## Where Does P&L Come From?
A bettor who makes money has generated that return from some combination of:
1. Model edge (probability estimates better than the market)
2. Price hunting (consistently finding better prices than the market average)
3. Timing (placing bets at better prices than later available)
4. Variance (lucky outcomes in excess of expected value)
Attributing P&L correctly to these sources tells you what to invest in to improve performance.
## The Attribution Methodology
**Model edge attribution:**
Compare model probability to Pinnacle closing probability. Average CLV across all bets is the model edge contribution.
**Price hunting attribution:**
Compare price taken to Pinnacle closing price. Average of (price taken / Pinnacle closing − 1) across all bets is the price hunting contribution.
**CLV = Model edge + Price hunting** (approximately)
**Actual ROI = CLV + Variance**
## The Insight
If CLV is positive but actual ROI is negative: variance is responsible. Continue operating — the model is working.
If CLV is negative but actual ROI is positive: variance is responsible for the profit. The model is not working — do not be deceived by the results.
If both CLV and ROI are positive: model edge confirmed. Scale appropriately.
## The Model Quality Benchmark
The model's contribution to CLV (estimated by comparing model probability to market probability) is the cleanest measure of model quality. A model that consistently outperforms the closing line is contributing genuine value — regardless of results in any specific period.
Create a free account to track your progress and save bookmarks.
