Wednesday, February 28, 2007

Earnings vs. Predictions

One of the old adages in Rotisserie baseball is that you can't simply look at what a player did last year and assume that is what he's going to do this year.

I disagree. On the contrary, last year's numbers will give you a suitable baseline to predict what most players are going to do this year.

If you look at most analysts, this is generally how they complete their predictions as well.

Since Alex Patton's predictions for this year are still
for sale, I can't give too much away by writing about them here. So I'll look to last year as a barometer for how 2005's earnings led to 2006's predictions.

Top 10 Alex Patton Predicted Hitters 2006

RankHitter'06 Prediction'05 Earnings+/-
1Albert Pujols$43$47-4
2Alex Rodriguez$40$49-9
3Carl Crawford$36$40-4
4Vladimir Guerrero$36$360
5Derrek Lee$36$48-12
6Ichiro Suzuki$35$34+1
7Juan Pierre$34$340
8Jason Bay$34$40-6
9Alfonso Soriano$34$37-3
10Todd Helton$33$25+8

These are the top 10 predicted hitters by Alex in 2006. With the exception of Todd Helton, all of them were expected to either slip a little or stay the same. Intuitively, this makes sense. Derrek Lee's $48 season in 2005 was almost definitely a career year, but predicting that he would do the same thing two years in a row would almost surely be folly. However, no one in Alex's 2006 top 10 is radical. Helton in hindsight might seem radical, but Alex was hardly the only tout who predicted that Helton would bounce back after a slightly down 2005.

Top 10 Alex Patton Predicted Pitchers 2006
RankPitcher'06 Prediction'05 Earnings+/-
1Mariano Rivera$41$49-8
2Billy Wagner$38$44-6
3Johan Santana$38$40-2
4Brad Lidge$36$38-2
5Chad Cordero$34$44-10
6Joe Nathan$33$42-9
7Francisco Rodriguez$33$36-3
8Huston Street$33$32+1
9Pedro Martinez$32$39-7
10Jason Isringhausen$31$32+1

Pitchers get ripped off even more when moving from earnings to predictions, but this again makes sense intuitively. The Top 10 predicted list of anything is bound to have some failures, and the pitching list is bound to be the most failure-oriented list of them all.

Just out of curiosity, I'll look at another grouping of 10 hitters, just to see if there's any further variability.

71-80 Alex Patton Predicted Hitters 2006
RankHitter'06 Prediction'05 Earnings+/-
71Adrian Beltre$20$14+6
72Raul Ibanez$20$22-2
73Jeremy Hermida$20$4+16
74Jacques Jones$19$18+1
75Ken Griffey Jr.$19$28-9
76Brady Clark$19$22-3
77Troy Glaus$19$25-6
78Luis Gonzalez$19$20-1
79Tadahito Iguchi$19$21-2
80Chipper Jones$19$21-2

This list is far less radical in terms of the predictions differing greatly from the previous year's earnings. The outliers are Hermida (rookie who had 41 AB in 2005), Griffey (injury history), Glaus (injury history), and Beltre (expected bounce back year).

Generally speaking, the guys who will require you (or your favorite expert) to do the most work are:

1) Rookies or second-year players
2) Players coming off of an injury or with an injury history
3) Players coming off a career year or an off-year
4) Players with a reduced or increased role

Let's look at an example.

Howie Kendrick batted .285 last year in 267 major league AB. His line was 4/30/6. Pumping that up to 500 AB (which I'll conservatively assume for him this year), gives me 7/56/11/.284. The Patton $ for such a projection is $14. Personally, I think Kendrick will put up a better season than that. I'm thinking he'll earn somewhere between $16-18. Assuming a midpoint of $17, I'll give him a 12/80/10 line with the same BA and see what happens.

That puts him up to $17. I'll bring his BA down (he is a great prospect, but he is only 23) to .274. He's now projected to earn $16.

I'll leave that for now.

One of the reasons I use Patton $ On Disk is because I don't have the time or inclination to do this for 1,000 players across both leagues. But you can (and should) do the same thing with any expert's projections who are using. If a projection looks high to you, don't be afraid to knock it down a couple of bucks. If a projection looks low, and you think your favorite rookie or oft-injured player is going to find the magic this year, pump him up a few bucks. If anything, projections are a safe place to play around and do this. Want to see what Albert Pujols looks like as a $50 player? The Patton software will let you play around until you can see it. That doesn't mean you have to pay $50, even if you think he'll earn it.

Which brings me to my next point.

If you've been playing Roto for long enough, you know that my $16 projection for Kendrick is not the price I'll necessarily pay for him in the auction. Tomorrow, I'll move on to predictions and bids and discuss how, and why, they're different.

No comments: