Correlated Season Win Totals
Parlaying divisional teams in MLB Win Total O/Us can create +EV bets
The outcomes of season win totals for teams are not independent of each other. Because there are a finite number of wins available, we know that it would be impossible to bet and win 100% of under wagers. In fact, we can come up with a distribution of possible outcomes by simulating past seasons, but that is something for another day.
What we’d like to focus on is the correlation of win totals between specific teams in the MLB. The MLB schedule is designed such that a team plays each of its four divisional opponents 19 teams each (76 games). That team also plays 66 games against the 10 other league rivals and plays 20 interleague games.
The concentration of divisional games, however, gives us a situation where there is a significant correlation between win total outcomes between divisional rivals. For example, if the SF Giants go 19-0 against the LA Dodgers, the Giants have a good chance of hitting their win total “over” while the Dodgers have a good chance of hitting their win total “under”. Thus, if you were betting these two teams, their win totals are clearly not independent of one another.
If win totals between different teams were truly independent of one another, we would expect each instance to have a 50.0% chance of happening (assuming no win inflation) and we’d be able to simply multiply the probability of each outcome to determine the probabilities associated with the outcome matrix as shown below.
As displayed above, the odds of the Team A going over their win total (50.0% likelihood) and Team B going under (50.0% likelihood) would be 25.0% (50.0% x 50.0%).
However, we know that these events aren’t truly independent of one another. In fact, we can measure the correlation between divisional rivals’ win totals by simulating the season using Monte Carlo simulation techniques. After 50,000 simulations, we have determined that on average, when one divisional rival goes over their win total, a divisional opponent would go under their win total approximately 54.0% of their time. That gives us the following outcome matrix:
As you can see, the probability of both teams going under their win totals decreases from 25.0% to 23.0% when we acknowledge the correlation between the two.
How to Incorporate Into a Betting Strategy
If you parlay two -110 wagers, you will typically get odds of +264 (+264.46 to be exact). These odds are determined by multiplying the breakeven win probability on each leg of the parlay. With each leg priced at -110, you need to hit each leg at 52.4%, and therefore win 27.4% of the time (52.4% x 52.4%) to breakeven at +264 odds. These calculations are displayed below.
The pricing on parlays implicitly assumes that each leg of the parlay is independent of each other and you therefore only have a 25.0% of hitting that parlay. As a result, the expected return for a randomly selected parlay is approximately -8.9%.
A correlated parlay is not random, however. In fact, due to the correlation between season win totals for divisional rivals, if we were to parlay one divisional over with one divisional under, our win probability would be approximately 27.0% The difference between 27.0% and 25.0% may not seem like much, but it increases our EV from -8.9% to -1.6%.
Still -EV but getting close. In fact, this edge is so small now, it only takes some basic line shopping to move into +EV territory. Let’s say one book offers the Dodgers at Over 101.5 -110, and one offers Over 101.5 -105. Just having that one leg at -105 increases our EV from -1.6% to +0.6%.
Where Can I Bet on Correlated Parlays?
Unfortunately, most sportsbooks are wise enough to recognize the correlation and don’t offer parlays on season win totals. A few do, however.
5Dimes is one of the offshore books that allows parlaying of season win totals for divisional rivals in the MLB (and other sports). Below we’re sharing a couple (significantly) +EV parlays that you can hit on 5Dimes. While most people have a single go-to sportsbook, this is another instance where having multiple sportsbooks at your disposal gives you a long-term advantage.
Parlay #1: Phillies Over 84.5 (-105) and Marlins Under 64.5 (+115)
Breakeven %: 23.8%
Est. Win %: 27.2%
Expected Value (%): 14.3%
Analysis: Across six different sportsbooks (offshore and U.S.) the next-best odds to be found on Marlins U64.5 is +104 (FanDuel and Pinnacle). 5Dimes is 11-cents better than the next best offering on the Marlins. Most books are offering the Phillies at 85.5 with odds between -110 and +110. We’d rather take the 5Dimes line of Over 84.5/-105 and spend the 15 cents to cross the 85 instead of taking Over 85.5/+110 based on our assumed push probability of 4.5% for win totals. With the correlative properties increasing our win % an addition ~2%, we find there to be considerable value with this parlay.
Parlay #2: Cubs Over 85.5 (+100) and Cardinals Under 86.5 (+110)
Breakeven %: 23.8%
Est. Win %: 28.0%
Expected Value (%): 17.7%
Analysis: This one is more about the value on the Cardinals that the Cubs, to be honest. The offshore books are sitting at a heavily juiced 87.5 for the Cardinals under (Pinnacle U87.5 -158). FanDuel is the only other book offering Under 86.5, and they are offering -110. At +110, there is tremendous value on the Cardinals leg. The Cubs leg isn’t as profitable, but still approximately 7-cents better than consensus.
Conclusion: Using this framework can lead to endless correlative parlay possibilities with +EV, but it is critical to have multiple sportsbook accounts - finding the right sportsbook to take your bet is a key blocker from extracting value.