Evidence · factor attribution
Is it selection, or factor exposure?
Two different questions: which market and style exposures contributed to the return, and whether the remaining alpha is distinguishable from zero after fees. FlightDeck publishes both, followed below by the separate comparison with factor funds.
- Academic window starts
- 27 Jun 2025
- Academic window ends
- 31 Aug 2026
- Trading days
- 296
- Academic basis
- SPY (cap-weight)
Academic factor attribution
What remains after the market and style factors?
FlightDeck separates the cumulative return into market, size, value, momentum and a selection residual using the Fama–French / Carhart factors. These contributions describe the record; the regression below asks whether alpha is statistically distinguishable from zero.
Published comparison window: 27 Jun 2025 to 31 Aug 2026 · 296 trading days · basis: SPY (cap-weight). This window lags the live record because academic factor data arrive later.
| Component | Contribution, gross | 95% bootstrap interval |
|---|---|---|
| Market | +22.2 ppt | 15.86 to 28.23 ppt |
| Size | −0.3 ppt | −1.49 to 0.42 ppt |
| Value | +8.1 ppt | 4.40 to 11.98 ppt |
| Momentum | 0.0 ppt | −1.34 to 0.54 ppt |
| Selection residual | +26.4 ppt | 20.46 to 32.88 ppt |
Contributions are percentage points of cumulative return, not annual alpha. The residual includes unexplained effects; its bootstrap interval measures estimation precision, not the conventional alpha significance test.
Conventional alpha test, gross
| Model | Alpha a year, gross | t, gross | p, grosstwo-sided | Gross at 95%two-sided |
|---|---|---|---|---|
| CAPM | +23.95% | 2.34 | 0.0193 | Significant |
| FF3 | +18.53% | 1.97 | 0.0485 | Significant |
| Carhart-4 | +18.05% | 1.92 | 0.0551 | Not significant |
| FF5 | +18.38% | 2.09 | 0.0370 | Significant |
| FF5 + Mom | +18.71% | 2.10 | 0.0354 | Significant |
Daily excess-return regressions use Newey-West HAC standard errors; p-values here are two-sided as FlightDeck publishes them. The five models are related diagnostics of one record, not independent trials, and supplementary to the primary test on the Statistical Validation page. All model results are shown, including those that do not meet the 95% threshold. A short sample cannot establish a durable edge. Computed on gross returns; does not reflect the deduction of all fees and expenses. Net total returns for the same period are shown on the Evidence page. Source and methodology on FlightDeck.
The same test, five factor funds
Alpha Convergence against funds that own the factors
Each row is the average daily alpha on Alpha Opportunity days, with its t-statistic and p-value. A factor fund that showed the same alpha would suggest the excess is factor exposure rather than selection.
| Portfolio | Alpha on AO days | t-statistic | p-value | At 95% |
|---|---|---|---|---|
| Alpha Convergence274 AO days, live | 8.1 bp/day | 2.19 | 0.028 | Significant |
| DFUSXDimensional US large company | 4.0 bp/day | 1.37 | 0.170 | Not significant |
| MTUMMomentum | −0.6 bp/day | −0.09 | 0.925 | Not significant |
| QUALQuality | −1.7 bp/day | −1.06 | 0.291 | Not significant |
| VTVValue | 0.8 bp/day | 0.28 | 0.783 | Not significant |
| VIGDividend growth | −2.3 bp/day | −1.07 | 0.284 | Not significant |
From FlightDeck’s AC Market Monitor feed (factor validation), never recomputed here. DFUSX is Dimensional’s US large company fund; MTUM, QUAL, VTV and VIG are momentum, quality, value and dividend-growth funds. A result that is not significant does not prove a factor played no part; it means the factor funds did not show a reliable edge on the same days.
After beta
Jensen alpha: the excess left after the market’s share is taken out
A simple excess return assumes a beta of one. Alpha Convergence runs at lower beta than the market by design, so the regression version, with standard errors robust to serial correlation, asks what remains once the actual beta is accounted for.
Backstage tour
See it running live
A live screen-share of FlightDeck with Mark Tennenbaum, CIO: the command center driven in real time, how today’s market is classified and what the book does about it, and the record with the tests behind it.