Copy-tradingKilled

A Bitcoin-direction entry filter for the copy-trading bot

abandoned

host strategy killed before the gate reached its sample floor

In plain English

We built a filter to improve the copy-trading bot, but the bot was shut down before the filter ever saw enough trades to be judged.

What was claimed

Only copy trades when Bitcoin's own direction agrees, and the copy-bot's results improve.

The bar we locked

30+ trades on each side of the filter

What the data said

far too few trades, and the strategy was shut down first

Why it's dead

The entry filter needed 30+ trades in each bucket to evaluate. The strategy it filtered (Leo) was killed first, so the sample stopped accruing, and the pre-reg (the test contract locked in writing before the data) forbids re-scoring a partial window against new cutoffs.

The detail

Closed INCONCLUSIVE_ABANDON (the pre-registered verdict for a test that ends before enough data arrives) per its own Section 7: at the Leo kill (n=44 live) the pass/fail buckets were almost certainly below the locked N≥30 each. Host strategy dead → no further trades; the anti-p-hacking commitment (no re-judging a finished window against friendlier thresholds) forbids re-evaluating the window against different cutoffs.

Kill date: 2026-05-09
Sample: n=44 (host killed)
Method: Pre-registered live test
Verdict: abandoned, sample stopped