Backtest P&L Reconciliation: Audit Matched Trades

Backtest P&L reconciliation: Match trade quantities and prices before assigning unexplained P&L to an execution cause.

The NQ results discussion compared live trades with a backtester using similar timing. Accessible replies questioned last-price fills and systematically adverse differences. The image figures were not independently validated; the original exercise below demonstrates how to investigate such a discrepancy.

The source discussion is a starting question, not verified strategy evidence. Accessible comments were reviewed incompletely; some replies remained unavailable. Author results are unverified. The exercise below is original educational reasoning with hypothetical inputs, not a reproduction of those results.

Backtest P&L reconciliation: matched-trade residual audit: align before averaging

Choose an identical date interval and strategy configuration. Match signals and orders using stable identifiers where available. Record unmatched signals, rejected requests and partial fills in separate rows. An inner join containing only completed pairs can hide the largest operational difference: one system traded while the other did not.

For matched trades, record side, quantity, decision timestamp, intended execution, observed execution and fees. Keep entry and exit differences separate. Consistent signal times are useful evidence about timing, but they do not prove consistent prices or quantities. Inspect the first divergent event before explaining the final profit-factor gap.

Worked example: correct times, adverse prices

Imagine the simulator buys ten shares at 100 and sells them at 102. Its gross profit is 20. The comparison ledger buys at 100.10 and sells at 101.90, producing 18 before fees. Define the P&L residual as observed minus simulated: it is negative 2. These invented prices expose an unfavorable difference at both ends despite matching trade times.

Repeat with five filled shares instead of ten. Separate the quantity effect from the price effect; do not describe the entire profit difference as slippage. Then add a fee and verify that the net residual changes by exactly that fee in this simple example. Each calculation should be independently reproducible.

Look for direction and concentration

Group residuals by entry versus exit, long versus short, order size, time bucket and exit reason. Preserve the sign convention in the report. Averages can look acceptable while a small set of emergency exits carries most of the discrepancy. Show individual extreme rows rather than relying only on a histogram.

Avoid declaring a universal acceptable tick tolerance. A tick's economic significance depends on quantity, instrument and the strategy's modeled advantage. Express residuals in both price units and cash. Compare their size with the net result being explained, while acknowledging the limited comparison interval.

Check accounting with deliberately simple cases

Use one trade with no fees, one with fees and one partial fill. Reconcile starting cash, purchases, sales and terminal positions. A cost setting should visibly affect the equity ledger, not just appear in configuration output. Add a duplicate event and confirm it does not double a position.

The audit can identify a pricing mismatch, missing order or state inconsistency. It does not prove the remaining strategy has an edge. Once the earliest divergence is understood, preserve the correction and evaluate later observations without retroactively selecting only the trades that now match.

Reference and first useful step

Last-traded prices can differ from market-order execution prices. Investor.gov: Types of Orders.

Open Advisory Research to explore a local sample report and inspect its assumptions before comparing your own saved evidence. Choose a synthetic sample before importing offline JSON. Keep outputs advisory; this exercise does not place trades. For background, read backtesting fundamentals and paper trading.

Continue the research

Intraday Trade Clustering: Resample Portfolio Days and Small Account Backtests: Test Capital Constraints develop adjacent checks. Use a small synthetic fixture first; these educational diagnostics do not validate a trading edge or change account settings.