You describe a strategy in plain English, and Secuora backtests it on real market data. No code, no formula builder. Type what you trade the way you would explain it to a friend, and you get a full backtest back: an equity curve, win rate, profit factor and drawdown. Then the edge detector tells you the part that actually matters, whether that result is a real edge or just luck.
The AI backtester, step by step
- You describe the strategy. For example: buy when RSI drops below 30 while price is above the 200 EMA, take profit at two times risk, stop at the prior swing low.
- The AI compiles it into exact rules. It reads your words and writes precise entry, exit, stop and target conditions. You see those rules before anything runs, so you can check that it understood you.
- A deterministic engine runs it on real data. The same backtest engine that powers the rest of Secuora steps through real historical candles one at a time and records every trade.
- You read the results. Equity curve, win rate, profit factor, average win and loss, drawdown, and the full trade list, all from real price history.
Why you can trust the numbers
The important thing to understand is what the AI does and does not do. The AI only translates your description into rules. It never makes up the result. Once your strategy is compiled, a plain, deterministic engine does the math on real candles, so the same strategy always returns the same numbers. The AI writes the recipe; the engine cooks it, on real ingredients.
That engine has been checked against a separate, independent simulator across thousands of trades with no differences. The one place to stay sharp is your wording: a vague description can compile to rules you did not quite mean. That is why the exact rules are shown to you up front, so you can tighten the wording and run it again.
The edge detector: real edge, or just luck?
A backtest curve that goes up and to the right is not proof of anything. Plenty of strategies look brilliant in hindsight and fall apart the moment you trade them live, because the good-looking result was luck or overfitting. The edge detector exists to catch exactly that.
It takes the returns from your backtest, or from your real logged trades, and runs the same checks professional quant teams use:
- Probabilistic and deflated Sharpe. How confident you can be that the Sharpe ratio is genuinely above zero, adjusted for how many strategies you tried.
- A Monte Carlo p-value. It reshuffles your trades thousands of times to ask how often pure chance would beat your result.
- Minimum track record length. How many trades you would need before the result is statistically convincing.
- Concentration and tail checks. Whether a handful of lucky trades carry the whole profit, and how the strategy behaves on its worst days.
You get one straight verdict: real edge, promising, mixed, or likely noise, with the numbers behind it. If you do not have enough trades yet, it says so rather than handing you a false green light.
The honest part most tools skip
If you test a hundred variations of a strategy and keep the best one, that winner looks good partly by luck. The edge detector knows this. The more combinations you searched to land on a result, the higher the bar it has to clear before it earns a real-edge verdict. Most backtesters quietly ignore this, which is the single biggest reason a great-looking backtest disappoints live. Secuora puts it front and center.
Frequently asked questions
What is the AI backtester?
It lets you describe a trading strategy in plain English, like "buy when RSI drops below 30 and price is above the 200 EMA, take profit at two times risk, stop at the prior swing low". Secuora reads that description, turns it into exact entry, exit and risk rules, and runs it on real historical market data. You get a full backtest with an equity curve, win rate, profit factor and drawdown, without writing any code.
Does the AI make up the results?
No. The AI only translates your words into precise rules. It never invents the numbers. Once your strategy is compiled into rules, a deterministic backtest engine runs it candle by candle on the real price history, so the same strategy always produces the same result. The AI writes the recipe; the engine does the cooking, on real ingredients.
How accurate is it?
The engine has been validated against an independent simulator across thousands of trades with zero differences, so the math is trustworthy. The one place to be careful is the description: vague wording can compile to rules you did not quite mean, so the tool shows you the exact rules it built before running, and you can refine the wording and run it again.
What does it cost?
Every account gets a couple of free runs to try it. After that, runs use credits from low-cost packs, so you only pay for what you use. One run is a single compile and backtest.
What is the edge detector?
A backtest curve going up is not proof of an edge. The edge detector takes the returns from your strategy and runs the same statistical tests quant funds use to separate a real edge from luck or overfitting: probabilistic and deflated Sharpe, a Monte Carlo p-value, the minimum track record length, and checks for whether a few lucky trades carry the whole result. It then gives you a plain verdict.
What does the verdict mean?
The edge detector returns one of: real edge, promising, mixed, or likely noise, along with the numbers behind it. "Likely noise" means the result sits within what random luck would produce, so you should not trust it yet. "Real edge" means the result is strong enough, and consistent enough, that luck is an unlikely explanation. It is honest about sample size: too few trades returns "insufficient data" rather than a false green light.
Why does the number of strategies I tried matter?
If you try a hundred variations and keep the best one, that best one looks good partly by luck. The edge detector accounts for this with a deflation penalty: the more combinations you searched to land on a result, the higher the bar it has to clear. Most tools quietly skip this, which is a big reason so many great-looking backtests fall apart live.