Baseball prediction systems are not new.  They’ve been around for a while.  But this year, they seem to be coming out of the woodwork.  Either that, or I’m just paying closer attention.

Baseball Prospectus has what they call their PECOTA system.  According to their website:

(PECOTA) (s)tands for Player Empirical Comparison and Optimization Test Algorithm. PECOTA is BP’s proprietary system that projects player performance based on comparison with thousands of historical player-seasons. PECOTA analyzes similarities with past player-seasons based not only on rate statistics, but also height, weight, age, and many other factors.

BP’s PECOTA system predicts that the Cubs will win between 77-82 games in 2010 and will finish third in the NL Central behind the Cards and Reds.  Ouch! 

But take heart Cubs fans.  PECOTA predicted that the Cubs would win 92 games last year.  That predication didn’t work out too well.

Since it’s inception in 2003, PECOTA has been off an average of eight games (+/-) on predicting the Cubs season record.  That means the Cubs could win between 71-90 games in 2010 and PECOTA would remain as accurate as they have been historically.  It’s not really much help, is it?

The CAIRO system (CAIRO v.03) predicts that the Cubs will win 83-84 games in 2010 and will finish in second place in the NL Central behind St. Louis.  CAIRO is touted as the most reliable of the baseball prediction systems (based on 2009 results).  They beat out THT, PECOTA, ZIPS, CHONE, and Marcel (poor Marcel).

Although I think I pretty much agree with the CAIRO prediction for the Cubs, I still have a hard time taking it seriously.  First, the predictions are made fairly early in the off season, even before all free agents are signed and rosters are set.  To CAIRO’s credit, they warn readers of this weakness right up front.

Second, there are simply too many variables that can happen during the course of the season to accurately predict what is going to happen.  For instance, what happens to the predictions for the NL Central if Albert Pujols goes down early in the year with a season ending injury?  What if Chase Utley and the Phillies have a falling out and Utley is traded to the Cubs at the trade deadline (my personal baseball fantasy)?  Each prediction system admits that they can not account for unforeseen injuries, but the truth is that each year, many teams have unforeseen injuries.  And the key word there is “unforeseen.”  They can’t be anticipated or fully planned for, either by the teams or the prediction systems.

I understand the desire to be able to predict the future.  Our desire to do this goes far beyond the confines of baseball.  Even so, our ability to predict the future has been shown time and time again to be somewhere between piss poor and futile.  Just ask all of the people who have designed a “system” to pick stocks.

My favorite sportswriter, Joe Posnanski, has developed a baseball prediction system of his own.  He describes it in perfect detail when he says:

In my prediction system, I basically use a statistical and scouting bouillabaisse to rank the players on each team. And then I … well, look, I can’t remember the whole thing right now. All I can tell you is that I rank players, add some stuff together, subtract some stuff, multiply by pi (or divide by pi) … and … voila … a baseball prediction system!

It’s the perfect Hot Stove system … perfect, because it’s pointless and ridiculously flawed and I’m fairly certain (and fairly hopeful) that people will have forgotten all about it long before the baseball season actually begins.

So, using the Posnanski system, how did the Cubs do?  Pos has the Cubs winning 86 games and finishing second in the NL Central behind the Cards.  I think that makes sense.  But here’s the interesting thing.  As Pos says in his quasi-explanation of his prediction system, he ranks players on each team.  Where he goes from there is anyone’s guess.  But in his article, he lists the best players at each position in each division.  And much to my surprise, Ryan Theriot was listed as the best shortstop in the NL Central.  Can that be right?

As you probably know, although I like Theriot’s style of play (he’s hard-nosed and gritty), I don’t think he is a particularly good shortstop.  In the very least, he’s not the quality of shortstop that a team with a $140 million payroll should have in their everyday lineup. 

As much as I like and respect Joe Posnanski, I had to take a closer look at his assertion that Ryan Theriot is the best shortstop in the NL Central.

In addition to Theriot, the other starting shortstops in the NL Central are:

• Tommy Manzella (Astros)
• Alcides Escobar (Brewers)
• Brendan Ryan (Cardinals)
• Ronny Cedeno (Pirates)
• Orlando Cabrera (Reds)

The NL Central has a rather weak crop of shortstops.  Manzella and Escobar are rookies, so it’s hard to say exactly how well they’ll do.  However, the Brewers traded a respectable J.J. Hardy so Escobar could take over at short.  That should tell you what Milwaukee is expecting out of Escobar.  Is he better than Theriot?  It’s hard to say, but I can just about guarantee you that Milwaukee would not trade Escobar for Theriot, even if the Cubs made up the salary difference. 

Comparing Brendan Ryan and Ryan Theriot is interesting.  Generally speaking, Ryan is considered a defensive specialist who is a sub-par hitter.  Theriot is considered an average (at best) defender and an average (at best) hitter. 

I don’t particularly like defensive statistics, but we need something to compare the two shortstops.  In this case, Ryan had a UZR150 rating of 13.8, while Theriot had an 8.3.  I don’t know how accurate those stats are, but I would agree that Ryan is the significantly better defensive shortstop.

At the plate, Ryan hit .292/.340/.400 with an OPS+ of 96.  Theriot hit .284/.343/.369 with an OPS+ of 83.  In all fairness, we’re only looking at one season and the season we are looking at (2009) was Ryan’s best offensive season.  If we look at their career numbers, we see that Ryan is a career .279/.333/.373 hitter with an 86 OPS+, while Theriot is a career .288/.356/.369 hitter with an 86 OPS+.  That’s fairly even.

As offensive players, there’s not much that seperates  Ryan and Theriot.  Defensively, Ryan is pretty clearly the superior shortstop.  Advantage:  Ryan.

The Pirates have Ronny Cedeno listed at the top of their depth chart for shortstops, but it very well could be that newly acquired Bobby Crosby will end up being their everyday shortstop.  It doesn’t take much analysis to come to the conclusion that Theriot is a better shortstop than Cedeno. 

Bobby Crosby is neither an elite defensive shortstop nor an above average (or even average) hitter.  Even if the Pirates go with Crosby over Cedeno, the Cubs will still have the better shortstop.

That leaves Orlando Cabrera.  For his career, Cabrera has been an exceptional defensive shortstop.  However, last year he had a UZR150 rating of -13.1.  Was that an aberration or has he lost his edge as a defender?  It’s hard to know for sure.  Cabrera is 35 years old and may be losing a step in the field.  Although we can’t be sure, I would still prefer Cabrera’s glove to Theriot’s.  However, I have to admit that it is not clear cut.

Last year, Cabrera hit .284/.316/.389 with an OPS + of 86 (split between the Angels & Twins).  For his career, Cabrera is a .275/.322/.398 hitter with an OPS+ of 86.  Offensively, I’d give the slight nod to Theriot.

So, Cabrera gets the slight nod on defense and Theriot gets the slight nod on offense.  It’s a wash.  If I was trying to figure out which guy I’d rather have in the future, Theriot, who is six years younger than Cabrera, would be my choice.  But since I’m only concerned with 2010 (at least for this article), who would I choose?  I’m not sure who the better shortstop is overall, but because Theriot has the slight edge offensively, and because the Cubs need all of the offensive help they can get, my choice would be Theriot.

So, with all due respect, I find myself disagreeing with Posnanski about who the best shortstop is in the NL Central.  I think Brendan Ryan is pretty clearly a better shortstop, and an argument could be made that Orlando Cabrera is better as well.  In addition, I would expect Alcides Escobar to be a better shortstop than Theriot in 2010.  But in Posnanski’s defense, Escobar is a rookie and doesn’t have a big league track record on which to base an opinion.

Which brings us back to baseball prediction systems.  Like any type of prediction system, the results you get are only as good as the information you put into the system.  If you rate players too highly (or not highly enough), the results are going to be flawed.   Plus, because it’s impossible to account for all of the variables that make up a baseball season, there’s no way that a baseball prediction system can consistently predict the outcome of a baseball season before the season even begins.