A VIX rule is built to be out of the market precisely during the panic and, through its resume threshold, to remain out through the early part of the rebound. It is structurally positioned to miss the exact window in which a momentum book earns back its crisis losses. The 2008-2009 pair in this study is that argument rendered in numbers: 14.9 points saved during the panic, 51.1 points forgone in the recovery.

Cooper, Gutierrez and Hameed (2004) offer the closest thing to a defence of the approach, finding that momentum profits are conditional on market state and disappear following down markets. That is a real conditioning result, but it conditions on the sign of a trailing multi-year market return, not on a volatility index crossing a level. The distinction is exactly the one this study surfaces: the shape and persistence of a volatility episode determines whether absence is correct, and the VIX level at the moment of crossing does not encode that. In 2008 volatility persisted and absence was right. In 2020 volatility spiked and reverted within weeks and absence was catastrophic. Both crossings looked identical to the rule.

There is also a lesson about study design that is worth stating plainly. This walk was first registered from a 2011 anchor by mistake. That window contains 2020 and 2022 — both cases where the cut fails badly — and none of 2008. A study run from 2011 would have produced the same directional conclusion for entirely the wrong reason, and would have had no way to observe the one regime in which the rule genuinely works or the recovery that undoes it. The 2006 anchor is what makes the mechanism visible rather than merely the verdict. Researchers testing crisis-contingent rules should choose the sample by which crises it must contain, not by convenience.

What would follow from this? The natural next tests are a continuous volatility scaler in place of the binary cut, which is the construction the supportive literature actually uses; an asymmetric rule that re-enters far faster than it exits, on the reasoning that the cost is concentrated in the recovery rather than the exit; and a version with transaction costs modelled, which this study omits and which would penalise the protected arm further. Each is a single-node change to this circuit and can be run against the identical twenty sealed windows, which is the point of registering the walk this way.

The practical conclusion for an investor is narrow and firm. On a long-only, concentrated, quarterly-rebalanced S&P 500 momentum book over 2006 to 2026, a VIX 30/22 circuit breaker reduced volatility by roughly a third and cost 7.5 percentage points of annual return, with a 0.8% bootstrap probability of having helped. The intuition that buying protection in panics is prudent is not wrong about the panic. It is wrong about what happens next.

No search record exists for this study: the design was not promoted from a recorded evolving search, so the number of alternatives tried before it is UNKNOWN — which is a different fact from one. No deflated Sharpe is claimed; the honest statement is the raw out-of-sample result plus this disclosure. The out-of-sample windows are historical.

5.3 Limitations

GROSS OF COSTS, AND THE BIAS RUNS AGAINST THE CONCLUSION'S FAVOUR. No transaction-cost overlay is wired, so neither arm pays spread, commission, slippage or market impact. This is not neutral between the arms: the protected arm liquidates and rebuilds a ten-name book at every threshold crossing, on top of the quarterly re-selection both arms pay. In 2022 that is four additional round trips of the entire portfolio. Charging realistic costs would widen the gap against the VIX cut, so the 7.5 point annual deficit reported here is, if anything, generous to the rule.

ONE THRESHOLD PAIR, NOT A SURFACE. The rule is tested at a single configuration: cut above 30, resume below 22. These were fixed before the study ran and never tuned, which protects the result from selection bias but also means the study speaks to this rule rather than to volatility gating in general. A higher cut threshold, a faster re-entry, or a partial rather than binary de-risking might behave differently, and this design cannot say by how much. The finding is that this rule fails, not that no such rule can work.

ELEVEN EFFECTIVE WINDOWS. Nine of the twenty windows never trigger the rule and carry no information about it. Eleven is a small sample for a conclusion about crisis behaviour, and it contains only three genuine volatility crises.

NO SEARCH RECORD, SO NO DEFLATED SHARPE. This design was not promoted from a recorded evolving search, so the number of alternatives tried before it is unknown rather than one. No deflated Sharpe ratio is claimed. The honest statement is the raw out-of-sample result together with this disclosure.

CONCENTRATION AND LONG-ONLY CONSTRUCTION. The book holds ten equal-weighted names, which is far more concentrated than the decile portfolios of the academic momentum literature and carries substantial idiosyncratic risk in every window. The strategy is also long only, so it omits the short leg where Daniel and Moskowitz locate most of momentum's crash risk. Results should not be read as a replication of published momentum premia.

THE PROJECTION-CALIBRATION RECORD DOES NOT SPEAK TO THE VARIABLE, AND IS OMITTED. The forward tester fits a Monte Carlo cone and a VaR estimate before each rebalance and scores them against the outcome. In this study that record is IDENTICAL across the two arms, to the integer, and the reason is structural rather than a coincidence worth reading into: the projection is fitted on the SELECTED BASKET, and both arms select identically — the VIX cut acts afterwards, on exposure. So the record measures the accuracy of the selection's projections and is silent on the overlay, which is the one thing this paper is about. It is therefore computed but not featured, and the computed-not-featured ledger records that choice. Reported for completeness: 80 scored rebalances per arm, 86.2% band coverage against an expected 90%, and a 6.77% VaR breach rate against an expected 5% — both properties of the basket, not of either arm's book. A calibration record that follows the exposure actually held is a separate measurement and would require re-running the walk.

PRICE RETURNS ONLY. Both the book and the benchmark exclude dividends. Over a twenty-year horizon this understates absolute returns for both arms roughly equally, but it is not exactly neutral for an arm that spends large fractions of some years in cash.

THE ONE-MONTH HORIZON IS INCLUDED. The composite weights the trailing one-month return equally with the longer horizons. The reversal literature finds this horizon predicts negatively at the cross-section, so its inclusion is likely a drag on both arms. Because both arms share it, the comparison is unaffected, but the absolute level of both is lower than a conventional 12-1 construction would produce.

Computed, not featured

The run produced the results below. The author chose not to feature them in this paper; they were computed all the same, are part of the frozen record, and are reproducible from the study's sealed circuit.

Projection calibration — 160 rebalance projections, 9,902 VaR days. The forward tester itself. Every portfolio backtest fits a Monte Carlo cone and a VaR estimate before each rebalance and scores them against what happened — it is instrumentation on the test, not a primitive on the canvas. No Monte Carlo primitive is required, and adding one would be a separate, standalone analysis.

References

As provided by QuanterLab

Bailey, D. H., & López de Prado, M. (2014). The Deflated Sharpe Ratio: Correcting for Selection Bias, Backtest Overfitting, and Non-Normality. Journal of Portfolio Management, 40(5), 94–107. doi:10.3905/jpm.2014.40.5.094
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.
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
Lo, A. W. (2002). The Statistics of Sharpe Ratios. Financial Analysts Journal, 58(4), 36–52. doi:10.2469/faj.v58.n4.2453

Added by the author?

Jegadeesh, N. and Titman, S. (1993). "Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency." Journal of Finance 48(1), 65-91. The origin of the cross-sectional momentum strategy tested here.
Daniel, K. and Moskowitz, T. J. (2016). "Momentum Crashes." Journal of Financial Economics 122(2), 221-247. Locates momentum’s left tail in panic states and the rebounds that follow them — the mechanism behind this study’s 2008-2009 pair.
Barroso, P. and Santa-Clara, P. (2015). "Momentum Has Its Moments." Journal of Financial Economics 116(1), 111-120. Scaling momentum by its own realised volatility halves crash risk; the continuous analogue of the binary rule tested here.
Moreira, A. and Muir, T. (2017). "Volatility-Managed Portfolios." Journal of Finance 72(4), 1611-1644. Extends volatility management across factors and finds it pays.
Cederburg, S., O’Doherty, M. S., Wang, F. and Yan, X. S. (2020). "On the Performance of Volatility-Managed Portfolios." Journal of Financial Economics 138(1), 95-117. Finds the gains largely fail out of sample; the counterweight to the two above.
Cooper, M. J., Gutierrez, R. C. and Hameed, A. (2004). "Market States and Momentum." Journal of Finance 59(3), 1345-1365. Momentum profits are conditional on market state — the academic precedent for gating momentum exposure at all.
Novy-Marx, R. (2012). "Is Momentum Really Momentum?" Journal of Financial Economics 103(3), 429-453. Argues the premium sits in intermediate-horizon past returns, which bears on this study’s equal weighting of the 1, 3, 6 and 12-month horizons.
Goyal, A. and Wahal, S. (2015). "Is Momentum an Echo?" Journal of Financial and Quantitative Analysis 50(6), 1237-1262. International evidence against the echo interpretation; the controversy this study’s combined score sits inside.
Jegadeesh, N. (1990). "Evidence of Predictable Behavior of Security Returns." Journal of Finance 45(3), 881-898. Short-horizon reversal — why including the one-month return with a positive sign is a known drag on both arms.
Israel, R. and Moskowitz, T. J. (2013). "The Role of Shorting, Firm Size, and Time on Market Anomalies." Journal of Financial Economics 108(2), 275-301. Momentum survives long-only implementation at reduced size — relevant to this book’s construction.
Whaley, R. E. (2009). "Understanding the VIX." Journal of Portfolio Management 35(3), 98-105. What the index measures and what it does not — the basis for reading a threshold crossing as a signal about the future rather than the recent past.
Bailey, D. H. and López de Prado, M. (2014). "The Deflated Sharpe Ratio: Correcting for Selection Bias, Backtest Overfitting and Non-Normality." Journal of Portfolio Management 40(5), 94-107. Why no deflated Sharpe is claimed without a search record.

Appendix A Reproducibility in QuanterLab

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

#	Commit	Report	Anchor	OOS window
1	203e7ae4468b	444	2006-01-01	2006-01-03 → 2006-12-29
2	0239a6609b05	445	2007-01-01	2007-01-03 → 2007-12-31
3	a4cb3ce259c7	446	2008-01-01	2008-01-02 → 2008-12-31
4	fac497d877fc	447	2009-01-01	2009-01-02 → 2009-12-31
5	c6388de6937b	448	2010-01-01	2010-01-04 → 2010-12-31
6	697b96bbdbd5	449	2011-01-01	2011-01-03 → 2011-12-30
7	aa15e88466fa	450	2012-01-01	2012-01-03 → 2012-12-31
8	462f4358beaf	451	2013-01-01	2013-01-02 → 2013-12-31
9	15add0943567	452	2014-01-01	2014-01-02 → 2014-12-31
10	0b308d4b22df	453	2015-01-01	2015-01-02 → 2015-12-31
11	af89f3002126	454	2016-01-01	2016-01-04 → 2016-12-30
12	c8e669665eb9	455	2017-01-01	2017-01-03 → 2017-12-29
13	6426ab562d7c	456	2018-01-01	2018-01-02 → 2018-12-31
14	47f3316125ef	457	2019-01-01	2019-01-02 → 2019-12-31
15	41860298994b	458	2020-01-01	2020-01-02 → 2020-12-31
16	4b1604710b1c	459	2021-01-01	2021-01-04 → 2021-12-31
17	774c622a2385	460	2022-01-01	2022-01-03 → 2022-12-30
18	710c0fba1255	461	2023-01-01	2023-01-03 → 2023-12-29
19	caf865eae653	462	2024-01-01	2024-01-02 → 2024-12-31
20	b98738f6dfc2	463	2025-01-01	2025-01-02 → 2025-12-31

Appendix A2 Sealed-hypothesis 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.

“A COMPARATIVE study — Arm A vs Arm B, walked on the same sealed out-of-sample windows. Arm A: S&P 500, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor — with no in-sample parameters to optimize, a walk-forward does not apply; its disposition is the realized forward path versus the benchmark. Arm B: S&P 500, rebalanced quarterly across the selected basket, and run out-of-sample from the anchor — with no in-sample parameters to optimize, a walk-forward does not apply; its disposition is the realized forward path versus the benchmark. The arms differ in: VIX Cut — removed in Arm B. The contrast under test: whether Arm A generates better risk-adjusted returns than Arm B over the identical out-of-sample windows.”

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

**Table 4.** Registration audit — one row per sealed 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 sealed once and re-sealed 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.
#	Anchor	Sealed at (UTC)	Run completed (UTC)
1	2006-01-01	2026-08-02 12:47:42	2026-08-02 13:09:03
2	2007-01-01	2026-08-02 13:09:03	2026-08-02 13:20:44
3	2008-01-01	2026-08-02 13:20:44	2026-08-02 13:26:05
4	2009-01-01	2026-08-02 13:26:05	2026-08-02 13:32:26
5	2010-01-01	2026-08-02 13:32:26	2026-08-02 13:38:26
6	2011-01-01	2026-08-02 13:38:26	2026-08-02 13:44:07
7	2012-01-01	2026-08-02 13:44:07	2026-08-02 13:49:28
8	2013-01-01	2026-08-02 13:49:28	2026-08-02 13:56:28
9	2014-01-01	2026-08-02 13:56:29	2026-08-02 14:02:49
10	2015-01-01	2026-08-02 14:02:49	2026-08-02 14:09:31
11	2016-01-01	2026-08-02 14:09:31	2026-08-02 14:15:33
12	2017-01-01	2026-08-02 14:15:33	2026-08-02 14:22:35
13	2018-01-01	2026-08-02 14:22:35	2026-08-02 14:29:16
14	2019-01-01	2026-08-02 14:29:16	2026-08-02 14:35:57
15	2020-01-01	2026-08-02 14:35:57	2026-08-02 14:41:37
16	2021-01-01	2026-08-02 14:41:37	2026-08-02 14:48:38
17	2022-01-01	2026-08-02 14:48:38	2026-08-02 14:54:59
18	2023-01-01	2026-08-02 14:54:59	2026-08-02 15:01:22
19	2024-01-01	2026-08-02 15:01:22	2026-08-02 15:06:43
20	2025-01-01	2026-08-02 15:06:43	2026-08-02 15:13:04

Appendix B Per-step diagnostics

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.

Step 1 · 2006-01-03 → 2006-12-29

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 17 names held · selection: reselect · 0.0% in cash · turnover 5.3× · 13 names dropped from the held union of 49 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 9.72% of 247 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2006-01-01	-5.6711%	8.7259%	26.96%	18.7356%	yes	1.8155%	3 / 61
2006-04-01	-9.5465%	8.3469%	29.9161%	-13.1691%	no	2.2576%	9 / 62
2006-07-01	-6.9501%	9.0459%	27.9096%	-3.9465%	yes	2.0156%	5 / 62
2006-10-01	-3.0828%	7.4064%	19.1019%	-0.4185%	yes	1.1043%	7 / 62
2007-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 17 names held · selection: reselect · 0.0% in cash · 13 names dropped from the held union of 49 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2006-01-01	-5.6711%	8.7259%	26.96%	18.7356%	yes	1.8155%	3 / 61
2006-04-01	-9.5465%	8.3469%	29.9161%	-13.1691%	no	2.2576%	9 / 62
2006-07-01	-6.9501%	9.0459%	27.9096%	-3.9465%	yes	2.0156%	5 / 62
2006-10-01	-3.0828%	7.4064%	19.1019%	-0.4185%	yes	1.1043%	7 / 62
2007-01-01	no segment follows this rebalance — not scored

Step 2 · 2007-01-03 → 2007-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 9.6% in cash · turnover 9.4× · 9 names dropped from the held union of 50 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 12.96% of 247 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2007-01-01	-5.9336%	9.7075%	26.1617%	6.7673%	yes	1.6983%	5 / 60
2007-04-01	-5.8482%	8.3658%	24.8276%	1.5929%	yes	1.6134%	5 / 62
2007-07-01	-5.4709%	9.8518%	27.7696%	5.2996%	yes	1.7704%	10 / 62
2007-10-01	-6.7457%	9.4828%	26.7356%	-7.2494%	no	1.7314%	12 / 63
2008-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · 9 names dropped from the held union of 50 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2007-01-01	-5.9336%	9.7075%	26.1617%	6.7673%	yes	1.6983%	5 / 60
2007-04-01	-5.8482%	8.3658%	24.8276%	1.5929%	yes	1.6134%	5 / 62
2007-07-01	-5.4709%	9.8518%	27.7696%	5.2996%	yes	1.7704%	10 / 62
2007-10-01	-6.7457%	9.4828%	26.7356%	-7.2494%	no	1.7314%	12 / 63
2008-01-01	no segment follows this rebalance — not scored

Step 3 · 2008-01-02 → 2008-12-31

Arm A

Portfolio book — rebalanced quarterly · 4 constructions · 19 names held · selection: reselect · 46.6% in cash · turnover 6.1× · 4 names dropped from the held union of 46 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 25.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 18.47% of 249 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2008-01-01	-7.1626%	9.7817%	27.843%	-17.034%	no	1.9698%	11 / 60
2008-04-01	-11.9681%	6.3827%	26.4369%	5.8398%	yes	2.3508%	1 / 63
2008-07-01	-12.6244%	10.1217%	35.9953%	-29.2969%	no	2.6135%	12 / 63
2008-10-01	-14.2329%	3.5296%	22.9203%	-24.3067%	no	2.2185%	22 / 63

Arm B

Portfolio book — rebalanced quarterly · 4 constructions · 19 names held · selection: reselect · 0.0% in cash · 4 names dropped from the held union of 59 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2008-01-01	-7.1626%	9.7817%	27.843%	-17.034%	no	1.9698%	11 / 60
2008-04-01	-11.9681%	6.3827%	26.4369%	5.8398%	yes	2.3508%	1 / 63
2008-07-01	-12.6244%	10.1217%	35.9953%	-29.2969%	no	2.6135%	12 / 63
2008-10-01	-14.2329%	3.5296%	22.9203%	-24.3067%	no	2.2185%	22 / 63

Step 4 · 2009-01-02 → 2009-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 83.7% in cash · turnover 2.7× · 13 names dropped from the held union of 16 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~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
2009-01-01	-19.195%	1.3654%	24.5475%	-8.2351%	yes	2.3079%	8 / 60
2009-04-01	-21.1152%	0.828%	29.0592%	4.1311%	yes	2.554%	0 / 62
2009-07-01	-45.9854%	-4.7965%	59.6469%	37.7637%	yes	6.8133%	0 / 63
2009-10-01	-40.9083%	0.835%	64.1716%	13.7356%	yes	6.339%	0 / 63
2010-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · 13 names dropped from the held union of 52 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2009-01-01	-19.195%	1.3654%	24.5475%	-8.2351%	yes	2.3079%	8 / 60
2009-04-01	-21.1152%	0.828%	29.0592%	4.1311%	yes	2.554%	0 / 62
2009-07-01	-45.9854%	-4.7965%	59.6469%	37.7637%	yes	6.8133%	0 / 63
2009-10-01	-40.9083%	0.835%	64.1716%	13.7356%	yes	6.339%	0 / 63
2010-01-01	no segment follows this rebalance — not scored

Step 5 · 2010-01-04 → 2010-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · turnover 5.5× · 7 names dropped from the held union of 50 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 0.0% of 248 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2010-01-01	-35.9838%	2.6348%	57.5967%	9.3484%	yes	5.8054%	0 / 60
2010-04-01	-34.3395%	5.7949%	70.9344%	-16.1643%	yes	5.8669%	0 / 62
2010-07-01	-36.9463%	4.2858%	65.0189%	15.0908%	yes	5.3897%	0 / 63
2010-10-01	-29.6911%	8.4474%	61.0221%	8.1188%	yes	5.0324%	0 / 63
2011-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · 7 names dropped from the held union of 50 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2010-01-01	-35.9838%	2.6348%	57.5967%	9.3484%	yes	5.8054%	0 / 60
2010-04-01	-34.3395%	5.7949%	70.9344%	-16.1643%	yes	5.8669%	0 / 62
2010-07-01	-36.9463%	4.2858%	65.0189%	15.0908%	yes	5.3897%	0 / 63
2010-10-01	-29.6911%	8.4474%	61.0221%	8.1188%	yes	5.0324%	0 / 63
2011-01-01	no segment follows this rebalance — not scored

Step 6 · 2011-01-03 → 2011-12-30

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 38.5% in cash · turnover 5.9× · 8 names dropped from the held union of 51 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 8.87% of 248 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2011-01-01	-15.9051%	15.015%	61.8771%	4.9252%	yes	4.2092%	0 / 61
2011-04-01	-15.045%	15.2821%	56.7131%	-5.846%	yes	3.6646%	0 / 62
2011-07-01	-8.6041%	12.7895%	36.6414%	-21.6024%	no	2.329%	14 / 63
2011-10-01	-11.0296%	8.3856%	32.1891%	8.5507%	yes	2.2143%	8 / 62
2012-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · 8 names dropped from the held union of 51 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2011-01-01	-15.9051%	15.015%	61.8771%	4.9252%	yes	4.2092%	0 / 61
2011-04-01	-15.045%	15.2821%	56.7131%	-5.846%	yes	3.6646%	0 / 62
2011-07-01	-8.6041%	12.7895%	36.6414%	-21.6024%	no	2.329%	14 / 63
2011-10-01	-11.0296%	8.3856%	32.1891%	8.5507%	yes	2.2143%	8 / 62
2012-01-01	no segment follows this rebalance — not scored

Step 7 · 2012-01-03 → 2012-12-31

Arm A

Portfolio book — rebalanced quarterly · 4 constructions · 19 names held · selection: reselect · 0.0% in cash · turnover 5.6× · 3 names dropped from the held union of 60 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~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	-10.8271%	7.1378%	30.9023%	8.2036%	yes	2.2986%	0 / 61
2012-04-01	-15.4307%	7.8198%	37.6568%	-7.7518%	yes	2.9953%	5 / 62
2012-07-01	-11.7939%	7.4488%	31.0397%	7.2462%	yes	2.2521%	0 / 62
2012-10-01	-11.0764%	8.3312%	34.3806%	4.7615%	yes	2.3917%	0 / 61

Arm B

Portfolio book — rebalanced quarterly · 4 constructions · 19 names held · selection: reselect · 0.0% in cash · 3 names dropped from the held union of 60 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2012-01-01	-10.8271%	7.1378%	30.9023%	8.2036%	yes	2.2986%	0 / 61
2012-04-01	-15.4307%	7.8198%	37.6568%	-7.7518%	yes	2.9953%	5 / 62
2012-07-01	-11.7939%	7.4488%	31.0397%	7.2462%	yes	2.2521%	0 / 62
2012-10-01	-11.0764%	8.3312%	34.3806%	4.7615%	yes	2.3917%	0 / 61

Step 8 · 2013-01-02 → 2013-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 20 names held · selection: reselect · 0.0% in cash · turnover 5.4×

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~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
2013-01-01	-15.4202%	6.0444%	28.7447%	11.8898%	yes	2.5495%	2 / 59
2013-04-01	-13.0306%	8.493%	32.7383%	1.4283%	yes	2.5442%	3 / 63
2013-07-01	-16.6667%	6.5401%	33.3%	10.936%	yes	2.7129%	1 / 63
2013-10-01	-3.4484%	14.0642%	32.7935%	7.0947%	yes	1.7194%	4 / 63
2014-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 20 names held · selection: reselect · 0.0% in cash

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2013-01-01	-15.4202%	6.0444%	28.7447%	11.8898%	yes	2.5495%	2 / 59
2013-04-01	-13.0306%	8.493%	32.7383%	1.4283%	yes	2.5442%	3 / 63
2013-07-01	-16.6667%	6.5401%	33.3%	10.936%	yes	2.7129%	1 / 63
2013-10-01	-3.4484%	14.0642%	32.7935%	7.0947%	yes	1.7194%	4 / 63
2014-01-01	no segment follows this rebalance — not scored

Step 9 · 2014-01-02 → 2014-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 20 names held · selection: reselect · 0.0% in cash · turnover 4.8×

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 7.26% of 248 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2014-01-01	-5.8717%	11.2571%	29.507%	7.2665%	yes	1.9675%	2 / 60
2014-04-01	-1.8901%	15.1559%	35.2894%	5.9766%	yes	1.7447%	6 / 62
2014-07-01	-5.9303%	11.0243%	29.14%	-3.0479%	yes	1.9302%	2 / 63
2014-10-01	-2.2098%	12.1278%	27.0307%	2.424%	yes	1.611%	8 / 63
2015-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 20 names held · selection: reselect · 0.0% in cash

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2014-01-01	-5.8717%	11.2571%	29.507%	7.2665%	yes	1.9675%	2 / 60
2014-04-01	-1.8901%	15.1559%	35.2894%	5.9766%	yes	1.7447%	6 / 62
2014-07-01	-5.9303%	11.0243%	29.14%	-3.0479%	yes	1.9302%	2 / 63
2014-10-01	-2.2098%	12.1278%	27.0307%	2.424%	yes	1.611%	8 / 63
2015-01-01	no segment follows this rebalance — not scored

Step 10 · 2015-01-02 → 2015-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 19 names held · selection: reselect · 6.3% in cash · turnover 6.7× · 4 names dropped from the held union of 51 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (an honest 90% band would contain ~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
2015-01-01	-2.7267%	11.0732%	25.3011%	8.4953%	yes	1.4523%	3 / 60
2015-04-01	-0.9964%	11.1723%	24.9209%	-0.2868%	yes	1.3866%	3 / 62
2015-07-01	-3.8166%	10.3218%	25.0223%	-5.7082%	no	1.6293%	6 / 63
2015-10-01	-6.089%	7.7988%	22.2493%	5.1729%	yes	1.5706%	4 / 63
2016-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 19 names held · selection: reselect · 0.0% in cash · 4 names dropped from the held union of 51 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2015-01-01	-2.7267%	11.0732%	25.3011%	8.4953%	yes	1.4523%	3 / 60
2015-04-01	-0.9964%	11.1723%	24.9209%	-0.2868%	yes	1.3866%	3 / 62
2015-07-01	-3.8166%	10.3218%	25.0223%	-5.7082%	no	1.6293%	6 / 63
2015-10-01	-6.089%	7.7988%	22.2493%	5.1729%	yes	1.5706%	4 / 63
2016-01-01	no segment follows this rebalance — not scored

Step 11 · 2016-01-04 → 2016-12-30

Arm A

Portfolio book — rebalanced quarterly · 4 constructions · 18 names held · selection: reselect · 0.0% in cash · turnover 5.7× · 4 names dropped from the held union of 60 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~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
2016-01-01	-6.5483%	8.5585%	24.3907%	-1.5155%	yes	1.6586%	9 / 60
2016-04-01	-6.3077%	5.4801%	17.5205%	3.3157%	yes	1.2573%	5 / 63
2016-07-01	-10.9142%	4.9032%	21.7671%	3.28%	yes	1.947%	2 / 63
2016-10-01	-12.35%	3.3747%	22.0376%	4.7531%	yes	2.0907%	1 / 62

Arm B

Portfolio book — rebalanced quarterly · 4 constructions · 18 names held · selection: reselect · 0.0% in cash · 4 names dropped from the held union of 60 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2016-01-01	-6.5483%	8.5585%	24.3907%	-1.5155%	yes	1.6586%	9 / 60
2016-04-01	-6.3077%	5.4801%	17.5205%	3.3157%	yes	1.2573%	5 / 63
2016-07-01	-10.9142%	4.9032%	21.7671%	3.28%	yes	1.947%	2 / 63
2016-10-01	-12.35%	3.3747%	22.0376%	4.7531%	yes	2.0907%	1 / 62

Step 12 · 2017-01-03 → 2017-12-29

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 19 names held · selection: reselect · 0.0% in cash · turnover 4.9× · 5 names dropped from the held union of 49 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 3.24% of 247 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2017-01-01	-14.5599%	2.9763%	26.2503%	-0.3768%	yes	2.3772%	2 / 61
2017-04-01	-13.6344%	4.9336%	27.638%	-0.2865%	yes	2.6614%	1 / 62
2017-07-01	-6.5353%	9.0714%	27.3985%	15.3986%	yes	1.941%	3 / 62
2017-10-01	-7.6575%	8.6135%	27.8722%	4.8924%	yes	2.0841%	2 / 62
2018-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 19 names held · selection: reselect · 0.0% in cash · 5 names dropped from the held union of 49 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2017-01-01	-14.5599%	2.9763%	26.2503%	-0.3768%	yes	2.3772%	2 / 61
2017-04-01	-13.6344%	4.9336%	27.638%	-0.2865%	yes	2.6614%	1 / 62
2017-07-01	-6.5353%	9.0714%	27.3985%	15.3986%	yes	1.941%	3 / 62
2017-10-01	-7.6575%	8.6135%	27.8722%	4.8924%	yes	2.0841%	2 / 62
2018-01-01	no segment follows this rebalance — not scored

Step 13 · 2018-01-02 → 2018-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 5.2% in cash · turnover 8.1× · 9 names dropped from the held union of 50 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 12.55% of 247 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2018-01-01	-5.1623%	8.4026%	22.4017%	4.5438%	yes	1.5706%	6 / 60
2018-04-01	-2.7493%	13.1846%	29.9841%	6.3045%	yes	1.652%	6 / 63
2018-07-01	-4.9239%	10.0943%	27.5934%	12.1033%	yes	1.7125%	3 / 62
2018-10-01	-2.4448%	10.1438%	24.4446%	-19.0559%	no	1.2777%	16 / 62
2019-01-01	no segment follows this rebalance — not scored

Arm B

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2018-01-01	-5.1623%	8.4026%	22.4017%	4.5438%	yes	1.5706%	6 / 60
2018-04-01	-2.7493%	13.1846%	29.9841%	6.3045%	yes	1.652%	6 / 63
2018-07-01	-4.9239%	10.0943%	27.5934%	12.1033%	yes	1.7125%	3 / 62
2018-10-01	-2.4448%	10.1438%	24.4446%	-19.0559%	no	1.2777%	16 / 62
2019-01-01	no segment follows this rebalance — not scored

Step 14 · 2019-01-02 → 2019-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · turnover 5.7× · 15 names dropped from the held union of 49 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 75.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 5.65% of 248 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2019-01-01	-5.776%	5.0647%	15.9915%	16.6919%	no	1.2424%	0 / 60
2019-04-01	-6.0794%	6.5053%	20.8645%	-2.5412%	yes	1.5203%	5 / 62
2019-07-01	-5.8564%	6.6209%	19.4382%	1.3964%	yes	1.4201%	6 / 63
2019-10-01	-4.7135%	7.3328%	19.6436%	3.6832%	yes	1.4281%	3 / 63
2020-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · 15 names dropped from the held union of 49 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2019-01-01	-5.776%	5.0647%	15.9915%	16.6919%	no	1.2424%	0 / 60
2019-04-01	-6.0794%	6.5053%	20.8645%	-2.5412%	yes	1.5203%	5 / 62
2019-07-01	-5.8564%	6.6209%	19.4382%	1.3964%	yes	1.4201%	6 / 63
2019-10-01	-4.7135%	7.3328%	19.6436%	3.6832%	yes	1.4281%	3 / 63
2020-01-01	no segment follows this rebalance — not scored

Step 15 · 2020-01-02 → 2020-12-31

Arm A

Portfolio book — rebalanced quarterly · 4 constructions · 20 names held · selection: reselect · 69.6% in cash · turnover 4.6×

Projection accuracy — realized outcome fell inside the P5–P95 cone in 50.0% of 4 scored rebalances (an honest 90% band would contain ~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	-10.0949%	5.4112%	25.4048%	-20.597%	no	2.1992%	14 / 61
2020-04-01	-16.5217%	3.6807%	28.9346%	31.9345%	no	2.4297%	3 / 62
2020-07-01	-12.792%	9.7038%	35.2468%	9.0022%	yes	2.3531%	3 / 63
2020-10-01	-19.1278%	8.0666%	40.7729%	11.4246%	yes	3.1457%	2 / 63

Arm B

Portfolio book — rebalanced quarterly · 4 constructions · 20 names held · selection: reselect · 0.0% in cash

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2020-01-01	-10.0949%	5.4112%	25.4048%	-20.597%	no	2.1992%	14 / 61
2020-04-01	-16.5217%	3.6807%	28.9346%	31.9345%	no	2.4297%	3 / 62
2020-07-01	-12.792%	9.7038%	35.2468%	9.0022%	yes	2.3531%	3 / 63
2020-10-01	-19.1278%	8.0666%	40.7729%	11.4246%	yes	3.1457%	2 / 63

Step 16 · 2021-01-04 → 2021-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 20 names held · selection: reselect · 4.0% in cash · turnover 10.6× · 5 names dropped from the held union of 54 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 1.21% of 248 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2021-01-01	-21.2293%	9.9512%	48.8629%	15.4513%	yes	3.4576%	2 / 60
2021-04-01	-28.8482%	8.429%	65.6413%	7.0911%	yes	3.9245%	0 / 62
2021-07-01	-29.5627%	7.6009%	58.3636%	-0.6734%	yes	3.9184%	0 / 63
2021-10-01	-10.1903%	18.0353%	51.4497%	4.8996%	yes	2.7712%	1 / 63
2022-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 20 names held · selection: reselect · 0.0% in cash · 5 names dropped from the held union of 54 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2021-01-01	-21.2293%	9.9512%	48.8629%	15.4513%	yes	3.4576%	2 / 60
2021-04-01	-28.8482%	8.429%	65.6413%	7.0911%	yes	3.9245%	0 / 62
2021-07-01	-29.5627%	7.6009%	58.3636%	-0.6734%	yes	3.9184%	0 / 63
2021-10-01	-10.1903%	18.0353%	51.4497%	4.8996%	yes	2.7712%	1 / 63
2022-01-01	no segment follows this rebalance — not scored

Step 17 · 2022-01-03 → 2022-12-30

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 20 names held · selection: reselect · 52.6% in cash · turnover 8.5×

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 7.29% of 247 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2022-01-01	-12.8441%	15.3645%	56.6728%	-4.3452%	yes	2.8731%	6 / 61
2022-04-01	-25.7087%	9.2357%	66.3853%	-13.8819%	yes	4.0687%	7 / 61
2022-07-01	-23.2613%	12.2015%	58.6655%	-1.2755%	yes	3.9072%	3 / 63
2022-10-01	-13.7707%	12.39%	46.7133%	2.9682%	yes	2.9624%	2 / 62
2023-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 20 names held · selection: reselect · 0.0% in cash

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2022-01-01	-12.8441%	15.3645%	56.6728%	-4.3452%	yes	2.8731%	6 / 61
2022-04-01	-25.7087%	9.2357%	66.3853%	-13.8819%	yes	4.0687%	7 / 61
2022-07-01	-23.2613%	12.2015%	58.6655%	-1.2755%	yes	3.9072%	3 / 63
2022-10-01	-13.7707%	12.39%	46.7133%	2.9682%	yes	2.9624%	2 / 62
2023-01-01	no segment follows this rebalance — not scored

Step 18 · 2023-01-03 → 2023-12-29

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · turnover 5.4× · 6 names dropped from the held union of 55 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~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
2023-01-01	-11.0551%	13.0297%	46.8214%	1.4062%	yes	2.6613%	1 / 61
2023-04-01	-12.0882%	7.4148%	33.6735%	8.3123%	yes	2.4506%	1 / 61
2023-07-01	-17.4584%	4.9158%	33.5411%	-1.9845%	yes	2.9243%	0 / 62
2023-10-01	-15.1206%	3.9442%	27.4411%	15.0229%	yes	2.2219%	2 / 62
2024-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · 6 names dropped from the held union of 55 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2023-01-01	-11.0551%	13.0297%	46.8214%	1.4062%	yes	2.6613%	1 / 61
2023-04-01	-12.0882%	7.4148%	33.6735%	8.3123%	yes	2.4506%	1 / 61
2023-07-01	-17.4584%	4.9158%	33.5411%	-1.9845%	yes	2.9243%	0 / 62
2023-10-01	-15.1206%	3.9442%	27.4411%	15.0229%	yes	2.2219%	2 / 62
2024-01-01	no segment follows this rebalance — not scored

Step 19 · 2024-01-02 → 2024-12-31

Arm A

Portfolio book — rebalanced quarterly · 4 constructions · 20 names held · selection: reselect · 1.6% in cash · turnover 8.4× · 1 name dropped from the held union of 60 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 4.84% of 248 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2024-01-01	-18.5185%	4.7379%	31.5753%	19.15%	yes	2.7461%	0 / 60
2024-04-01	-13.7415%	10.8678%	42.7061%	4.4618%	yes	2.835%	1 / 62
2024-07-01	-7.72%	12.9596%	35.8378%	2.1042%	yes	1.9065%	10 / 63
2024-10-01	-7.3444%	9.0491%	26.5184%	-0.7281%	yes	1.6555%	1 / 63

Arm B

Portfolio book — rebalanced quarterly · 4 constructions · 20 names held · selection: reselect · 0.0% in cash · 1 name dropped from the held union of 60 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2024-01-01	-18.5185%	4.7379%	31.5753%	19.15%	yes	2.7461%	0 / 60
2024-04-01	-13.7415%	10.8678%	42.7061%	4.4618%	yes	2.835%	1 / 62
2024-07-01	-7.72%	12.9596%	35.8378%	2.1042%	yes	1.9065%	10 / 63
2024-10-01	-7.3444%	9.0491%	26.5184%	-0.7281%	yes	1.6555%	1 / 63

Step 20 · 2025-01-02 → 2025-12-31

Arm A

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 10.0% in cash · turnover 6.9× · 9 names dropped from the held union of 48 (weights renormalised onto the rest)

Projection accuracy — realized outcome fell inside the P5–P95 cone in 100.0% of 4 scored rebalances (an honest 90% band would contain ~90.0%) · 95% VaR breached on 10.16% of 246 days (expected ~5.0%)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2025-01-01	-5.2879%	13.2623%	32.0411%	-3.9018%	yes	1.8048%	12 / 59
2025-04-01	-4.183%	9.4735%	26.611%	12.2527%	yes	1.545%	7 / 61
2025-07-01	-7.1917%	14.5785%	38.8596%	11.3459%	yes	1.8754%	0 / 63
2025-10-01	-8.0837%	17.7195%	47.523%	4.536%	yes	2.4395%	6 / 63
2026-01-01	no segment follows this rebalance — not scored

Arm B

Portfolio book — rebalanced quarterly · 5 constructions · 18 names held · selection: reselect · 0.0% in cash · 9 names dropped from the held union of 48 (weights renormalised onto the rest)

Rebalance	P5	Median	P95	Realized	In band	VaR 95 (1d)	Breaches
2025-01-01	-5.2879%	13.2623%	32.0411%	-3.9018%	yes	1.8048%	12 / 59
2025-04-01	-4.183%	9.4735%	26.611%	12.2527%	yes	1.545%	7 / 61
2025-07-01	-7.1917%	14.5785%	38.8596%	11.3459%	yes	1.8754%	0 / 63
2025-10-01	-8.0837%	17.7195%	47.523%	4.536%	yes	2.4395%	6 / 63
2026-01-01	no segment follows this rebalance — not scored

QuanterLab · Study f51ba5def904 · compiled August 02, 2026. Point-in-time constituents and hypothesis-registration timestamps are enforced by the platform; transaction costs are not modelled in this study. 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.

A VIX Circuit Breaker on Momentum: One Crisis Averted, Every Recovery Missed (S&P 500, 2006–2026)

1 Methodology

2 Results

2.1 Headline

2.2 Per-step results

2.3 Search accounting

2.4 The comparison

3 The circuit

The objective and the search

3.1 Universe

3.2 Price Loader

3.3 Factor Loader

3.4 Factor Momentum

3.5 Factor Composite

3.6 Factor Top Tier

3.7 Portfolio Backtest

3.8 Vix Cut

5 Discussion

5.1 Findings

5.2 Interpretation

5.3 Limitations

Computed, not featured

References

Appendix A Reproducibility in QuanterLab

Appendix A2 Sealed-hypothesis record

Appendix B Per-step diagnostics

Step 1 · 2006-01-03 → 2006-12-29

Step 2 · 2007-01-03 → 2007-12-31

Step 3 · 2008-01-02 → 2008-12-31

Step 4 · 2009-01-02 → 2009-12-31

Step 5 · 2010-01-04 → 2010-12-31

Step 6 · 2011-01-03 → 2011-12-30

Step 7 · 2012-01-03 → 2012-12-31

Step 8 · 2013-01-02 → 2013-12-31

Step 9 · 2014-01-02 → 2014-12-31

Step 10 · 2015-01-02 → 2015-12-31

Step 11 · 2016-01-04 → 2016-12-30

Step 12 · 2017-01-03 → 2017-12-29

Step 13 · 2018-01-02 → 2018-12-31

Step 14 · 2019-01-02 → 2019-12-31

Step 15 · 2020-01-02 → 2020-12-31

Step 16 · 2021-01-04 → 2021-12-31

Step 17 · 2022-01-03 → 2022-12-30

Step 18 · 2023-01-03 → 2023-12-29

Step 19 · 2024-01-02 → 2024-12-31

Step 20 · 2025-01-02 → 2025-12-31

Run a study like this one