No black box, no "trust the algorithm." Here's exactly what decides every entry, every exit, and every stop — in plain language.
Every bot in the fleet runs the same underlying idea: prices that move to a statistical extreme — unusually oversold, unusually overbought — tend to snap back toward their recent average. That's mean-reversion, and it's one of the oldest, most-studied ideas in trading.
The catch is that "statistically extreme" only means something when the market is actually moving enough for the signal to be meaningful. A quiet, low-volatility market can sit at "oversold" for hours without doing anything — trading that noise is how mean-reversion strategies bleed small losses. So before any bot considers an RSI or Bollinger Band signal at all, it checks a volatility gate: is this instrument's current volatility unusually high relative to its own recent history? Only if that gate is open does a signal get evaluated in the first place.
Every position — long or short — goes through the same five stages, entirely in code:
Every bot here trades on an OANDA practice account — real market prices, real order execution logic, real slippage and spread, but no real money changing hands. That's deliberate: the fleet's approach is to run each strategy against live markets for long enough to build a genuine track record before any real capital would ever be considered, not to skip that step and hope.
The Performance page shows that track record in full — including the stretches where it's losing, because a strategy is only worth trusting if you can see it fail honestly, not just when it's winning.
Every claim on this page is checkable against real data published automatically as it happens.