Archive
The living record.
Every paper Nullberg covers ends up here with one of four verdicts: replicated, degraded, failed, or inconclusive. Verdicts are not one-shot. When a replicated factor decays out of sample or a failed factor re-enters the conversation, the entry is updated and the change is logged.
-
April 10, 2026 · failed
Accruals, out of sample, 2016 to 2026
Sloan's 1996 accruals anomaly claims that low-accruals firms (earnings backed by cash) outperform high-accruals firms by ~10.4% per year. On 123 months of US equity data with cash-flow-statement accruals, the primary VW NYSE-breakpoint spec produces +0.883% per month — essentially exactly the canonical 0.87% monthly magnitude — but the i.i.d. t is +1.85, just below the rubric threshold, and the Newey-West 12-lag t is +1.74. Verdict: FAILED by pre-registered rubric. Headline caveat: two of the three EW sensitivities REPLICATE robustly under both i.i.d. and Newey-West (Sens A: +3.81 NW t, Sens B: +2.39 NW t), implying the accruals effect has migrated to smaller stocks where the VW primary underweights it. This is a new kind of near-miss in the archive: FAILED at the canonical magnitude, with sensitivities that clearly replicate.
-
April 10, 2026 · failed
Gross profitability, out of sample, 2016 to 2026
The original claim is that gross profits-to-assets predicts the cross-section of returns about as strongly as book-to-market does. In 123 months of value-weighted NYSE-breakpoint US equity data, the long-short spread goes in the right direction at 0.503% per month (above Novy-Marx's own canonical 0.31%) but with t = +0.93, below the conventional significance bar. Equal-weighted variants flip sign on smaller stocks. This is the closest of Nullberg's first three verdicts to REPLICATED, but the rubric requires t > 2 so it lands as FAILED with an underpowered-survivor interpretation. Backfill update 2026-04-11: Newey-West 12-lag t is +1.05 (slightly above the i.i.d. t of +0.93, implying mildly negative monthly autocorrelation), and both sub-sample halves are positive and consistent (+0.42% and +0.58%), strengthening the underpowered-but-consistent interpretation.
-
April 10, 2026 · failed
The MAX factor, out of sample, 2016 to 2026
The original claim is that the lowest-MAX decile beats the highest-MAX decile by over 1% per month. On 122 months of US equity data, value-weighted with NYSE breakpoints and stocks-only filtering, the spread is not merely absent but inverted: high-MAX beats low-MAX by 1.80% per month with t = -2.57. The finding is robust across three equal-weighted sensitivities with t-statistics from -1.81 to -4.12. Backfill update 2026-04-11 confirms the inversion survives the Newey-West 12-lag HAC adjustment at t = -2.18, and is concentrated in the 2016-2021 first half (t = -2.29) versus the 2021-2026 second half (t = -1.38).
-
April 10, 2026 · failed
Momentum, out of sample, 2017 to 2026
The original abstract says past winners beat past losers by a significant margin over three- to twelve-month holding periods. On 111 months of US equity data, value-weighted with NYSE breakpoints and stocks-only filtering, the winners-minus-losers spread is -0.13% per month with t = -0.16. The effect has decayed to zero in the standard spec and only approaches significance in one shorter-lookback equal-weighted sensitivity at t = +1.45. All four specifications fail to reach significance. Backfill update 2026-04-11: Newey-West 12-lag t is -0.26 on the primary, and both sub-sample halves are near-null (+0.02% and -0.25%), confirming the decay is consistent across time rather than regime-driven.
-
April 10, 2026 · replicated
Value (book-to-market), out of sample, 2016 to 2026
First REPLICATED verdict in the Nullberg archive, and also the most heavily caveated. The Fama-French 1993 HML 2x3 construction on 122 months of US equity data delivers +1.698% per month with i.i.d. t = +2.82, clearing the pre-registered rubric. But the effect is regime-driven: the first half of the sample is statistically null (+0.585%, t = +0.73) while the second half is huge (+2.810%, t = +3.20), with 2022 alone averaging +9.24% on an annualized-monthly basis. The Newey-West 12-lag t is +1.94, just below the rubric threshold. The simple D10-D1 sensitivity replicates under both i.i.d. and Newey-West, but the equal-weighted sensitivity is flat at t = +0.35, consistent with value being a value-weighted large-cap phenomenon in this regime.