Monday, March 02, 2009

Thoughts on Pre-Season Projections

After my write-up on the CBS Sportsline analyst auction two weeks ago, Eugene thought I was being a little greedy.
I don't know Mike. Winning by nearly 20% more points than 2nd place seems like domination by just about any standard.

Maybe you were looking for a 100 point team...
In reality, I'd love to buy a 120-point team, but I know it's not going to happen.

The biggest problem with pre-season projections is that they don't prove anything. Some of them are likely to be fairly accurate, some will be uncannily so, while others will be so far off base that when we look back in October at our respective leagues, we'll be dumbfounded at how bad some projections were.

Tom Kephart of Baseball HQ put together an amazing write-up about his strategy and tactics that I strongly recommend reading. One thing that jumped out at me was the fact that Tom believes that:
I left the auction with 59 of 60 possible projected hitting points, using Baseball HQ player projections. The only offensive category where I am not projected first is SB, where Greg Ambrosious of the NFBC...is projected to edge me out by one SB.
Both the Sportsline and Alex Patton projections don't see it this way. Patton's projections put HQ at 47 points on offense, with a first place finish in RBI, second place in HR and runs, sixth in batting average and seventh in steals. Sportsline's projector is even less kind, giving HQ 41 points and putting them 3rd in runs, 4th in HR/RBI, sixth in batting average and seven in SB.

The Patton projection agrees with Kephart and puts him a good 200 AB ahead of the next team on offense, which is always a good start. Your goal in 5x5 is to keep moving the chains, every day of every week of every year.

The top finish in SB/BA for HQ is disputed by two other projection models, so it's worth looking into. As far as steals go, HQ looks a little light. Rios could steal 32 again, but 20-25 seems more like it for him. After that, there isn't a true big-timer runner on Kephart's team. It's also hard to believe that HQ might win batting average, as Placido Polanco is the only true .300 hitter on the HQ's team, and there is a lot of potential for variability elsewhere.

I don't bring this up to malign Kephart's auction (which I thought was very good) but to point out that pre-season projections tend to be optimistic, don't take injuries into account, and probably don't do much in Roto except for give us a sense of confidence that may or may not be deserved.

So, to get back to Eugene's statement, I'd sure love a 100 point team, because this means that I've done everything right, but because I also know that the law of averages is going to knock one or two of my players down a peg if I'm lucky and knock more off if I'm not.

My preference is to use bid limits for this reason. I'd rather not see what middle reliever is projected to have a $15-20 season, or at least not pretend that I'm going to "get" those $20 worth of stats because that's what it says on the sheet. Just like in "real" baseball, projections are pretty, but there's a reason Major League Baseball plays all of those games once the season starts: to see which fantasy baseball owner actually has the best Rotisserie team.

1 comment:

rafi said...

Can you describe what system you use to arrive at your bid limits? Like you mentioned in a past post your limit on Pedroia was $31. Because I know you approach this systematically and mathematically, I am curious how you arrived at that figure? Ultimately it's still projections that set your bid limits, right?