Sanity checks for every fair price you compute
Four cheap checks catch every broken margin removal: the sum, the bounds, the ordering and the distance from the raw price. Run them per market.
PNCLAPI
Dated notes on the numbers behind the Pinnacle odds API: fair odds, margins, conversions and request budgets.
Four cheap checks catch every broken margin removal: the sum, the bounds, the ordering and the distance from the raw price. Run them per market.
Decimal is the format the API serves and the only one your database should hold. American and fractional are display languages, translated at the edge.
Your model says 62%. What price makes that a bet? The conversion from probability to a fair decimal floor, and the margin check that follows.
The overround is the margin made visible. What its size says about a market, why it moves, and how to use it as a data-quality check on API prices.
Margin removal belongs on every snapshot, not the average. The per-row rule, the completeness check and the cutoff that make a fair series trustworthy.
Two margin-removal methods turn the same prices into different fair odds. See where proportional and power no-vig diverge, with a worked example.