Neural Trading Performance

Live forecast MAE plus Sharpe, Sortino, drawdown, PnL, and win-rate stats for our trading algorithms

30-Day Return

Sharpe Ratio

Sortino Ratio

Max Drawdown

Win Rate

Total Trades

Forecasting Accuracy

1H MAE

1H Error Stdev

7H MAE

7H Error Stdev

1D MAE

1D Error Stdev

7D MAE

7D Error Stdev

Lookback: -- days • Samples: -- (1H), -- (7H), -- (1D), -- (7D)

MAE is the mean absolute percent error between predicted and realized prices at each horizon. Lower MAE means the forecasts were closer to the realized market price.

Buy/Sell Levels Over Time

Trading Bot Playthrough

Time: -- Equity: -- Return: -- Trades: --

Watch the bot's walk-forward strategy replay over real market data: buy/sell levels update each hour, triangles mark simulated entries and exits (fills at next candle open with fees + 5bps slippage), and the lower panel tracks equity. Drag the slider to scrub.

Historical Forecast Explorer

Hover or scrub over any candle to see the 7-day forecast made at that moment, its confidence band, and what actually happened

Hover a candle to inspect a historical forecast

Every candle stores the 7-day-ahead forecast our model produced at that close. The dashed line is that forecast, the shaded cone is the realized MAE band at each horizon, and the dotted line is what the market actually did next.

Performance by Trading Pair

Pair	Return	Sharpe	Sortino	Win Rate	Trades
Loading performance data...

About Our Metrics

Sharpe Ratio measures risk-adjusted returns. Values above 1.0 indicate good performance, above 2.0 is excellent. It compares excess returns to total volatility.

Sortino Ratio is similar but only penalizes downside volatility. Higher values indicate better downside risk management - important for capital preservation.

Forecast MAE tracks prediction error separately from trading performance, so you can see both forecast precision and whether the strategy converted those forecasts into risk-adjusted returns.

Buy/Sell Levels show the neural network's predicted optimal entry and exit prices over time, generated by our transformer models trained on historical market patterns.

All metrics are calculated from live simulated trades on our neural trading system. Past performance does not guarantee future results.