What is the Optimal Bets core betting model?
The Optimal Bets model uses player projections to find inefficiencies in betting lines.
Step 1: Project a Team Total
Let's say LeBron James has a player prop of O/U 28 points with the Over and Under each set to -110. That player prop would imply LeBron is projected for 28 points.
If we can do this for all players on a given team, we can calculate an implied score for that team. Continuing with our example, let's say all the player projected points for the Lakers add up to 105 points.
Great! Now we have an implied total for the Lakers. But calculating a team's projected score is only the start.
Step 2: Project a Team Distribution
Our next step is to build a distribution of scores for a given team.
We do this by adding standard deviations into the mix. For brevity, we're not going to go into details, but ultimately we end up with a range of scores.
Let's say we calculate the Lakers to score anywhere from 60 to 150 points, with the numbers closer to the middle (105 points) occurring most commonly.
Awesome, now we have something useful!
Step 3: Project an Opponent Distribution
Next, we go through the same process for the Lakers' opponent.
We find the Clippers are projected for 108 points with a range of 63 points to 153 points.
In this overly simplified example, it should be clear that our betting model expects the Clippers to be a 3-point favorite.
Step 4: Compare Team Distributions
Once we have our team distributions, we can compare them to calculate the win percentage for given Spread and Moneyline lines. We can also merge those distributions and calculate the win percentage for Over/Under lines.
And once we have the win percentage for a given bet, we can compare that win percentage to the betting odds to calculate the expected value for that bet.
We can also take the win percentage and the betting odds for a given bet and plug them into the Kelly Criterion method to calculate the optimal bet size.
In our mobile app, we use the 1/4 Kelly Criterion to display bet units.
Wrapping it up
This is a simplified breakdown of how our core betting model works. It's worth noting that we're always learning from our data and looking for tweaks to make to our betting models. There are more variables we account for in our algorithm that would require a deeper dive like sport-specific adjustments.
For the NBA, if we don't have player projections for a full team, we'll normalize the projection data based on a full 48 minutes of play.
For the MLB, we normalize batting runs scored projections vs. pitching run allowed projections.
A great thing about this algorithm is it already takes into consideration many elements bettors look for like home field advantage, player hot streaks, and more because they're already baked into the player projections.
If you have any questions or would like to learn more, please feel free to send us an email at email@example.com