Earlier in 2015, I proposed a system of projecting wins by using two sabermetrics projection systems, fWAR & Steamer, and applying them to the MLB futures for win totals. The results are in.
To recap, to calculate these projected win totals I used a regression based on the projected total WAR for each roster. WAR stands for Wins Above Replacement, or how valuable a player is related to a replacement level player. The regression formula to calculate the win total was based off of the statistical work of sabermetrics expert, Glenn DuPaul.
First let’s will review the results using the Steamer WAR projections. This projection system was more bullish than the fWAR system, as it projected that 21 teams would go over the O/U total listed at Bovada. That left just 9 teams with a projected record that warranted a bet on the Under season win total. However, of those 9 teams suggested to go under the win total 8 of them in fact did. 8-1 is a pretty strong result and the only team that bucked the trend is the New York Mets. The Over suggestions were 11-10, not a disaster, and overall the system was 19-11 in choosing the right direction for the season MLB picks win total.
Here are the full results using the Steamer WAR projections:
Steamer Projected Win Totals vs Bovada O/U Wins
|TEAM||Steamer Projected Wins||O/U (Bovada)||Diff.||Lean||Actual Wins||Result|
The fWAR projections from Fangraphs had more of a lean towards picking the under, but still was a system that was successful with an overall 20-10 record. Again, the Under leans had the best results at 13-5, against just 7-5 for the Over leans. The reasons for this is likely due to the public betting markets allowing the books the ability to inflate O/U win totals on popular teams. True talent evaluation and mathematics will throw that bias out.
Here are the full results using the fWAR projections:
fWAR Projected Win Totals vs Bovada O/U Wins
|TEAM||fWAR Projected Wins||O/U (Bovada)||Diff.||Lean||Wins||Result|
I’ll definitely be repeating this next year as I made quite a bit of coin wagering on MLB futures this year. Next year I’ll also likely hammer the Under projections a bit more, as they appear to be have the strongest correlation in the model.