Algorithmic strategy Simulation
Algorithmic trading strategy
A rule-based strategy that fades stretched price across every Hyperliquid market, but only when the market as a whole is moving — and always on the opposite side of that move. Every signal was weighted by measured predictive power rather than convention, and the result held up on data it was never tuned against.
The strategy is running forward on daily data but is not yet executing on-chain, so the book below is simulated. Nothing here is investment advice.
Live on-chain
Live trading
0 long · 0 short open
No open positions right now — the strategy is selective and opens only strong setups.
How the strategy works
Every signal was measured before it was used: an edge scan over ~72,000 bars ranked market structure, levels and sweeps, volume, momentum, volatility and candle triggers by how well each actually predicted the next move. Several textbook readings measured inverted on daily crypto — momentum, breakouts, structure-break "change of character", swing-failure patterns and pin bars all fade rather than follow — so the strategy carries them with a negative weight. The result is mean reversion: it buys weakness and sells strength, inside the long-term trend. On top sit two timing filters. The market regime decides the side: when the whole universe has moved at least 4% over five days, the strategy takes the opposite side, and when the market is flat it does not trade at all, because that regime measured a losing profit factor. The stall filter decides the moment: it refuses to fade a move that is still accelerating, and waits for the flush or the rip to lose steam. Exits are deliberately plain, because measurement said the exit shape is not where this edge lives: every sane variant lands within 0.06 of the others out of sample. What survived is a −12% stop and a break-even lock that moves the stop to +1% once a trade is 8% ahead, so a position that already worked cannot end red — worth about ten points of win rate. There is no take-profit and no trailing stop; both measured as subtracting money or as noise once stop fills were made honest (a stop is filled at the open when the market gaps past it, never at the level itself). Finally, the strategy watches its own equity: after a losing month it halves its risk instead of guessing when the market will suit it again.
Past performance does not guarantee future results. This is a rule-based simulation on daily data, published for research and transparency — not investment advice or a solicitation.