You test the ideas you already believe in. That's the problem.
This generates and evolves thousands of complete strategies you'd never have thought to try, then puts every survivor through a robustness gauntlet designed to break the ones that only worked by luck.
Download SabrTrader free See pricingIncluded in Ultimate and Lifetime, with AlgoStudio Pro.
Generation by generation, live. Green is in-sample, magenta is out-of-sample, yellow is incubation on data the strategy has never touched.
Finding an edge is a numbers game, and you're playing it by hand
Hand-building strategies is slow. Worse, it's biased. You test the ideas you already like, using the indicators you already trust, on the timeframes you already trade.
Which means the enormous space of combinations that might actually work goes unexplored, forever, because you'd never think to put those three things together in that order.
The strategy that works is probably one you have no intuition for. That's exactly why you haven't found it.
So you set the rules of the game, pick the indicators it's allowed to use, and let a genetic algorithm search. The best performers survive and combine, mutations introduce ideas nobody had, and each generation is fitter than the last. Every candidate gets backtested on real tick data. You watch it happen, generation by generation.
Not one hopeful backtest. The top of thousands, ranked.
Sharpe and MAR for risk-adjusted return. Max drawdown in dollars, not percent. Win rate, trade count, commission, and an overall Monte Carlo rating from Poor to Outstanding. Sort, filter and pick columns until you've isolated exactly the profile you're willing to trade.
The fitness score that drives the ranking blends the metrics that actually matter, so what rises to the top isn't just whatever made the most money in the fitting window.
Anyone can find a strategy that looks perfect on history
Telling a real edge apart from a curve-fit fluke is the only hard problem here. So every candidate runs a gauntlet.
In-sample and out-of-sample
Built on one slice of data, validated on a slice it never saw. A ratio filter rejects anything whose edge evaporated the moment it left the fitting window.
Incubation
A final held-back slice of the most recent history acts as a forward test, so you see how it did on the freshest data of all.
Monte Carlo
Thousands of permutations of the trade sequence stress the equity curve, producing distributions for net profit, drawdown and consecutive losers. You see the range of outcomes, not one lucky path.
Parameter stress
Re-run with perturbed parameters to confirm the edge isn't balanced on a knife-edge of one specific setting that happened to fit.
Take away its five best trades. Is it still a strategy?
That question kills a startling number of good-looking backtests, and it's the one the remove-top-winners test asks directly.
The robustness dashboard rolls everything into a single score across eight metrics: profit consistency, drawdown, slippage survival, win rate stability, profit factor and more.
A strategy that clears out-of-sample validation, survives Monte Carlo, holds up under perturbed parameters, performs in incubation, and doesn't collapse when you remove its luckiest trades has genuinely earned your attention.
Most of them don't. That's the tool working.
You set the constraints. It does the looking.
The algorithm
Population size, number of generations, what to optimise for, and whether to search freely or bias toward a trading style.
The building blocks
Choose which indicators it's allowed to use, from standard studies to the premium order flow and TradeDevils tools.
Data splits
In-sample percentage, out-of-sample, and the size of the incubation slice held back for forward testing.
Filters
Minimum trades in and out of sample, minimum indicators, minimum trades per day, and the out-of-sample to in-sample ratio floor.
Sessions and risk
Trading-hours windows, stop and target rules, and prop-firm style daily loss and profit limits.
Monte Carlo depth
Permutation count for the robustness distributions. More permutations, tighter confidence, longer run.
Search for strategies that pass the evaluation
If you're trading a funded or evaluation account, clearing the firm's rules is the entire job. A profitable strategy that breaches trailing drawdown on day four is a failed evaluation, not a good strategy.
So every candidate can be checked against prop firm requirements automatically: trailing drawdown, profit target, daily loss limits and consistency rules. You get told which strategies would pass, which would be rejected, and why.
Instead of finding a strategy and then discovering it can't survive your firm's rules, you search inside the rules from the start. It's a small change in workflow that removes an entire category of wasted evaluation fees.
These are runnable strategies, not suggestions
A ranked list is only useful if you can trade what's on it. Three ways out.
Open it in the designer
Load any strategy into AlgoStudio Pro and inspect every rule it discovered. You can read exactly what it built, and change it.
Save it out
Keep it, version it, come back to it. The search results aren't trapped inside the search window.
Run it live
Straight to your broker through the built-in engine, with session control and risk limits. No re-coding, no translation step.
The two halves of the lab. The Auto Strategy Builder searches and hands you survivors. AlgoStudio Pro lets you build and refine by hand. Same engine underneath, so you can move freely between building by machine and building by yourself.
Put strategy discovery on autopilot tonight
Pick your indicators, set the constraints, and let it run while you sleep. Read the leaderboard in the morning.
Download SabrTrader free Compare plansIncluded in Ultimate and Lifetime.
The questions we get
Does it actually generate the strategies, or just optimise mine?
It generates them. You choose the indicators and the constraints, and a genetic algorithm builds and evolves thousands of complete strategies, entry and exit logic included, backtests each on real tick data, and ranks the best on a live leaderboard.
How do I know these aren't curve-fit?
You don't take our word for it, you read the columns. Every candidate is validated out-of-sample on data it was never built on, run through Monte Carlo permutations and parameter stress, forward-tested in incubation, and scored across eight robustness metrics including a remove-top-winners test. The ones that only worked by luck are filtered out before they reach you, and you can see the numbers that made the call.
How long does a search take?
Depends on population size, generations, how much history you point it at and how many Monte Carlo permutations you ask for. All four are yours to set. Most people start a run in the evening and read the leaderboard the next morning.
Can it find strategies that pass prop firm rules?
Yes. It evaluates candidates against trailing drawdown, profit target, daily loss limits and consistency rules, and tells you which would pass and which would be rejected, so you can search specifically for evaluation-ready strategies.
What do I actually get at the end?
A ranked list of real, runnable strategies. Load any of them into AlgoStudio Pro to read and refine the rules, save it, or run it live at your broker. Nothing needs re-coding.
Is this a money printer?
No, and be suspicious of anything described that way. It's a search engine with an unusually strict filter on the results. It will find you candidates worth trading and it will reject far more than it keeps. That rejection rate is the feature.
Which plan includes it?
Ultimate and the Lifetime licence, along with AlgoStudio Pro and the full premium suite. See pricing.