The All-Star Break presents us with a great opportunity to review how well our MLB projections have done so far this year. Unlike our highly profitable 2011 NFL projections and 2011 College Football projections, our 2012 MLB projections are just doing … fine. They’re basically right at par compared to the Vegas lines, which is about what we expected — and one of the main reasons we didn’t hype them at all heading into the season.
Before we get to the numbers, let’s take a quick overview of what we think are the main reasons our 2012 preseason projections (our first year doing them algorithmically) aren’t significantly outperforming the Vegas preseason win totals lines.
Areas For Future Improvement In The Preseason Ratings
1) Roster turnover is more difficult to quantify in pro sports than in college. All of our preseason ratings use a weighted average of the team ratings from the past few seasons as a starting point. This works great for college sports — nearly every new addition can be quantified in a “recruiting class” variable, and most truly important departing players can be taken into account in either a “returning starters” or “draft value” variable. That’s not the case for MLB, NBA, or NFL projections, where free agent signings muck up the waters. In the NFL, we manage to capture the most important position (QB) in a separate variable, but there is no single position in baseball that’s far more important than the others (counting each starting rotation spot as a separate position).
2) Individual players matter a lot in MLB, and our 2012 projections didn’t take rosters into account at all. Roster projection certainly ranks up there as a top priority for future enhancements of our MLB preseason predictions. Aside from the QB variable in the NFL, and some “returning stats” variables in the college sports, we haven’t incorporated individual player stats into our preseason ratings. That’s not a deal-breaker in the sports I just mentioned, because the rosters have a lot of continuity, and because coaches and systems matter quite a bit. In baseball, however, the whole really does equal the sum of its parts, so until we add player ratings into the mix, we’ll have a tough time creating market-beating projections.
The Current Projections Should Be Much Better
By now, actual games played are the most important factor in our MLB projected standings, and the preseason projections only play a small part. That should mean that the current projections will fare a bit better than those from the preseason.
Though it’s not as if those preseason projections did badly. If we grade our preseason projections by assuming that the current Vegas win totals are 100% accurate, then we basically broke even on those preseason projection, as you’ll see in the table below.
This shows the preseason win total projections of both Vegas and TeamRankings (TR) along with how far we differed from Vegas, whether each projection was “correct” (based on the current Vegas lines), and what our latest projections — and the latest Vegas lines — say about all the races.
|Team||Vegas (pre)||TR (pre)||Diff||Correct direction?||Vegas (now)||TR (now)||Diff|
|Boston Red Sox||87.5||87||-0.5||YES||85.0||83||-2.0|
|Chicago White Sox||77.5||76||-1.5||no||88.5||88||-0.5|
|Kansas City Royals||78.5||78||-0.5||YES||73.5||74||0.5|
|Los Angeles Angels||89.5||82||-7.5||no||91.5||87||-4.5|
|Los Angeles Dodgers||81.5||82||0.5||YES||86.0||85||-1.0|
|New York Mets||74.5||80||5.5||YES||84.5||85||0.5|
|New York Yankees||93.0||91||-2.0||no||96.0||96||0.0|
|San Diego Padres||70.5||81||10.5||no||66.5||68||1.5|
|San Francisco Giants||87.5||82||-5.5||no||88.0||85||-3.0|
|St Louis Cardinals||87.0||86||-1.0||no||87.5||89||1.5|
|Tampa Bay Rays||87.5||86||-1.5||YES||84.0||84||0.0|
|Toronto Blue Jays||81.5||81||-0.5||YES||78.5||81||2.5|
What do you think about the mid-season Vegas lines? Do you see anything you disagree with, or do you think Vegas is right on the money? Which numbers do you like better — ours or Vegas? Leave any thoughts in the comment section below!