Gross profitability long-only v1 preregistration

Ledger sources: trial #36 hash 4dc542091c8dd4e1 · trial #41 hash 65fa381edc9bbc83 · preregister event rule hash fa519c91f793a718fdb92004879116922cc60da87b2c302e2386722717b05e66 · Screening window 1998-01-01→2021-12-31 · Lockbox 2022-01-01→2026-08-21 (opened once)

LOCKBOX STATUS AT REGISTRATION (2026-08-28T02:56Z): UNOPENED. The 2022-01-01→2026-08-21 lockbox has not been inspected, computed, printed, or used in this document. Every performance or capacity number below comes only from the screening-window gross-profitability confirm-card for trial #41; trial #36 is recorded only as the positive lead.

1. Mechanism (who pays, and why they keep paying)

The proposed payer is delayed incorporation of operating profitability into prices: investors and constrained institutions may underweight the durability of gross profits relative to the asset base, especially when accounting data arrive with reporting lags and compete with more salient narratives. The effect should appear as higher subsequent returns among the most profitable eligible firms and should weaken when profitability is rapidly repriced, later restated, or overwhelmed by distress and financing shocks. The negative control is the bottom-quintile portfolio: it must not outperform the registered top-quintile portfolio, because that would reverse the mechanism's predicted ordering.

2. Exact rule (frozen)

• Universe: US common stocks with unadjusted price ≥ $10 and trailing 20-session ADV (close × volume) ≥ $10M at formation, with 20 observations; no category filter.
• Signal and availability: latest SF1 ARQ GP/A = (revenue − COR) / assets, using deterministic permaticker/datekey collapse. The first exchange session on or after datekey is session 0; the record becomes available at session 2 close (datekey+2 sessions, pessimistic).
• PIT-safe structural exclusion: latest available ARQ revenue, COR and nonzero assets must all be non-null; missing COR is not imputed.
• Portfolio and execution: long-only equal-weight top GP/A quintile; buy-and-hold annual cohorts staggered monthly, with 12 live cohorts equal-weighted as 1/12 tranches. Decide at the formation month-end close, enter at the next session adjusted open, and exit at the tranche-anniversary adjusted open. A delisted name exits at its last observed close; monthly aggregate trades are netted.
• Costs: charge 10 bps per side for current-metadata Mid+ names and 25 bps per side otherwise. Cohort-sleeve wealth drift, forced exits, and final liquidation are included. The size bucket is current rather than point-in-time and is used only for costs, never eligibility.
• Sizing overlay: apply the pre-registered prior-60-completed-unscaled-net-daily-return framework with a one-session lag, 10% annual volatility target, and 1.0 leverage cap at the PORTFOLIO level to the GP/A sleeve. Alongside momentum, each strategy sleeve is sized separately under live-risk-policy/v1; the overlay cannot pre-empt an overnight gap.

3. Predicted shape (from the screening window — the confirm-card numbers)

• Primary fundable result: GP/A active net excess versus the separately costed equal-weight PIT-safe structural universe was +2.83%/yr, with Sharpe 0.58, max drawdown −15.4%, and Newey–West t = 2.46.
• Costs: active net excess was +2.80%/yr at 2× costs and +2.76%/yr at 3× costs.
• Halves: active net Sharpe was 0.55 in 1998–2009 and 0.62 in 2010–2021.
• Right-tail dependence: removing the best 14 months left +0.82%/yr with Newey–West t = 0.91. The screening edge therefore depends materially on the right tail; this is a central fragility, not a robustness success.
• Capacity: the median binding live sleeve at 1% of ADV was $27.3m.

4. Falsifier (pre-committed lockbox gates — any miss = REFUSED, no "fix and retry")

• Lockbox net excess versus the equal-weight PIT-safe structural universe must be > 0 and at least 50% of the screening-window +2.83%/yr estimate.
• Newey–West t must be ≥ 2 on the lockbox alone.
• At least 3 of the 5 lockbox calendar years must have positive net return.
• At 2× the frozen per-side costs, lockbox net excess must remain positive.
• Deflated Sharpe Ratio against the full 41-trial ledger must be ≥ 0.95.
• Negative control: the bottom quintile must not outperform the registered top-quintile portfolio in the lockbox.
• These gates are conjunctive. Any miss yields REFUSED; the lockbox may be opened once and may not be repaired, re-specified, or reopened for a retry.

5. Live contract (if it passes)

• Stage: a passing candidate enters the same paper-promotion and live-graduation ladder as momentum. This registration itself grants no broker or live-trading authority.
• Sleeve: GP/A shares the combined strategy-sleeve cap of £10k with momentum.
• Sizing: start GP/A at 10 bp of NAV. Any later ladder step requires the separate sealed approval demanded by live-risk-policy/v1.
• Kill framework: before any live order, bind live-risk-policy/v1 by its policy-document SHA-256 01c2102758b44067c92715ce67613f41d6ab2ef471f86e8b097c815872b5cc0d. Its hard-flat, soft-breach, retirement, and reconciliation/data-failure framework applies to this sleeve; each strategy sleeve remains separately sized.

6. Trial accounting

• Source free-look trial #36: gross-profitability, event preregister_free_look, registered at 2026-08-27T22:10Z, one allowed variant, ledger-entry hash 4dc542091c8dd4e1.
• Confirm-card trial #41: gross-profitability-confirm, event confirm_card, registered at 2026-08-28T00:50Z, ledger-entry hash 65fa381edc9bbc83; verdict CONDITIONAL_PRE_REGISTER; pre-registration choice GP_A_ALONE.
• Trial-count basis: the full ledger contains 41 trials and has SHA-256 db97b21bdf843622a1766c212e7619a08e3ef944022e8980b858cc7ad1902ae7. The lockbox Deflated Sharpe Ratio uses that full ledger without a post-opening trial-count change.
• This pre-registration's ledger rule hash is fa519c91f793a718fdb92004879116922cc60da87b2c302e2386722717b05e66. Every lockbox calculation, risk-policy binding, live decision, scaling approval, and later run must log against gross-profitability-long-only-v1.

The companion preregistration event binds this document without pretending a self-referential file can hash its own printed digest. Hash the final Markdown bytes after replacing only the next line with the same label and no value; the newline remains.

• Preregistration document SHA-256 (self-reference line blanked): 2d9916fd9c9037c555bfc1c461161694b43c35d536a21e680433b60c16738058
• Verification: perl -0pe 's/^- Preregistration document SHA-256 \(self-reference line blanked\):.*$/- Preregistration document SHA-256 (self-reference line blanked):/m' docs/research/prereg-gross-profitability-long-only-v1-2026-08-28.md | shasum -a 256