The platform
Platform · AI Architect
Describe the system. The AI designs it — and proves it on real data.
Tell the AI Architect your goal in plain language. It measures the live market on 25 indicators, designs a full system, then backtests it on real candles and returns an honest verdict — paper, small-live, or a refusal. Every number is measured, never promised.
Every number is measured on real history, never promised. The Architect self-corrects a weak design and refuses a setup it cannot prove — it recommends, you decide.
Plain-language intake
State a goal, capital and risk appetite. No config forms to fight — the AI turns words into a complete, working design.
Measured, not promised
It reads the live market regime, then backtests the design on real candles — return, drawdown, win rate and alpha are measured by the engine, never invented.
Self-correcting & honest
If the backtest is weak it redesigns automatically; the verdict is honest — paper, small-live, or a refusal with the reason. It says no to what it cannot prove.
Beyond a single system
Position sizing from the measured edge (fractional Kelly), a best-pair scan, safer/aggressive what-ifs, and a whole portfolio built from a capital amount with measured correlations.
Where it sits in the lifecycle
01
Goal → measured market
You describe the system; the AI measures the live regime on 25 indicators — trend, volatility, macro — before designing anything.
02
Design → real backtest → self-correct
It designs the config, backtests it on real history, and redesigns until the result is sound or it honestly can’t prove the idea.
03
Honest verdict → deploy
A paper/live/refuse verdict with a graduation protocol; deploy the exact proven config into the builder in one step.
Design your first system
Describe a strategy in plain words — the AI designs and backtests it in seconds.