I ran 20 pre-registered backtests of retail trading strategies. The pass/fail criteria were written down before each test, and the goalposts stayed welded in place after. Seventeen died, the last of them the one family nothing had tested, killed on 33 years of the exchange's own data. Two passed and still didn't get my money, because a tiebreaker written in advance said they hadn't earned it. Every strategy died in the pipeline before it touched a live account: total tuition paid to the market, $0.
This kit is that pipeline. It will not make you money. It stops your backtest from lying to you about making money, which, if you trade, is the same thing wearing work clothes.
20 gates run · 17 kills (16 straight, then the last untested family) · 1 pass · 2 certified, then benched by their own pre-registered tiebreaker · $0 lost live
counted from the public gate files; every number on this page is in the repo
Each of these looked plausible. Several looked great, right up until honest fills, split-adjusted data, and a de-survivorship universe were applied. The picture first, then the ledger; full writeups are free in the repo.
DISTANCE FROM THE GATE — best honest profit factor per family
fast-strategy gate, pre-registered: PF ≥ 1.30 at the stated cost · every dot is a documented result
Not one family reached the line. The catalyst pair is the expensive lesson: a passing primary sample (hollow dot) collapsing to negative expectancy on the holdout it never touched. Exact figures, verdicts, and causes of death: table below.
| strategy family | best honest result | verdict |
|---|---|---|
| Momentum scalping (213-trade window) | 49% win, ±1.18%: a fair coin | FAIL |
| Opening-range breakout, long+short ("Sharpe 2.4" paper) | PF 0.82–0.85 @ 0.10%/side | FAIL |
| Catalyst gaps (≥5% + volume) | PF 1.23 primary, 0.93 holdout | FAIL |
| Earnings drift (real SEC 8-K dates) | PF 1.11, worse than random gaps (1.29) | FAIL |
| Mean reversion (RSI-2) | PF 0.77 @ 0.25%/side | FAIL |
| News sentiment (19.5k articles, LLM-scored) | made every config worse | FAIL |
| Crypto funding carry | real mechanism, ~0.4% net in 2026 | FAIL |
| Cross-asset trend, monthly (Faber, zero tuned params) | 8.1% CAGR, −1.1% in 2022 | PASS |
| Same rule, sampled daily | 7.31%; whipsaw ate it | FAIL |
| Options premium selling (5 pro income ETFs + CBOE's own indices) | every fund's Sharpe below its own underlying | FAIL |
PF = profit factor. The pattern: nearly every price-derived signal is ~PF 1.05–1.10 gross, and retail friction eats it whole.
GATE 20 · THE LAST UNTESTED FAMILY: OPTIONS PREMIUM SELLING
tested twice over: 5 professional covered-call/put-write ETFs, then CBOE's own benchmark indices (real traded SPX option prices back to 1990)
Every income fund had a worse Sharpe than the index it writes options on. QQQ made 21.6%/yr over QYLD's decade; QYLD kept 10% of it with 70% of the drawdown. But the 33-year index record is the real exhibit: a mechanism dying in public.
| index (vs SPY, Sharpe) | pre-2012 | post-2012 |
|---|---|---|
| BXM covered-call | +0.20 vs +0.16 · beats | +0.61 vs +0.90 · lags |
| PUT put-write | +0.24 vs +0.06 · beats | +0.70 vs +0.90 · lags |
| BXMD 30-delta buy-write | +0.50 vs +0.33 · beats | +0.73 vs +0.90 · lags |
The volatility premium WAS real; the academic papers were written on the pre-2012 sample. Then it got institutionalized, and the edge went to whoever collects the flow. Same life-cycle as crypto funding carry (2021: ~22% → 2026: ~0.4%), two decades slower. The crash tell: in 2020 these indices fell HARDER than SPY (BXM −54.6% annualized vs −32.4%). Short vol is short the gap: insurance that pays out in drizzle and fails in the flood. Full writeup and both prediction audits in the repo.
A backtest is a witness with every incentive to lie to you. Interrogate it like one.
These aren't hypotheticals. Each one is an artifact my own pipeline produced, believed for a while, and then caught. Each one became a rule in the free checklist.
The split artifact. Raw price bars plus a 1:25 reverse split inside the hold window turned a $0.17 stock into the trade of the decade. It never happened. A strategy "passed validation" for six days on trades like this before the split-adjusted rerun killed it.
rule: split-adjusted bars only, no exceptions
The fantasy fill. A backtest that fills your stop at the stop price is fiction: names that gap through a −5% stop routinely close down 10 to 15%. Half my "edges" were fill fantasies that evaporated under a checked-at-close, filled-at-close model.
rule: model the fill you actually get, not the one you want
The false-positive machine. A slot-capped portfolio simulation printed profit factor 3.96 by accidentally cherry-picking under 10% of its own trades. It did this twice, on two unrelated strategies, before the pattern was caught and banned from the pipeline.
rule: slot-capped stats are a false-positive generator
Every verdict above traces to a gate file written before its test ran. The receipts, directly:
| artifact | what it is |
|---|---|
| The full graveyard writeup | all 20 gates, every number, the three expensive lessons |
| Gates 17–20: the passes and the last family | the first PASS, the benched survivors, the split decision that falsified my own prediction, and the 33-year mechanism autopsy |
| The artifact-hunting checklist | the free tier: every rule priced by the disaster that created it (landing in the repo this week) |
I pre-registered the launch the way I pre-register backtests: the paid tiers ship only if enough people demonstrably want them. There's no fake scarcity here and no countdown timer; the incentive runs the other way. The kit is currently working scripts on my machine, packaging them properly takes two honest weekends, and I'll spend those weekends only on real demand.
The demand gate: if the tools would be useful to you, star the repo or open an issue saying what you'd use, or say so in the comments below (which files the issue for you). That's the vote. When it trips, the NOT FOR SALE stamps above become direct checkout links on this page, and the checklist tier stays free either way.
Comments here are powered by GitHub: each one is filed as an issue on the graveyard repo, which means posting a comment literally advances the demand gate. Say what you'd use, what's missing, or why the whole thing is wrong. Arguments count as interest.
One retail trader who spent months building a bot, watched honest testing kill every strategy it ran, and kept the machine that did the killing. The graveyard writeups converted the skeptics first, including the ones who showed up to argue. The receipts are public, the code produced them, and when my own pre-registered prediction got falsified by my own gate last week, that went in the repo too.
Questions, arguments, or a strategy you want dead? Open an issue on the repo. Arguing in public is the quality-assurance process.
© 2026 · strategygraveyard.com · The Falsification Kit · Educational software and research writeups. Not financial advice. No performance claims made or implied. Backtested results are historical simulations and do not predict future returns. The kit tests strategies; it does not provide, recommend, or execute them.
Widget not loading (some previews block it)? Post directly as a GitHub issue →