The Bake-Off Scoreboard: 37 Signals In, Nobody's Winning Yet

We run three paper-traded strategies against each other — Ignition (momentum), Phoenix (mean-reversion), and Boost (attention arbitrage) — and publish every signal with its 1-hour and 24-hour returns. Here's where the scoreboard stands, with no cherry-picking:

The strategy leaderboard (24h returns)

Read that table the honest way: Boost "leads" only in the sense that it loses the least. Four of its 42 signals cleared +50% within 24 hours, versus three for Ignition and none for Phoenix — and Phoenix has just two resolved signals, far too few to judge anything. No strategy has a positive average. None is anywhere near proven.

What the data is actually telling us

Three patterns are emerging, and all three argue for caution, not confidence:

1. The first hour decides almost everything. Median 1h return −6.8%; by 24 hours the average is −45.9%. Whatever edge exists in these signals evaporates fast — which is why the bots run 1-hour fast exits instead of hoping.

2. Winners are rare but real. Sixteen percent of signals cleared +20% within an hour. That's not zero — it's the entire reason the experiment continues. The question is whether disciplined exits can harvest enough of them to beat the other 84%.

3. Everything is still paper for a reason. A track record with three negative averages is not a sales pitch. It's a lab notebook. The bake-off exists so that if one strategy ever does separate from the pack, the proof will be public from day one — not manufactured after the fact.

The scoreboard updates every scan. If the numbers ever turn green, you'll see it here first — along with everything that came before it.