Sunday, March 2, 2008

Why Projection Systems are Useless

When someone says projection systems, I think of words like Chone, Marcels, and Pecota. A lot of baseball sites pridefully boast that they feature the new year's player projections from these systems. I immediately leave those sites.

The reason I dislike these systems is not because they are wrong. In fact, according to what I've heard, they are fairly accurate. The reason I dislike them is because the systems themselves are extremely predictable. As a hitter gets older, decrease his power and speed; as a pitcher gets older decrease his K rate. I don't need a projection system to tell me Carlos Lee's numbers will be around .300 avg, 30 homers, 100 rbi/runs, and 10 SB. When it comes to bold predictions, projection systems like Chone, Marcels, and Pecota have none of them. For a fantasy owner, that makes them irrelevant.

Any moron can draft dependable players and finish respectably in a fantasy league. But if you have the desire to win, you must pick players with upside, the Ryan Brauns and Fausto Carmonas of last year. Those are obviously extreme examples, and those two were highly improbably; so improbable that in fact it would have been unwise to spend even a last round pick on them. Those players are the rewards for persistent owners that constantly check updates and are ready to pounce on the f/a market.

I'm talking about players this year like Frank Thomas, Jason Giambi, Evan Longoria and Clay Buchholz. There's a fair chance they do nothing or play mediocre, however there's that small chance they play amazingly. It's those value picks that win you fantasy leagues. A quote from Matthew Berry puts it best, "You can't win the league in the first round but you certainly can lose it."

Notice how my player examples included older veterans like Thomas and Giambi. I'm not a big fan of drafting rookies like Longoria, as they usually take a few years to adjust. Last year was simply a uniquely crazy year where many rookies thrived in the bigs. Don't expect that to happen again, to the same extent at least. However proven players, usually ones who have recently suffered injuries, often prove to be the best sleeper candidates. I'm not saying target players who were recently injured, that's stupid, but certain formerly talented players fall so far in drafts that they become worth a late round gamble.

The aforementioned quote by Mr. berry leads me to my second point of this article, avoid players who have a high risk rate. That means don't pick Albert Pujols in your draft this year! It's simply just not worth it. Your first/second round picks should have virtually no risk attached to them. Pujols is very, very risky with that elbow, and even his best case scenario isn't much better than your alternative options. His worst case scenario is, however, much worse. Avoid.

Projection systems usually fail to recognize these types of players. Lists of break-out/collapsing candidates are much more valuable than viewing whole projection systems. Most of them just average a player's stats over the last few years, placing more weight on recent years. Although fairly accurate, that isn't exactly useful, as long as you are not a moron.

I believe baseball prospectus (a site I am not a subscriber to, because of monetary issues) lists a collapse rate for all the players. This is extremely useful, and apparently they do decent job since they listed Jason Bay as probably candidate for a collapse last year. If you are fortunate enough to be a subscriber, avoid any player with a high collapse rate, and target those with high break-out rates. That was an obvious statement, but too many people ignore those percentages and just listen to a projection system that takes no risks. Don't be that guy.




Anonymous said...

Great post. I agree 100%.

Swishmynish33 said...

Thanks for your comment, its always nice to hear some positive feedback. I've noticed you added a link to my site on yours, so I've returned the favor.