Tell Strategy AI your symbol, risk level, and goal. It evaluates 13 strategies against the current market conditions, tunes parameters for each candidate, and runs multi-window backtests so you see how every recommendation would have performed. You review the ranked list and deploy what you trust.
Five stages, fully transparent. Every recommendation comes with the reasoning and the numbers behind it.
Pick a symbol or asset class. Set risk level and trading intent. That is the only input the recommender needs.
The current market is classified across multiple timeframes — trend direction, trend strength, volatility, and over-extension. The classification drives which strategies fit.
Every candidate gets a fit score against the classified market plus your risk and intent. Strategies that do not suit current conditions are flagged or filtered.
Parameters are tuned per strategy and per market — not one-size-fits-all defaults. Each candidate runs through multi-window historical backtests to validate consistency.
Recommendations are persisted with the score, the reasoning, and the backtest summary. You review the ranked list and deploy the ones you want.
A focused universe of proven, broadly applicable strategies — spot and perpetual. Specialty strategies (arbitrage, sentiment, risk management) run directly rather than via the recommender.
A trend-following strategy is the right answer in one market and the wrong answer in another. Strategy AI classifies the market before it recommends — so a recommendation in a ranging market won't be a momentum strategy waiting to fail.
The current market is summarized along directional bias, trend strength, volatility, and whether price is over-extended. That summary is attached to every recommendation so the why is visible.
The classification does not read a single chart — it averages signals across short, medium, and long timeframes so a noisy short-term swing does not overwhelm the actual regime.
Each candidate runs over a meaningful historical period and is evaluated across multiple sub-windows. A strategy has to hold up consistently — not just in the one window that happened to look good.
A candidate has to perform in more than one historical window. A single fluky window does not earn the pass tag.
Each evaluated window needs a real number of trades. A single lucky win does not get to carry the metrics.
A window with excessive drawdown fails — even if return was positive. Passing means the strategy earned its return without heavy losses along the way.
Backtest results are an annotation on each recommendation, not a hard filter. Candidates that didn't clear the consistency check still appear in the ranked list — clearly marked — so you can decide whether to deploy them anyway. A historical fail is information, not a verdict: a strategy can underperform historically and still be the right call live, and vice versa.
Strategy AI is not static. It periodically reviews the real performance of deployed recommendations and adjusts how it scores future picks — both for everyone and for specific market conditions.
When a strategy consistently performs well in a given market condition, future recommendations of that combination are scored higher.
Strategy and condition combinations that consistently underperform are temporarily suppressed from new recommendations until they recover.
Parameter risk levels are adjusted for combinations that are running too hot or too conservatively versus actual outcomes.
Pick perpetual futures as the asset class and Auto as the venue, and Strategy AI runs the full pipeline once per venue — Hyperliquid plus every centralized futures account you've connected. Each venue produces its own recommendation.
One request can return, for example, a trend-following recommendation on Hyperliquid at 5x and a mean-reversion recommendation on your Bybit futures account at 10x — both scored, both backtested, each ready to deploy independently. Symbol naming gets translated automatically per venue, so you don't have to think about USDC vs. USDT, bare symbols vs. settlement-suffixed pairs, or any of the other venue-specific quirks.
Strategy AI is precise about the job it's doing. Reading a recommendation the way it's intended keeps expectations aligned with what the system can actually tell you.
Strategy AI classifies conditions and picks methods that fit. It is not a price forecaster — the recommendation is about which strategy suits the regime, not where Bitcoin is headed tomorrow.
A passing tag means the strategy held up across multiple sub-windows in the past. Markets change, so a passing recommendation can still lose money live — and the learning loop adapts from real outcomes.
Thirteen broadly applicable spot and perp strategies where the recommender can do useful work. Arbitrage, sentiment, risk management, custom, and TradingView strategies run directly rather than through the recommender.
Free to run recommendations against your connected venues. No credit card.