Although there have been recent talks about restarting the MLB season in May, there remains considerable uncertainty around what the MLB might look like. Shortened season? 7 inning games? Expanded rosters? questions than answers, but one thing we know is that a standard 162-season is likely not in the cards. As a result, most sportsbooks have pulled their season win total wagers for the MLB season.
To still give sports bettors something to wager on, DraftKings sportsbook replaced its MLB Win Totals with MLB Win Percentages. As long as a minimum of 60 games are played, these wagers will have action. We give DraftKings credit for getting creative and offering these wagers to their customers.
Assessing the Market
We decided to take a closer look at this offering from DraftKings and compare it to the win totals offered by sportsbooks in March. To make an apples-to-apples comparison, we converted the March consensus Win Totals to winning percentage and then compared them with the Win Percentage offerings from DraftKings. As you can see below, the Win Percentages are almost identical to the market consensus win totals offered in March.
Not a single offering was more than a one percent difference from the implied win percentages offered by sportsbooks in March.
The next things we wanted to assess was the DraftKing’s theoretical hold (or vig) on these markets. As you can see below, the standard offering is -112 on each side of these wagers.
The sportsbook hold (or vig) on a two-sided -112 market is around 5.4%. If you want to learn how to calculate the hold yourself, you can learn here.
For comparison, in March, the hold on the MLB Win Totals market ranged from 4.5% - 5.2% across legally operating U.S. sportsbooks. Although the 5.4% hold is slightly higher than the hold in the Win Totals markets, it’s still a far superior way to speculate on a team’s performance than the World Series futures, which routinely have holds north of 20% (currently 25.2% at one unnamed U.S. sportsbook).
The next thing we want to assess is whether the lines or odds are shaded toward the over. This is a very common strategy by sportsbooks as they take advantage of sports bettors’ tendency to prefer betting “Over” win totals rather than under. We discussed this strategy previously in our win inflation article.
For the Win Percentage offering from DraftKings, we calculated the average Win Percentage and odds across all teams. If there was no shading in either direction, we would expect the average Win Percentage to be exactly 50.0% and the average odds to be the same on the over and under. This is what we found:
To the naked eye, it looks like nominal win inflation, at best. However, if we normalize a 50.2% win percentage across a normal 2,430 game season, this is the equivalent of 8.9 games of win inflation across the league. Still may not sound like a lot, but using push probabilities, we can reallocate the estimated vig between over and under bets as follows:
Ah – well these numbers start to be meaningful. Yes, the average vig for this market is 5.4%, but the over wagers carry the majority of the vig due to the shading. As a result, under bets are burdened with only 2.9% vig. This is similar to the Win Totals market, where we calculated that the average hold on Win Total unders to be 0.5 % - 2.9%.
Great - we’ve confirmed that the Win Percentage markets are very similar to the Win Totals markets. So how are we going to find an edge?Typically finding an edge requires some creative thinking. Well - here’s one idea we’ve come up with:
Over the last two seasons, the National League has been superior in interleague play winning 52.7% of interleague games in 2018 and 55.3% of interleague games in 2019. What does the market expect to happen in 2020? Let’s turn to the Win Percentages offered by DraftKings.
The average win percentage offered by DraftKings for NL teams and AL teams is 50.7% and 49.7%, respectively. After removing win inflation and normalizing for a 2,430-game season, this equates to 1,227 wins for the NL and 1,203 wins for the AL. Since there are only 300 interleague games played each season, we can attribute the win differences between the two leagues to those interleague games (since the average win percentage in league games for each league will be 50.0%). This implies that the NL is expected to go 162-138 in interleague games, good for a 53.9% win percentage.
If MLB cannot play a 162-game season, however, which games are you going to chop? My guess is the interleague games are the first games that will be removed from the schedule. Apparently, I’m not the only one who thinks that interleague games could go this year.
If we were to assume that all interleague games were removed (and everything else remained), we might estimate the vig allocation for the NL as follows:
At average odds of 50.7% would estimate the win inflation at almost half a win per team in the NL, bringing the vig on NL Win Percentages close to only one percent. If other books start offering lines on this, you can almost certainly get to positive expected value with a little line shopping. Even if this is the only book that has this offering, it would be a good place to focus your handicapping efforts as this is a pretty good market, all things considered.
Caveats and Extensions
Certainly, this analysis relies on a variety of assumptions. One of those assumptions is that interleague games are removed. If interleague games aren’t removed, betting the NL Unders is probably no better than betting the AL Unders.
Second, we make the assumption that only 20 games are removed from each schedule per team. Should fewer games be played, we would expect the push probability for win totals to increase, but we’d also expect more variance in final win probabilities. It’s hard to tell the net effect of fewer games.
Third, if this wager type is maintained once the total number of games is determined, it might behoove you to pay close attention to exact win percentages for a particular number of wins. For example, if the season is 110 games and a team wins 56 games, their win percentage is 50.9%. If you bet under 50.5%, you’re a loser. But if you bet under 51.0%, you’re a winner. That extra 0.5% between 50.5% and 51.0% is significant. On the other hand, if that team wins 57 games (one additional game), their win percentage is 51.8%. 51.0%and 51.5% are both losers. Thus, there is no difference between 51.0% and 51.5%if there are 110 games. Just something to keep in mind.
Last thought – this same thought process can be applied to the analysis of win percentages for particular divisions. The AL Central, for example, is particularly weak this season. Their average posted win percentage on DraftKings is 47.8%. If for some reason, the schedule is changed to have a much higher percentage of divisional games, there may be some value on the Win Percentage Over for AL Central teams.
If you’re betting these markets, I hope this helps you build a strategy to squeeze out as much value as possible.