Strategy backtesting for retail traders
Send us your strategy and we'll test it properly, with clean data, real costs and a long list of checks built to break it. If it survives, you'll know why. If it doesn't, better to find out here than in your trading account.
What we do
Bring any strategy. We'll try our best to break it.
Most backtests are built to look good. Ours are built to find the problems: look-ahead bias, missing costs, a parameter that only works at one exact value, a year that carried all the profit. We run your strategy through the same nine-phase process every time and write up what we find in plain English.
Request a backtest →Monte Carlo simulation
IllustrativeOne backtest is one path through history. Reshuffle the trades a few thousand times and you see how bad the drawdowns could realistically get.
You get the numbers, the risk limits and a forward-testing plan to take it further.
We'll show you exactly where and why. That's money you didn't lose.
A backtest is a hypothesis, not proof. We treat it that way.
Our process
Every phase is there to catch one specific way strategies fail. Most failures happen in the boring middle steps, so those are the ones we never skip.
01
Before any testing, we write down why this should make money and who's on the losing side of your trade.
Catches strategies with no real reason to work
02
Bad data gives you confident, wrong answers. So we clean it first, and split it before anyone looks.
Catches survivorship bias and dirty data
03
The test only gets information you'd actually have had at that moment, and it pays every rupee of cost.
Catches profits that only exist on paper
04
Total return is the least interesting number. We look at how the money was made, and when.
Catches a few lucky trades posing as an edge
05
Most skipped
The most important phase, and the one most people skip. This is where we really try to break it.
Catches curve-fitting, the big one
06
A good strategy with bad sizing can still empty an account. We set the limits while everyone's calm.
Catches one bad run wiping you out
07
Weeks to months on live data, without real money, to check that reality agrees with the backtest.
Catches gaps between the model and the market
08
Start at 10 to 25% of the size you want. The goal here isn't profit yet. It's checking execution and your own nerves.
Catches execution surprises, early and cheaply
09
Markets change and edges fade. A strategy that worked last year owes you nothing this year.
Catches an edge that quietly stopped working
Phase 05, up close
Tweak a strategy long enough and it'll fit any stretch of history perfectly. Then it falls apart on the next one. Here's how we tell a real edge from a lucky fit.
Parameter sensitivity
A 20-period average should still work at 18 or 22. If only one exact value makes money, it's been curve-fitted.
Walk-forward analysis
We optimise on a rolling window, test on the unseen data right after it, then roll forward and repeat. That's how you'd really re-tune and trade over time.
The data split
The most recent 20% stays locked until the very end, and we open it once. If the strategy fails there, we don't tweak it and try again. That would just make it more in-sample data.
Cleaned out before the first test
Monte Carlo
Reshuffling the trade order shows the realistic range of drawdowns and losing streaks, not just the one path that happened.
Stress tests
Count the attempts
If it took 200 variations to find one that works, it's probably luck. We log every version we test, and we'll take fewer rules over clever ones.
Kill switch
For example: if live drawdown goes past 1.5× the backtest's worst, or results leave the Monte Carlo 95% range, trading stops and we review.
Charts on this page explain concepts. None of them are real performance data.
Your report
You won't get a single big return figure and a nice-looking curve. You'll get the numbers that tell you whether the strategy is actually any good, explained so they make sense.
Fewer than about 100 and it's hard to tell skill from luck.
The average profit per trade after costs. This is the number that pays the bills.
Gross profit over gross loss. Around 1.3 to 1.5 after costs is decent. Above 3, we get suspicious.
How deep it went and how long it took to recover. Could you honestly sit through that?
How much return you got for the risk you took.
Neither means much on its own. A 30% win rate can be great, and so can 80%.
Expect it to be worse live. Better to know now.
Did it beat simply buying and holding the Nifty, after costs and risk? If not, why bother?
Before real money
Twelve boxes. A strategy goes live with us only when every one is ticked.
A backtest is a hypothesis, not proof. Every step after the first one is there to try and break the strategy. Only what survives deserves real money. Evidence over promises.
How we work
Same process, every strategy, every time. No shortcuts because a result looks good.
We show our working, and past results never guarantee future ones.
If we can't explain a result simply, we haven't understood it well enough yet.
Every test has limits, and so does automation. We'll always tell you what they are.
Follow along
We share what we learn from testing strategies: the mistakes, the methods and the results that didn't go our way.
Questions
If we can't answer something honestly, we'll say so.
Pretty much anything that can be written as rules: indicator-based, price action, breakouts, mean reversion, intraday or positional, equities or F&O, or an EA you already have. If part of your strategy is gut feel, we'll work through it with you until it's precise enough to test.
Then the process did its job. We'll show you where it broke, whether that's costs, one lucky year or a fragile parameter, and whether there's anything worth rethinking. Finding out in a report is a lot cheaper than finding out in your account.
It's when a strategy has been tuned so closely to past data that it's learned the noise instead of a real pattern. It looks amazing in the backtest and then stops working. Parameter sensitivity, walk-forward testing and a one-time holdout are how we catch it.
Your backtest is one sequence of trades. Shuffle that order or randomly skip trades a few thousand times and you get a realistic range of drawdowns and losing streaks. That range is what you should plan your risk around, not the single path that happened.
No. Markets change and edges fade, and nobody can promise future returns. A properly tested strategy just gives you much better odds of knowing what you're getting into. That's also why we forward test and go live small.
No. Explain how you trade and we'll handle the rules, the code and the testing. If you plan to automate later, check your broker's and SEBI's current rules for retail algo trading, because API access and registration requirements apply.
Got a strategy?
Tell us how you trade. We'll come back with a plan for testing it and an honest idea of what to expect.
