No black-box claims. Both strategies the bot trades are openly documented setups from published market-structure research, refined and backtested across 24 months of real candle data.
Two correlated instruments — Nasdaq and Russell 2000, Gold and Silver, EUR and GBP — almost always move in step. When one of them sweeps a recent high or low and the other refuses to confirm, that asymmetry frequently marks a session reversal.
The pattern itself is well-known. The edge is what's been done with it — years of refinement against real market data, multi-tier filtering, and chunked stability testing on independent out-of-sample windows. The specific filter thresholds, session gates, per-pair direction rules, and risk-multiplier triggers are not published — that's the work we've put in.
When a recent 4-hour high or low gets swept and the move fails immediately, the last opposing candle that held that level often becomes a high-probability reversal zone — price frequently retests it on the way back the other direction.
Our variant runs this setup on the 4-hour reference candle with a fixed reward target. As with the divergence strategy, the framework is well-known. The specific filtering — which instruments pass quality screening, which direction edges hold per pair, how the entry is timed, how risk is sized when it stacks with the paired-asset signals — is the result of years of testing, and isn't published.
This setup runs as a single-instrument strategy on the same instrument universe as the divergence strategy, plus a few additional pairs. Some instruments execute live; others are watched but not auto-executed pending further out-of-sample validation. Subscribers see the live ones as actionable signals; the rest as observation-only data.
Every trade is structurally protected — not by hope, by code:
"R" is the standard risk-adjusted unit: 1R = the dollar amount risked on the trade. A +0.92R-per-signal edge across 864 trades means that on average, every trade returned 92% of what it risked — losers are clipped at -1R, winners run to +3R-9R depending on the pair.
Both strategies were validated through chunked-stability testing: the 24-month dataset was split into three 8-month chunks, and the strategy had to produce positive returns in all three independently. This is to filter out strategies that look profitable overall because of one freak month — and is far more conservative than typical "show the equity curve" backtest reporting.
What it is: a research bot publishing every signal it generates, transparently, with the methodology openly documented. Subscribers can use it as one input in their own informed decision-making.