Signals with
a paper trail.

Most signal services are vibes in a Discord server.

You get an alert. You don't know if it backtests. You don't know the historical hit rate. You don't know if the person posting it has any idea what they're doing. By the time you find out, you're down 8 points and the channel has moved on.

Fonflo Lab takes the opposite approach. Every strategy starts as a hypothesis and ends as a published result. If it grades out, it goes live in Discord with a real-time alert. If it degrades over time, you see that too — the kill is just as public as the promotion.

Live alerts. Posted methodology. No mystery.

• Hypothesis. A pattern is identified — opening range failure, volume exhaustion, late-session reversal — and written as detection logic.
• Walk-forward backtest. Trained on the first ~4 months of a year, tested out-of-sample on the last ~2. No lookahead. Stops checked intrabar. Slippage and commission baked in.
• Random-control test. 200+ randomized entries on the same days with identical risk geometry. If random beats the signal, the signal isn't real.
• Regime & day-type breakdown. A strategy that only works in one VIX regime or on balance days isn't a robust edge — it's a conditional bet, and we treat it like one.
• Grade A or B only. Promotion threshold: profit factor > 1.3, OOS PF > 1.0, OOS win rate > 45%, beats random. Anything weaker stays in the lab.
• Probation, then live. First 10 trades after promotion run on a tighter kill threshold. Pass and it's tracked indefinitely. Fail and it's killed publicly.

Quant-minded traders.

If you read backtest reports, argue about walk-forward methodology, and have ever screamed at a "guru" who couldn't show you a hit rate — this server is built for you. We post the work, we post the kills, and we expect you'll push back when something looks fitted.

If you want black-box magic, this isn't that. We'd rather kill a strategy in front of you than pretend it's still working.