Free backtesting is genuinely useful for learning how testing works and whether an idea is obviously broken. Where it stops being sufficient is specific and predictable, and knowing the boundary tells you when a free result is enough to act on.
What free tools do well
- Teach the mechanics. How a strategy is specified, what a report contains, what the metrics mean.
- Reject obviously bad ideas. A strategy that loses badly across any reasonable sample does not need a paid tool to establish that.
- Build intuition. Seeing how changing a stop loss or an entry time moves the result is valuable regardless of the tool.
- Compare rough alternatives. Whether a straddle or a strangle suits a period, at a coarse level.
None of that requires precision. For learning, a tool that is directionally right is adequate.
The five limits that usually appear
| Limit | Why it matters | When it stops mattering |
|---|---|---|
| Short historical range | Cannot contain multiple market regimes | Never — this one is fundamental |
| End-of-day data | Intraday stop losses cannot be evaluated at all | If your strategy has no intraday rules |
| Two-leg cap | Excludes condors and butterflies | If you only trade simple structures |
| Gross results, no costs | Flatters every strategy, multi-leg worst | If you subtract costs yourself |
| Few runs allowed | The plateau test needs several runs of one strategy | If you only ever run once, which is itself a problem |
The second and fourth rows are the ones that most often make a free result unusable rather than merely imprecise.
Why end-of-day data is disqualifying for intraday rules
If your strategy has a per-leg stop loss, a target, or a trailing stop, those rules are evaluated during the session. End-of-day data has no during.
A tool working from daily closes can tell you what a position entered in the morning would have been worth at the close. It cannot tell you whether the stop loss fired at 11:40, which for most intraday options strategies is the thing that determines the result.
Why gross results mislead so badly on options
Brokerage, exchange charges, STT, stamp duty and GST all apply per leg, on entry and again on exit.
| Structure | Legs | Cost events per round trip |
|---|---|---|
| Single option | 1 | 2 |
| Straddle or strangle | 2 | 4 |
| Vertical spread | 2 | 4 |
| Iron condor or butterfly | 4 | 8 |
A gross result understates a four-leg strategy's cost by eight charge events plus eight crossed spreads. A thin edge can be entirely consumed by that, which means a gross backtest can show a profitable strategy that is reliably unprofitable in practice.
This is one of the main reasons live results trail backtests, and it is the easiest of those reasons to correct for yourself.
Making a free result usable
If a free tool is what you have, these adjustments recover much of the value.
1
Establish your per-leg cost once
Brokerage, exchange charges and taxes, plus an estimate of the spread you cross. Compute it properly once and reuse it.
2
Subtract it manually
Multiply by legs and by two for entry and exit, then by the trade count. Deduct from the gross result.
3
Check the trade count before anything else
Below roughly a hundred, treat every ratio as provisional regardless of the tool.
4
Run the longest range the tool allows
And note explicitly which market conditions the sample does and does not contain.
5
Treat the result as a filter, not a decision
Good enough to reject bad ideas; not good enough to fund one.
When free is genuinely enough
There are real cases, and it is worth naming them rather than implying everyone needs a paid tool.
- You are learning. Precision is not the point yet.
- Your strategy has no intraday rules. Entry and exit at fixed times with no stops means daily data loses less.
- You are rejecting rather than selecting. A clearly bad result is informative on any tool.
- You are not going to trade it. Testing an idea out of curiosity needs no rigour.
What Arthalab includes
For comparison: backtesting is included in the plan with 10 credits a day, resetting at midnight IST. A normal backtest uses one credit; a bucket backtest uses one per active strategy inside it. Failed runs are refunded and unused credits do not carry over.
Backtests use the lot size applicable to each historical date, so a report spanning a revision stays internally consistent. Coverage is NIFTY and SENSEX index options on NSE and BSE.
The limitation worth knowing
The honest constraint is the daily cap. Ten runs supports genuine iteration including the plateau test, but a workflow sweeping dozens of combinations in an afternoon will hit it.
Questions to ask any tool, free or paid
- What data granularity does it use?
- Are costs applied, and per leg or per trade?
- How many legs does it support?
- How does it handle lot-size revisions within the test period?
- How easy is it to re-run at different parameters?
- Does the report show drawdown duration and trade count, not just profit?
The full evaluation guide covers what each answer tells you.
The short version
- Free tools teach mechanics and reject bad ideas well
- End-of-day data cannot evaluate intraday stops — it answers a different question
- Gross results understate a four-leg strategy by eight charge events per round trip
- Subtract your own cost estimate to recover most of the value
- Use a free result as a filter, then test properly before funding anything
Frequently asked questions
For learning the mechanics and rejecting obviously bad ideas, yes. It stops being sufficient when your strategy has intraday rules, more than two legs, or a thin edge that costs would consume.
Because a stop loss is evaluated during the session and daily data has no during. The backtest effectively tests what would have happened if the rule had not existed.
Substantially on multi-leg structures. A four-leg strategy incurs charges on eight legs per round trip, plus eight crossed spreads. A thin edge can disappear entirely.
Yes. Work out your per-leg cost once, multiply by legs and by two for entry and exit, then by the trade count, and subtract. It recovers most of the value.
10 credits a day, resetting at midnight IST. Bucket backtests use one credit per active strategy inside the bucket, and failed runs are refunded.
When you are rejecting an idea rather than selecting one, or when your strategy has no intraday rules. A good result on a limited tool is a reason to test further, not a reason to fund.
Yes. If the tool applies today's lot size to a historical period that used a different one, the rupee figures are wrong — and those are the figures you use to judge whether a drawdown is tolerable.
Start with a free 3-day trial
Build a strategy, backtest it and run it on paper — no broker, no IP and no money needed to try it.
