From Research Lab To Your Trading Account

You create API keys on your exchange (Binance, Bybit, OKX) with trade-only permissions — no withdrawal access. We execute trades on your behalf, but your funds never leave your exchange account. You can revoke API access anytime.

30% performance fee on profits only — we earn when you earn. For accounts over $50,000, there's an additional 2% annual management fee. No setup fees, no hidden charges.

Binance, Bybit, OKX, Gate.io, Hyperliquid. More coming soon.

You set your maximum acceptable drawdown (risk level). Our AI analyzes your parameters and selects the optimal trading strategy that fits within your risk tolerance. You can adjust your risk settings anytime.

New users receive virtual bonus credits — up to $45 total — to cover platform performance fees. Earn $10 on registration and verification, $15 for connecting an API key, and $20 for launching your first strategy. Each bonus is valid for 90 days. Bonus credits cannot be withdrawn — they can only be used to pay platform performance fees.

All performance metrics shown in strategy cards (total return %, Sharpe ratio, max drawdown %) come from the same backtest pipeline. Window: the most recent 12 months of historical data, single window. Timeframe: 1-minute OHLCV bars. Parameter selection: hyperparameters are tuned with Optuna on this same 12-month window. Metrics are therefore in-sample — the strategy is evaluated on the data it was fit on. There is no walk-forward validation and no held-out test split at this time. Commissions: fixed 0.1% per side (0.2% round-trip), applied as a taker fee. Real exchange fees may be lower (maker rebates, VIP tiers) or higher depending on the venue and account. Slippage: not modeled. Backtest fills assume the bar's reference price. Funding, borrow costs, and listing/delisting effects: not modeled.

Usually not, and sometimes by a wide margin. The gap between the figures on the strategy cards and what you see in live trading comes from several sources. Market regime change: the backtest window is the last 12 months. When volatility, correlation, or liquidity shift, a strategy fit to the recent past can underperform — sometimes immediately after deployment. In-sample optimization bias: parameters are tuned on the same window used to report results, so the metrics are biased upward. Without walk-forward or out-of-sample testing, we cannot quantify how much of the reported edge is signal versus fit to noise. Slippage: live fills happen on a real order book, not at the bar's reference price. On thin or volatile pairs (e.g. HYPE, PEPE) the realized fill can deviate by tens of basis points or more per trade, which compounds quickly for high-frequency strategies. Execution timing: a 1-minute bar hides intra-bar price movement; the bot's actual entries and exits may land at different prices than the backtest assumed. Fees: backtests use 0.1% per side as a default. Real fees depend on your exchange, fee tier, maker/taker mix, and any rebates — they can be lower for high-volume accounts or higher on smaller venues.