purrtfolio

The strategy graveyard

Most trading strategies don't work. We test them anyway, and publish how they died.

Claims from X, Reddit, vendors and academic papers, re-run on years of real market data with real prop-firm costs. Every rule is written down before we look at the results.

Browse the graveyard Get your strategy tested
200studies run end to end
4,529backtest variants logged, so the luck of trying many is priced in
188rule sets pre-registered before the data was read
10autopsies published so far, more every week

Recent autopsies

See all studies →
Gold · intradayNo gross edge

"CCI mean reversion on gold, London session: $10,000 → $46,519"

Before costs it earns nothing: within 0.08 points of zero in all 15 parameter settings. After costs, −0.42 points a trade at 19 trades a day.

Source: a popular X day-trader
Nasdaq · opening rangeJust geometry

"64.5% of NQ sessions break both sides of the 8–9am range, so fade to the other end"

True, and meaningless. With the bars shuffled into random order, both sides still get broken 69.2% of the time, against 69.4% in real data.

Source: a trading-analytics vendor
Bitcoin · session timingDied live

Buy bitcoin from 21:00 to 23:00 UTC, while stock exchanges are shut, every day

Our strongest backtest ever (deflated Sharpe 0.980). Traded forward for real: 40 trades averaging −$7.21, against +$61.62 expected. The backtest still reproduces to the cent, so it was not a data error. A great backtest is not proof.

Source: QuantPedia (2022)

Free tools

Run the same checks on your own idea. Everything runs in your browser. Nothing you type is sent anywhere.

Prop challenge simulator

Your win rate, average win and loss, and trades per day. Out comes your chance of passing, or blowing, a prop-firm challenge.

COMING SOON

True cost calculator

Measured spread, commission and overnight swap for each instrument. See how much of a 5-point scalp the costs take.

COMING SOON

Edge reality check

Enter your backtest stats. We show what coin flips would score with the same rules, and how many years it takes to tell skill from luck.

COMING SOON

Market data, updated daily

Each page shows the data, and also what happened when we tested trading on it.

Futures positioning (COT)

Who is long and short in Nasdaq, S&P, Dow, gold, euro, pound and bitcoin futures, from the CFTC's weekly report.

Volatility & regime monitor

The VIX term structure, plus our regime reading: calm or stressed, trending or choppy.

Event calendar

FOMC, CPI, payrolls, ECB and BoE dates, with how far markets have historically moved on each.

How we test

  1. 01

    Take the claim literally

    We use the author's own rules and numbers. Where they left a gap, we write down every reasonable reading.

  2. 02

    Lock the rules first

    Every test is fixed in writing before we read a single result, and the most recent data is held back unseen.

  3. 03

    Beat luck, then beat costs

    The strategy must beat random entries at the same times of day, survive real spreads and swap, and hold up when its settings are nudged.

  4. 04

    Publish the verdict

    Pass or fail, the numbers go up, together with what it would take to change our mind.

Get your strategy tested

Send us your rules. You get the same battery every study here goes through, delivered as a private report.

  1. Send your rules, market and timeframe.
  2. We reply with a quote. Nothing is paid until you accept it.
  3. Pay through a secure link and get your private report.

From US$149 · one rule set, one market

Requests open soon.

Members

Full write-ups, downloadable statistics and early access to new studies. Free summaries stay free.

The list opens soon.