The Factor Models tab is the fundamental-research side of the platform. Instead of reacting to price, you evaluate companies on their financials — profitability, cash flow, earnings quality, valuation, balance-sheet strength — and build portfolios from the names that score well on the factors you care about.
What lives in this tab
Four tiers that roughly correspond to the order you use them:
- Screening — Multi-Factor Screener (FM099MFSX). Rank every ticker in a universe (S&P 500, NASDAQ 100, FTSE, etc.) on the factor weights you choose: Value, Quality, Momentum, Growth, or a custom blend.
- Quality Analyzers — Seven deep-dive modules, one per dimension. Profitability (FM003PSSP), Cash Flow (FM004CSSC), Earnings (FM005ESSE), Capital Allocation (FM006CSSA), Piotroski F-Score (FM007PSSF), Value Multiples (FM008VSSP), Intrinsic Value Deviation (FM009VSSA). Use these to validate each shortlisted name before committing.
- Factor Backtesting + Portfolio Construction — Historical Factor Backtesting (FM101FBKT) tests whether your factor definition would have produced excess returns historically, with survivorship-bias adjustment. Then Factor-Ranked MVO / HRP / Inverse Volatility (FM102) turns the top-ranked names into an actual portfolio with computed weights.
- Risk and Audit — Monte Carlo Simulation (FM095MCSX) and Value at Risk (FM094VARX) for downside assessment. Autopsy Analysis (FM103APSX) for reviewing what went wrong when a backtest disappointed — contribution by factor, by name, by period.
Canonical workflow
- Shortlist with the Screener. Pick your universe, pick or weight the factors, run. Top 10–20 names are your working shortlist.
- Validate each shortlisted name with one or two Quality Analyzers. For a growth thesis, run Profitability and Cash Flow. For a value thesis, run Value Multiples and Intrinsic Value Deviation. For a turnaround, run Piotroski.
- Test the factor historically. Move the factor definition to Historical Factor Backtesting. Rebalance frequency, lookback, top-quantile count — the module rebuilds the rolling top-N portfolio through history and reports returns, drawdowns, and factor decay.
- Construct the portfolio. Feed the top-ranked names into Factor-Ranked MVO (classical mean-variance), Factor-Ranked HRP (hierarchical risk parity — more robust for noisy correlations), or Factor-Ranked IVOL (inverse volatility — simplest, no covariance estimation). Pick one based on how much you trust the correlation structure in your universe.
- Stress-test the weights. Run Monte Carlo on the weighted portfolio to see the distribution of possible one-year outcomes. Run VaR for the tail percentile you care about (usually 5% or 1%).
- If results disappoint, autopsy. Autopsy Analysis breaks down the historical backtest by factor contribution, sector, and time period. It answers "did my factor actually work, or did one sector carry the whole thing?"
Which Quality Analyzer when
- Profitability — Growth names, checking whether returns on capital are actually improving.
- Cash Flow — Any name where earnings look suspiciously strong. Cash is much harder to manipulate than reported earnings.
- Earnings — Quality of earnings: accruals, one-off items, revenue recognition.
- Capital Allocation — Mature companies. Are they returning capital effectively or destroying it on bad acquisitions?
- Piotroski F-Score — Value / turnaround situations. The classic 9-point score.
- Value Multiples — Relative-value screening across a peer set.
- Intrinsic Value Deviation — DCF-derived intrinsic value vs. market price, with explicit assumption guidance so the output is reproducible.
Reading the outputs
- Screener composite score — Factor-weighted rank. Rank is more reliable than absolute score; the top quintile is more meaningful than any cutoff.
- Quality Analyzer scorecards — Each analyzer produces a pillar score with sub-component breakdown. A high pillar score with one visibly weak sub-component is usually worth investigating more than a uniformly mediocre score.
- Historical Factor Backtest — Watch factor decay: does the top-quintile minus bottom-quintile spread persist, or does it collapse in recent years? A collapsed spread is the factor fading.
- Monte Carlo distribution — The percentile bands matter more than the mean. A high expected return with a fat left tail is often a worse risk profile than a lower expected return with a compact distribution.
Common pitfalls
Watch for
- Screener without validation. Screening produces a rank, not a buy list. Quality Analyzers are what turn a ranked name into a position sized on actual financial health.
- Mistaking MVO for a neutral optimizer. Mean-variance is extremely sensitive to the expected-return inputs. On realistic estimates it produces concentrated, unstable weights. HRP is usually the safer default.
- Ignoring factor decay. A factor that worked 2000–2010 may not be working now. Historical Factor Backtesting shows this if you look at the rolling spread, not just the full-period average.
Deeper reading
- Factor Models → Quant Approach Theory / Platform Usage.
- Factor Models → Key Ratios Essential Guide / Interpreting Results.
- Factor Models → Scoring Methodology and Portfolio Construction Theory.