TL;DR: A backtest measures the system. A journal measures you running the system. They answer different questions, and the difference between them is where most trading money quietly goes. A backtest assumes you took every signal, at the modelled price, with no charges and no hesitation. Your journal records what actually happened: the trades you skipped after two losses, the entries you chased by forty paise, the STT and brokerage nobody models. Run both, subtract one from the other, and the gap has a name — execution drift — and it is fixable in a way that a bad system is not.


Last updated: 10 September 2026 · 9 min read

By Pulkit Mangal — F&O trader since 2017, founder of TradeDiary. Traded across Zerodha, Kotak and Dhan; built TradeDiary after losing ₹14L in 2021 to mistakes a journal would have caught — including six months spent optimising a backtest while my live account did something completely different.


My backtest said 62% win rate and ₹1,840 average profit per trade. Six months of live trading said 41% and ₹390.

Same system. Same instrument. Same rules, written down and taped to the monitor. I spent three weeks assuming the backtest was broken — re-checking the data, the fill logic, the position sizing. The backtest was fine. I was the variable I had never measured.

That is the whole story of backtest versus journal. One tells you whether a set of rules made money on history. The other tells you whether you made money following them. Traders spend months on the first question and almost no time on the second, which is strange, because only the second one has your name on it.

💡 Short on time? The fastest way to find your execution gap is to let a journal do the subtraction. Start a free trading journal, import your tradebook, and it will show your real win rate, average R and expectancy per setup — after charges — so you can compare them against what your backtest promised.

What "backtest vs journal" actually means

A backtest is a simulation of a rule set against historical prices. A journal is a record of decisions you actually made with real money. They overlap far less than people assume.

A backtest answers: if these rules had been followed perfectly, what would have happened?

A journal answers: what did I do, why did I do it, and what did it cost?

The word doing the heavy lifting in the first question is perfectly. A backtest has no fear. It never skips a signal because the last two lost. It never sizes up after a good week. It never widens a stop at 3:10pm because the trade is about to work. It takes every entry at the modelled price, pays whatever costs you told it to model, and never once considers not trading.

Three India-specific gaps make that assumption worse here than in most markets:

Why the gap matters more for Indian traders

Because the margin for error is thinner here, and the regulator has published exactly how thin.

SEBI's January 2024 study of individual F&O traders found that 93% of individual traders lost money in equity F&O between FY22 and FY24, with average losses of about ₹2 lakh across three years. Buried in that data is a detail that matters for this argument: transaction costs alone consumed a meaningful share of turnover for active traders. A strategy with a genuine but small edge can be pushed negative by costs a backtest never counted.

Tax reporting needs the journal, not the backtest. Trading income in India is business income. F&O is non-speculative, intraday equity is speculative, and they are reported separately, with the audit thresholds under Section 44AB depending on turnover and declared profit. Your backtest is irrelevant to your ITR. Your actual trade record is the only thing that produces the numbers, and reconstructing a year of it in March is how people end up filing wrong.

Broker statements are not a journal. A contract note tells you what filled. It does not tell you which setup it was, what your stop was, whether you followed the plan, or how you felt. Without those fields, you can see that you lost money but never which decision lost it.

Algo traders are not exempt — they are more exposed. If you deploy an automated strategy, you still choose when to switch it on, when to pull it after a drawdown, and when to "just override this once". Those interventions never appear in the backtest and are frequently the entire difference between the equity curve you modelled and the one in your account.

The seven fields that close the gap

To compare a backtest against reality you need the journal to capture what the backtest assumed. These are the minimum:

Field Why it matters
Setup / strategy tag Without it you cannot compare like with like. A blended win rate across three systems tells you nothing about any of them.
Planned entry vs actual entry The difference is slippage plus hesitation. Backtests model the first and ignore the second.
Planned stop Lets you compute R. Without a stop recorded at entry, "risk" is a story you tell afterwards.
Actual exit and reason Target, stop, time, or discretion. The fourth category is the one that quietly costs the most.
Quantity in lots Exposes the rounding your backtest never had to do.
All-in charges Brokerage, STT, exchange, GST, stamp duty. The gap between gross and net is where small edges die.
Signals skipped The trades you did not take. This is the field almost nobody keeps, and it is the single biggest source of drift.

That last row deserves emphasis. A backtest takes 100% of signals by definition. If you took 80 of 100 and the 20 you skipped were disproportionately the uncomfortable ones — the gap-downs, the entries right after a loss — then your live results are not a worse version of your backtest. They are a different strategy, one your backtest never tested.

Optional but useful: market regime, time of day, whether you were in drawdown at entry, and a one-line note on your state of mind. See trade tagging software for how to structure tags so they stay comparable over time.

How to actually run the comparison

  1. Fix the backtest's cost model first. Re-run it with real all-in charges for your broker and segment, not just brokerage. If the edge disappears at this step, stop — you do not have an execution problem, you have a strategy problem, and no amount of journaling fixes it.
  2. Export your live trades from the broker for the same period. Zerodha, Kotak, Dhan, Upstox and Angel all provide a tradebook; see the Zerodha trade journal guide for the export path.
  3. Tag every live trade with the setup it belongs to. Untagged trades cannot be compared to anything.
  4. Compute the same four numbers on both sides: win rate, average win, average loss, and expectancy per trade. Use R-multiples rather than rupees so position size differences do not distort the comparison.
  5. Subtract. Backtest expectancy minus live expectancy is your execution drift, in R per trade.
  6. Attribute the drift. Split it into slippage (planned vs actual entry), skipped signals (count them), early exits (exit reason = discretion), and unmodelled costs. One of these is almost always dominant.
  7. Fix the dominant one, then re-measure after 30 trades. Not before — anything less is noise.

A real example, with numbers

Rohit is 34, a systems engineer in Pune, trading BANKNIFTY weekly options around a ₹6.5 lakh account. He backtested a simple opening-range breakout over two years: 58% win rate, +0.42R expectancy per trade. Live, over 118 trades in five months, he made ₹31,400 — well below the roughly ₹1.6 lakh the backtest implied.

He tagged his trades and ran the subtraction:

Source of drift Trades affected Cost in R Share of gap
Signals skipped after a losing day 23 not taken −0.19R/trade 46%
Entry slippage vs modelled price 118 −0.11R/trade 27%
Discretionary early exits 31 −0.08R/trade 19%
Charges beyond the model 118 −0.03R/trade 8%

The finding that changed his year: the 23 trades he skipped were, in aggregate, his most profitable cohort. They came disproportionately after losing days, which is precisely when he did not feel like trading. His system was fine. His attendance was not.

Fixing attendance is a different project from fixing a strategy — it is closer to the work described in revenge trading psychology — but he could only see it because the journal recorded what the backtest assumed away.

Six mistakes that make the comparison useless

  1. Comparing different periods. A backtest over two years against six live months compares a system to a market phase. Match the windows, or match the regime.
  2. Comparing in rupees instead of R. Position size changed over those six months; rupees hide that, R does not.
  3. Leaving trades untagged. A blended number across setups cannot be compared to a backtest of one setup.
  4. Modelling brokerage only. Add STT, exchange charges, GST, SEBI fees and stamp duty, or your "edge" is partly imaginary.
  5. Not recording skipped signals. The most valuable field, and the easiest to omit because there is no contract note to remind you.
  6. Re-optimising the backtest to match live results. This is the worst one. It curve-fits the simulation to your bad execution and destroys the only clean benchmark you had.

How TradeDiary helps

TradeDiary imports your tradebook from Zerodha, Kotak, Dhan, Upstox, Angel One and others, applies real all-in charges rather than estimates, and computes win rate, average R and expectancy per setup — so the live half of the comparison takes minutes instead of a weekend in a spreadsheet. Tag your setups once and every future import lands in the right bucket, which is what makes the backtest-versus-live comparison repeatable rather than a one-off exercise you never redo. Start free — import your trades and see your real expectancy per setup.

You may also like: What a trading edge actually is · R-multiples explained · Trade journal vs spreadsheet · Max drawdown, and what it really costs · Algo trading slippage · Walk forward analysis

Frequently asked questions

Is backtesting useless if I keep a journal? No — they answer different questions and you need both. A backtest tells you whether a rule set has any edge at all, which is worth knowing before you risk money on it. A journal tells you whether you are capturing that edge. Skipping the backtest means trading a system nobody ever validated; skipping the journal means never learning why the validated system is not working for you.

How many live trades before the comparison means anything? Around 30 gives you a first signal, 100 gives you something you can act on with reasonable confidence. Below 30, the difference between your backtest and your account is mostly variance, and you risk "fixing" a problem that was never there.

Should I forward test instead of going live? Forward testing — paper trading the system in real time — sits usefully between the two. It removes the historical-data assumptions while keeping real money out of it. Its limitation is that it also removes the emotion, which is exactly the variable you are trying to measure, so paper results tend to sit closer to the backtest than your live account will.

My backtest edge disappears once I add full charges. What now? Then you have your answer, and it is a useful one. Either the strategy needs a higher average win to clear costs (longer holds, wider targets), fewer trades of higher quality, or a cheaper cost structure. What you must not do is trade it anyway and hope execution makes up the difference — costs are the one part of the gap that never improves with practice.

Do algo traders need a journal at all? Yes, and arguably more than discretionary traders. The code follows the rules; you decide when it runs. Switch-offs during drawdowns, manual overrides and parameter changes mid-deployment are all decisions that never appear in the backtest, and they are usually where a live algo diverges from its simulated equity curve.


Risk disclaimer

This article is educational and is not investment advice. Trading in equity, derivatives and commodities carries substantial risk of loss. SEBI's January 2024 study found that 93% of individual traders in the equity F&O segment made losses between FY22 and FY24. Past performance, whether in a backtest or a live record, does not guarantee future results. Consult a SEBI-registered investment adviser and a qualified chartered accountant for advice specific to your situation.

Written by Pulkit Mangal — F&O trader since 2017 and founder of TradeDiary.

Last updated: 10 Sep 2026.