Market Signals and EV Calculations
Figuring out if a bet is profitable or not involves far more than just probability calculations.
Here’s a sneak preview of a new product we’re working on, an itemized list of best bets according to our distribution-based analytics along with their estimated expected value:
A reminder of our color-coding scheme: green indicates a low-hold market, where the best odds across all sportsbooks result in a market with less than 2% vig, and orange indicates an arbitrage market, where the best odds across all sportsbooks result in a market with negative hold. Here’s a question for you, the reader: if you had to guess which bet is the most likely to be profitable from this list, which one would you pick?
You might reasonably guess Ivica Zubac’s prop, since it has the highest expected value in the list. But my guess would be Zach Lavine’s prop, since it has the most critical characteristic: it’s occurring in an arbitrage market. There are many concepts worth exploring in this exercise that illustrate how to think about finding value in betting markets, so we’ll go through it step by step.
First, let’s talk about how the expected value percentages in this list are derived. Betscope is synthesizing market information in near real time, aggregating what the sharp consensus is for each market, and estimating what the true line should be. Each book’s price is compared to the true line estimate, and bets that show a positive expectation are shown in our best bets list. We put a lot of thought into how to derive our true line estimates, and this process is always being refined and improved based on actual betting results. It’s a pretty good process! It’s made money in our live betting testing! But at the end of the day, our true line estimate is exactly that- an estimate, in the same way that literally everything in sports betting is an estimate. And like all estimations, the process to derive those estimations contains many potential sources of error. It could be any number of factors- a lag in the data feed, extenuating circumstances about a particular game that doesn’t fit our aggregation methodology, etc. At a high level, this applies to any calculations used in sports betting, whether it’s someone else’s best bets feed, a prop calculator or projection source, or handicapping your own numbers- whatever the process is, it’s going to have some sources of error in it, no matter how hard the creator of that process works to identify and minimize them (which they should absolutely do!) Successful sports betting is not about eliminating error from your process entirely, that’s impossible to do. Instead, it’s about embracing the reality that every process has error baked in, and navigating the markets while remaining conscious of this fact.
Which brings us back to the example from above. Some supplemental data points not previously shown using our chosen two bets: our estimated true line for the Zubac prop is -111 for the under and +111 for the over, and the best available price for the over is -115; our estimated true line for the Lavine prop is -105 for the under and +105 for the over, and the best available price for the over is +100. You can do the math yourself to see how those numbers produce the expected value estimates for each bet, as well as why the Zubac market is coded green and the Lavine market as orange. If our true line estimate was perfect, the Zubac bet would be an obviously much better bet than Lavine’s. But as we’ve outlined, our true line estimate is not perfect, because estimates by definition are never perfect, no matter how hard we try to make them. There’s a universe in which the actual true line- a fundamentally unknowable quantity that we’re all trying to make our best guess at- is actually something like +110 for the under and -110 for the over, which would result in negative expected value for both best prices of the Zubac prop.
By way of contrast, the Lavine market is an arbitrage market, which introduces a brand new level of certainty to our estimates. Here’s the thing about arbitrage markets: it is mathematically impossible for an arbitrage bet to exist without at least one book being far enough from the true line to produce a positive expected value bet. Go ahead, try it out for yourself: if the best price on Lavine’s under is +102 and the best price on the over is +100, plug in literally any value for a true line, and you’ll always get at least one of the bets to show positive expectation. An arbitrage market contains a powerful piece of information: at least one book has mispriced their market compared to the true line for that market, even if we don’t know exactly what the true line is. We can make some pretty reasonable guesses, as we do in our upcoming product, for what the true line is and which side of the market is profitable, but knowing that at least one book has a guaranteed mispricing brings a much higher level of confidence that our selected bet is profitable. This is also a way for you, the sports bettor, to utilize additional information in your betting process: you can incorporate signals from other sportsbooks to inform your process, even if you don’t have access to those sportsbooks.
If you just ran a set of EV calculations based on our true line estimates, the EV values themselves would be the same in that screenshot, and the calculations wouldn’t “know” that one of the markets is an arbitrage markets. That’s because expected value calculations by definition don’t need to incorporate that information, they’re just multiplying probabilities and prices together. But when you incorporate the additional layer of information that one of the markets is in an arbitrage situation, you have higher confidence that your expected value calculations are correct. It’s very difficult to convey that nuance into a betting calculator product, to say nothing of the fact that you’re arguably incentivized not to if you want to grow your business- people want certainty and locks of the week in their products, not some meta-reminder that your estimates might be wrong. But the best sports bettors in the world know their processes have errors in them, and seek every angle possible to minimize and reduce the cost of those errors- and that’s the best practice we’re incorporating into our products at Betscope. Proudly trumpeting that you have some errors in your process could very well be a marketing disaster for us, but it doesn’t change the fact that it’s the right thing to do to make everyone who comes here a more competent sports bettor, so we’re going to do it. As I’ve mentioned before, it’s actually liberating once you embrace error in your process- you’re less attached to seemingly pretty-looking products and gravitate more towards honesty in the process. And in the long run, marketing doesn’t make profitable sports betting tools, but honesty does.