Was It The Signal Or The Execution?

A trade loses money and the post-mortem needs to answer one question before anything else: was the idea wrong, or was the idea right but the execution gave it away? These call for completely different fixes. If the signal was wrong, you go back and question the model or the thesis. If the signal was right but execution ate the edge, the model is fine — you need a better trading plan, not a better idea. Conflating the two is how a genuinely good strategy gets abandoned for the wrong reason, or a genuinely bad one keeps getting re-run with "better execution" that was never the problem.

The split

Compare three prices: the price when the signal fired (the decision price), the price you actually got filled at, and the price where the position eventually closed.

$$\text{PnL}_{\text{total}} = \underbrace{(P_{\text{decision}} \to P_{\text{exit theoretical}})}_{\text{signal PnL}} - \underbrace{(P_{\text{fill}} - P_{\text{decision}})}_{\text{entry slippage}} - \underbrace{(P_{\text{exit theoretical}} - P_{\text{exit fill}})}_{\text{exit slippage}}$$

In words: what the trade would have made at the prices the signal actually pointed to, minus whatever you gave up getting in late or expensive, minus whatever you gave up getting out late or expensive.

Worked example

A signal fires to buy at $50.00 (the decision price). By the time the order works through the market, you're actually filled at an average of $50.18 — that's $0.18 of entry slippage per share. The position is held and the signal's target exit price is $52.50, but by the time you actually sell, the market has slipped to $52.10 — another $0.40 of exit slippage. You traded 20,000 shares.

• Signal P&L (decision price to theoretical exit): $20{,}000 \times (52.50 - 50.00) = 50{,}000$, i.e. $50,000.
• Entry slippage cost: $20{,}000 \times 0.18 = 3{,}600$.
• Exit slippage cost: $20{,}000 \times 0.40 = 8{,}000$.
• Actual realized P&L: $50{,}000 - 3{,}600 - 8{,}000 = 38{,}400$.

The signal was worth $50,000. Execution gave back $11,600 of it, about 23 percent. This trade wasn't a mediocre idea that happened to make money — it was a strong idea that execution meaningfully taxed. The fix is a better order-working strategy (more patient entries, smaller clips, better timing around liquidity), not a change to the signal.

Making it routine

Log decision price, fill price, and theoretical versus actual exit for every trade, not just the losers — a strategy that's consistently strong on signal P&L but consistently weak after slippage is telling you something structural about how it trades (too aggressive, too large relative to liquidity, too slow to react), and that pattern only shows up once you've done the split many times over.