ShowMeTheTrades
Long-form

Every full write-up.

The lab notes below each take one question and answer it properly — charts, the rows behind them, and the cases that argue against the conclusion. If you arrived from the daily log, this is where the long versions live.

The daily log →  ·  every test, one short entry per testing day
007
Eleven free EAs cleared every filter. Three are now trading on a demo.
The whole shortlist, why the account is $10,000, and the failure modes that would void the test.
29 Oct 2026
006
Forty-one accounts ended under $1,000. Thirty-five were still trading.
Where the $1.9 million actually went, file by file, and why only 6 of the 41 were margin calls.
29 Oct 2026
005
372 free EAs, one harness: the leaderboard is a drawdown contest
The funnel nobody publishes, return against drawdown, and the 11 files a real account could survive.
29 Oct 2026
004
A 74% win-rate EA blew up in six days. Three rescues, all failed.
Hedging, cutting the lot, and volatility timing, tested against the same collapse.
21 Oct 2026
003
Sample audit: the EA that passed every test and still worries me
Structure, tail test, source review, exposure arithmetic and four variant rebuilds on one file.
19 Oct 2026
002
A 5-pip target needs 66.7% before the spread is mentioned
Break-even win rate, the spread as geometry rather than cost, and a calculator you can put a vendor's numbers into.
16 Oct 2026
001
The 77-file batch: 34 of them never opened a trade
The first full sample, published unfiltered, including every file that produced no trade record at all.
14 Oct 2026

Every number in every note is recomputable from the files on this site: batch-b.csv for the 372 free downloads, data.json for the 77-file batch.

The next batch, as it finishes. One note when a batch is done: what was tested, what failed, and the number that decided it. No product offers, no “top 10” roundups, no affiliate mail.