To calculate the expected value of a bookmaker's price for a specific outcome, we must first determine the probability of that outcome occurring. The expected value is based directly on the probability of a sporting event occurring, along with the price offered by the book. How can we actually determine these probabilities? We ultimately look to sports betting experts, or 'sharps', who use their own complex modeling to influence market prices on their favorite books.
Let's say we are evaluating Cowboys -3.5 spread over the Eagles. Specifically, we need to ask: What is the probability of the Cowboys beating the Eagles by 4 or more points?
Outlier conducts this analysis by crawling tens of thousands of prices in sharp markets, where market makers adjust to expert sports bettors. By continually shifting lines, these market making sports books reach a state of near-perfectly priced markets. These prices, once translated into implied probabilities, represent the likelihood of various outcomes.
However, sportsbooks need to make money. Thus, embedded within these prices is a 'vig,' or guaranteed profit margin for the book. To reach an accurate set of probabilities, this vig must be removed through a process known as 'devigging' (otherwise, each outcome appears more likely than it actually is). The way the vig is allocated to each outcome during devigging becomes essential to the predictive accuracy of the market price.
There are four popular methods for devigging, and choosing the right one depends on the specific market circumstances and your historical success rates. The method you select should align with your evolving EV betting strategy.
Multiplicative Method
The vig is spread proportionally among each outcome, so outcomes with higher implied probability (lower odds) get more vig.
This method is the most popular due to its mathematical simplicity, and is considered the standard approach to devigging.
This method can be problematic as it fails to account for the known tendency of bettors (even sharps) to overbet on long-shots and underbet on favorites.
Additive Method
The vig is divided equally among all outcomes.
While this takes bias to overbet on long-shots into account, it can sometimes overweight this bias, resulting in negative probabilities for underdogs.
Shin Method
Utilizing an iterative algorithm, this method seeks to correct the favorite-longshot bias, where favorites are underbet, and long-shots are overbet.
Generally offers improved predictive accuracy, especially over the Multiplicative method.
For markets with two outcomes, it is equivalent to the Additive method.
Power Method
This method is an extension of both the additive and multiplicative methods and involves raising the probabilities to a constant power.
Beneficial because it always maintains probabilities within the 0 to 1 range, avoiding feasbility issues present in other devigging methods.
However, it can overcompensate for betting biases, adjusting long-shot probabilities more than Shin, while adjusting middle-range outcomes less than Shin.
By understanding these methods, one can better assess which technique may be the most effective for a given market. What do we recommend? The honest answer is we are still conducting our own research on these methods. By analyzing millions of historic lines we plan to build models of when each method carries the most predictive weight, based on a variety of market criteria, and incorporate those predictions in the Outlier Pro offering.
Want to explore further? Outlier’s algorithms are based on this research: Stephen Clarke, Stephanie Kovalchik, and Martin Ingram: "Adjusting Bookmaker’s Odds to Allow for Overround." American Journal of Sports Science, Vol. 5, No. 6, 2017, pp. 45-49. doi: 10.11648/j.ajss.20170506.12