ArdenAlgoHow we'll know if the engine has edge
The dirtiest secret of trading-tool marketing is that with enough freedom after the fact, anything can be made to look like edge: pick the lucky window, drop the bad symbols, tweak a threshold, quote gross. The defense isn't better intentions — it's removing the freedom. That's what pre-registration does.
Frozen before data
Our evaluation design — sample triggers, pass criteria, the benchmark, even the random seed — was written and frozen while the ledger was still empty. Mechanization v1 itself is frozen too; if we ever change a parameter, that's a new version with a dated note, never silently pooled with old results.
The trigger
The pooled test runs when the ledger reaches 60 closed trades, or 42 days with at least 30 — whichever comes first. Until then, nothing on the site is presented as evidence of anything; the pages say "descriptive" because that's all small samples can be.
The benchmark: beating randomness, after costs
The engine's pooled net result is compared against thousands of duration-matched random entries: simulated trades at random times, on the same symbols, held for comparable horizons, paying the same modeled costs. If random entries do about as well, the signals carry no information worth paying for — full stop. We also cross-check with a second framing of the same test and will flag any disagreement between the two rather than quoting the friendlier one.
What pass and fail actually mean
- Pass doesn't mean "guaranteed profits" — it means the live sample cleared a null benchmark after costs, once. It buys the engine four more weeks of forward confirmation, not your trust.
- Fail means we say, in public, that this mechanization did not beat randomness in its live test — and the product remains a scanner with an honest ledger, priced and described as exactly that.
A sibling research project of ours ran ~170 pre-registered experiments across 9 signal families and found zero real edges — every one of them died in exactly this kind of test. (That work is internal and unpublished, so treat the number as our claim about ourselves, not independent evidence — what it explains is why our bar sits where it sits, and why we're comfortable publishing whatever lands.) The four gates →