It starts on its own
A first search runs in the background after the first backtest of every new strategy.
The AI Optimizer runs thousands of backtests to search for better settings, then brings back two variants to compare with the original before anything changes.
A first search runs in the background after the first backtest of every new strategy.
Indicator settings, entries, exits, mid-trade rules and order methods.
Up to 5 weighted criteria out of 12, from win rate and drawdown to Sharpe, profit factor and Runbot Score.
The winning settings are replayed on an earlier window the optimizer was never shown.
Start it. From the strategy page or by asking the agent; a short search also starts on its own after the first backtest.
Choose its length and speed. From 2 minutes to 4 hours of search; a faster search finishes the same work sooner.
Thousands of backtests. It runs in the background, and the app says when it is done.
Two variants come back. Best Score and Best APR, each with its return, drawdown, win rate, trades, score and Sharpe.
Compare, then apply. A variant's equity curve is drawn over the original, metric by metric. Applied, it becomes a new version; the previous one stays in the history.
Results are ranked by up to 5 weighted criteria chosen among 12: win rate, drawdown, flat drawdown, Sharpe, profit factor, Runbot Score, monthly performance, risk/reward, number of trades, time in position, drawdown variance and PnL variance.
The optimizer tunes parameters, not fundamentals. It does not change the sizing, the timeframe or the market unless you ask. To change the indicators or the rules, talk to the AI strategy builder; the AI analyst also proposes fixes after every backtest.
Walk-forward validation optimizes on one window, then replays the winning settings on an earlier window the optimizer was never shown, so a variant that only fits its own data shows it. Every search runs on the tick-level backtester.
The settings of the indicators, entry rules, exits, mid-trade rules and order methods, with no limit on the number of parameters. It leaves the sizing, the timeframe and the market alone unless you ask.
Walk-forward validation replays the winning settings on an earlier window the optimizer was never shown, and every variant is compared with the original before you apply it.
Yes. Optimize is one of the tools of the MCP API, and an agent connected to a Runbot account can ask for it too.
Tell us about your community, the markets you want to open and your timeline. We will show you the app under your brand.
Choose how you want to access the Runbot ecosystem.
Thanks. We will reach back to you by email.
You are about to be redirected to an external website that is not affiliated with Runbot. This third-party platform operates independently and is not managed, endorsed, or controlled by Runbot.
Runbot is not responsible for the content, security, or services provided by this external site.