P&L Tracker: Reviewing Performance Without Fooling Yourself

P&L tells you the outcome and almost nothing about the cause. Used as a starting point for diagnosis it reliably produces wrong conclusions, and used as a review tool at the right interval it is genuinely useful. The difference is in how and when you read it.

How the figures are captured

Daily P&L is captured after the market close rather than computed on demand. That keeps historical figures stable rather than shifting every time prices move.

A number that changes when you look at it twice is not useful for reviewing past performance, which is why the snapshot approach is worth the small delay.

Why the separation matters

Combining them would be convenient and misleading in a specific direction.

Paper results are structurally better than live ones — fills are assumed, margin is never a constraint, and nothing is ever rejected. Mixing the two produces a figure flattered by the half that cannot lose for operational reasons.

The measurement it enables

The gap between paper and live is a number worth measuring deliberately, which is only possible if the two are kept apart.

What P&L cannot tell you

A losing day has many possible causes, and the P&L screen distinguishes none of them.

What happened	How P&L looks	Where to actually look
The strategy traded and lost	Down	Expected behaviour — check against the backtest
Orders were rejected	Flat or odd	Execution logs and the order book
The bot was never started	Flat	Logs — an empty log is the signal
A multi-leg entry partly filled	Unexpected	Positions, immediately
The strategy correctly did nothing	Flat	Logs — evaluations with no trigger

Three of those five produce a flat or near-flat day and have completely different causes. Reasoning from the P&L alone cannot separate them, and guessing usually leads to changing a strategy that was never at fault.

The right interval for review

Checking P&L daily encourages reacting to noise. Checking it rarely means missing a genuine change in behaviour. A workable split is to separate the two questions by interval.

Interval	Question	What to look at
Daily	Did the system work?	Logs — did bots run, were there rejections
Weekly	Is behaviour changing?	Rejection counts, days run, drift against the backtest
Monthly	Is the strategy still sound?	Drawdown against what the backtest showed

The daily question is operational and the monthly one is about the strategy. Conflating them means treating a bad week as evidence about the strategy when it is usually evidence about nothing.

Comparing against the backtest

The most useful review is against what the backtest said, rather than against zero.

Is the current drawdown within what the backtest already showed? If yes, the strategy is behaving normally and there is nothing to decide.
Is it deeper than anything in the test period? That is a real signal, either about the strategy or about conditions the test did not contain.
Is the trade count roughly as expected? Far fewer trades usually means an operational problem rather than a strategy one.
Has the gap to paper widened? If you run a paper copy, a widening gap points at execution rather than at the strategy.

Point one is what the drawdown figure in your backtest report is for. Without it, every losing stretch feels unprecedented.

Deciding whether to stop

The hardest use of a P&L screen is deciding when a strategy has stopped working, and the honest answer is that the screen cannot tell you.

What can tell you is a stop condition written in advance, in rupees, before you deployed. Reaching it means the strategy did what it was always capable of doing, not that something went wrong.

What to track beyond the total

Days the strategy actually ran, against days it should have
Number of rejections and their reasons
Whether any multi-leg entry filled partially
Drawdown against the backtest's figure, in rupees
The gap between live and paper, if you run both

The first item catches the most common and least visible problem. A strategy that ran on sixteen of twenty intended days is underperforming for reasons that have nothing to do with the strategy.

The short version

Daily figures are captured after the close so history stays stable
Live and paper totals are never combined, which keeps the comparison honest
P&L shows outcome, not cause — three different problems all look flat
Debug from the logs outward, never from P&L backwards
Review operations daily, behaviour weekly, the strategy monthly
Judge a drawdown against the backtest, and against a stop condition written in advance

Frequently asked questions

So that historical figures stay stable instead of shifting every time prices move. A number that changes when you look at it twice is not useful for reviewing past performance.

No, never. They are stored separately, because paper results are structurally better and combining them would flatter the total.

It could be three different things: the bot never started, the strategy correctly found no entry, or orders were rejected. The P&L screen cannot distinguish them — the logs can.

Check the logs daily for operational problems, review behaviour weekly, and assess the strategy monthly. Checking P&L daily mostly encourages reacting to noise.

Compare it against the max drawdown your backtest showed. If the current stretch is within that, the strategy is behaving as tested.

At the stop condition you wrote in rupees before deploying. Deciding during a drawdown reliably produces a different and worse answer.

How many days the strategy actually ran against how many it should have, rejection counts and reasons, any partial fills, and the gap to paper if you run both.

Some gap is structural — slippage, costs and margin constraints. A gap that is large, or the opposite sign, suggests the backtest was measuring something other than what you are running.

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.

Create a free account

All articles More in Live Trading