Market-Anchored Probabilities: Starting From the Price
Why start from the market
A betting market pools the views of many people who put money behind them, and its closing price is hard to beat. A model built only on team ratings usually scores worse than the market's own price, so the honest starting point for a winner probability is the price itself, with the margin taken out.
From a moneyline to a probability
- Implied chance, home (150 / 250)
- 60.00%
- Implied chance, away (100 / 230)
- 43.48%
- Sum (the margin is the part over 100%)
- 103.48%
- No-vig home
- 57.98%
- No-vig away
- 42.02%
- No-vig home as a fair price
- -138
The no-vig home chance, 58.0%, is the anchor. A model may move it, but only by as much as testing has justified.
What Probetrics does
Where a sportsbook prices a game, the published winner probability follows the market's no-vig price. In football the model's own rating is mixed in lightly, in basketball it decides only games with no price, and for UFC no blend of the model with the market beat the market alone in testing, so a priced bout uses the market's no-vig chance and the model decides only unpriced bouts. The model's own number stays visible on the prediction page. Exact weights are not published.
This is also why Probetrics shows no edge over the market on UFC and why it makes no spread or over/under call. The record is on winners, locked before the start and graded in public.
Questions and answers
- What does market-anchored mean?
- A probability that begins with the sportsbook's price, margin removed, and departs from it only when a model has been shown to add information.
- Why not just use the model's own probability?
- In testing, a model built from ratings alone scored worse than the market's price. Anchoring keeps the strength of the price and uses the model where it is useful, such as games with no price.
- Is a market-anchored pick the same as the market favorite?
- For a priced game it is close to it, because the probability follows the price. The record is on winners, graded in public, and it is compared against the market rather than against nothing.
Read next
- No-vig oddsNo-vig odds, also called fair odds, are a two-sided price with the sportsbook's margin (the vig, or juice) taken out, so the two sides' chances add up to 100%.
- Implied probabilityImplied probability is the chance a price says an outcome has, found by converting the odds, and it puts American, decimal and fractional prices on one 0 to 100% scale.
- CalibrationCalibration is whether stated probabilities match what happens: of the picks a model gave about 70%, about 70% should win.
- Walk-forward backtestingWalk-forward backtesting scores a model on games it has not seen by pricing each game only from what was known before it, moving forward through time.
This page explains arithmetic and vocabulary. It does not recommend a bet, a side, an amount or a sportsbook, and the examples are illustrations. For information only, not betting advice. 21+ only. Gambling problem? Call or text 1-800-GAMBLER.