Why do EV betting tools disagree? Devig methods compared

EV betting tools disagree because each one estimates the "true" probability its own way. They can use different reference sportsbooks, remove the margin with different devig methods, and capture odds at different moments. The same bet can look clearly +EV on one tool and break-even on another, with neither tool making an arithmetic mistake.

Why do EV betting tools disagree? Three choices

Every EV number is your price × an estimated fair probability − 1. The price is the same everywhere; the fair probability isn't. It depends on:

1. The reference market. Which sportsbook or blend of sportsbooks is treated as the best estimate of the truth.
2. The devig method. How the reference market's margin is removed (below).
3. Timing. Odds change constantly. Two tools that read the market seconds apart can see different prices, which matters most for fast-moving markets such as live betting; see How to evaluate scanner alerts and stale odds.

Devig methods compared

Researchers have compared several ways of turning odds into probabilities. The normalization (proportional) method scales all implied probabilities by the same factor, which "does not account for favorite long-shot bias." The additive method subtracts an equal amount from each, which can produce negative probabilities. The power method raises each implied probability to a common power and allows for that bias.^[1]

Here's what that means for one market. The reference sportsbook prices a favourite at 1.25 (-400) and a long shot at 4.50 (+350). Another sportsbook offers the long shot at 5.00 (+400).

At 4.80, the same bet is +4.3% under one method and slightly negative under another. The further a price is from even money, the more the methods disagree.

Which method is right?

None is right for every market. In the comparison above, the authors found that "the power method universally outperforms the multiplicative method and outperforms or is comparable to the Shin method."^[1] For two-way markets, they note that the Shin and additive methods are equivalent.^[1]

The practical test is robustness: a bet that stays +EV under several methods and reference books is more trustworthy than one that only appears under the most generous assumption.

Check a number yourself

Enter the reference market's two prices in the no-vig calculator (it uses the proportional method) and compare the result with what your tool shows. If they differ a lot, the tool is probably using another method, another reference, or newer odds. For how no-vig probabilities work, see Implied probability and no-vig odds, explained.