Predicting How Pujols Might Age, Revisited

Earlier, I posted a small study concerning how Albert Pujols might age by examining similar players. That study used the most similar players lists available at the time: the current Baseball Reference lists for Pujols and the 1-year old 2010 Baseball Prospectus list of 10 most similar players. This is a follow up that uses the 2011 Baseball Prospectus list (generated after the 2010 season). This time I also use Baseball Prospectus’s WARP, instead the similar WAR, since I was already on their site, and I think WARP is probably good enough, as we are looking at shapes of data.

As I said before, Pujols is awesome and will hopefully be a productive Cardinal for a long time. I’m not intending him any disrespect with this post. I know new study is not that statistically sophisticated and uses small sample sizes—I offer this for thought and discussion, not as certainty.

Part 1: The Most Similar Players

Baseball Prospectus uses factors such as a player’s statistics, position, age, and body type to generate similarity scores.

The 10 most similar players, listed from most to least, are these: Lance Berkman, Vladimir Guerrero, Todd Helton, Barry Bonds, Frank Robinson, Wade Boggs, Frank Thomas, Jason Giambi, Jeff Bagwell, and Chipper Jones.

One problem with this list is that few players are in reality very similar to Pujols. A similarity score of 70 is high, and a score over 50 means a player is substantially comparable. Only one player, Lance Berkman, scored over 70. The next two highest are Vladimir Guerrero at 59 and Todd Helton at 58. The other scores ranged from 42-49. By comparison, the other active players in my study have lists of similar players mostly in the 70s or higher. None had a players in the fifties or lower.

I decided to omit Barry Bonds and Jason Giambi from the study due the possibility of their PED use tainting the data. This leaves me with an admittedly small sample, but nonetheless this is a sample of superstars who rake. For active players, I do not count statistics past 2010. I also will not mention the first month of anyone’s performance this year in my discussion.

Part 2: What can we expect from the first 4 years of the contract?

The question is how much Albert Pujols might be expected to produce after his current contract expires. When discussing this, people often say that if he signs a contract of 7 or more years, of course his production will likely be diminished towards the end of the contract. But if you sign him you are paying for the earlier years of the contract when Pujols can be expected to continue his established high level of performance.   

Since Pujols was 30 last season, I established his level of performance  by averaging his WARP for his age 27-30 seasons. I then did the same for his similar players. From ages 32-35—what will be the first four years of Pujols’s contract, the players averaged a per season WARP value of slightly over 60% of their established value. This is pretty disappointing. If Pujols follows this same trend, he will average a WARP of 6.28, meaning he will be worth a bit more than 6 wins above a replacement player. This value is lower than that of any season of his career (his lowest was a WARP of 6.4 in 2002). Now, 60% of Albert Pujols is quite good--to give you an idea of what that looks like, in 2002, he had a slash line of .314/.394/.561. However, that was in a higher scoring era than today, so the numbers would look lower today, as WARP is normed. But it would still be good.

But he still would be producing only 60% of what he is producing now, and these would likely be the most productive years of his contract. From age 36 on, he can be expected to decline, as all humans do. As my previous study noted, few players can avoid serious decline at these ages.

Part 3: Another approach: individual season performance

One problem with aggregating data is it can mask individual differences. If some exceptional players beat the effects of time, the failings of others can mask it. So I tried other ways to look at this. One question is, how many more seasons at above his established level can we expect from Pujols?

The players in my study have played a combined 54 seasons while aged 32 or older. In six of those seasons these players managed to equal or exceed their established level of production—an average of less than once per player. One player reached his established level twice. Three players never hit at their established level again.

Of course, many of these players could perform at a slightly lower level and be quite productive. So I examined how many times these players reached or exceeded 75% of their established level of production. The 8 players collectively reached this level only 10 times. One player reached it three times, and two players never produced at even 75% of their established level again. In short, producing at a level of 75% of their established level was a difficult feat for these players.

Part 4: Age 32-36 performance, by player  

Even taken individually, the players do not present an encouraging picture. I’ll discuss them in order of similarity.

Two players are not yet old enough to discuss from ages 32-36.

During the ages of 32-34, Big Puma has produced at a level equal to almost 95% of his established value. This is a bit misleading, as his established value was driven down by a poor age 29 season (and Pujols’s established value is not driven down by a poor season). This makes it easier for Berkman to reach his established level. Also, since he only has three years of data at age 32 or older, his mammoth season at age 32 completely distorts the data. In contrast to the retention of his value, the trend of his WARP scores during the three years was quite troubling: 8.2, 4.9, 2.0. So he is not really that encouraging of a comp. However, so far he sure has been fun to watch this year!

During the ages of 32-35, Guerrero has produced at a level slightly less than 50% of his established level.

The following players will have their age 32-36 seasonal production examined.

Todd Helton produced at a level equal to 28% of his established level.

Mr. Robinson produced at a level equal to 69% of his established level.

Wade Boggs produced at 37% of his established level.

Frank Thomas produced at 50% of his established level.

Jeff Bagwell produced at a level equal to 74% of his established level.

Chipper Jones, the least comparable player to Pujols, also is the best case scenario. He produced at a level equal to almost 88% of his established value.  He had two seasons that exceeded his established level, and another one that was at least 75% of his established level.

Jones, Bagwell, and to a lesser extent Robinson, did fairly well at retaining their value. Still during the 15 seasons they were aged 32-36, they combined for only 4 seasons that exceed their individually established levels of production, and only 6 seasons (40% of the total) reached a level of at least 75% of their established level. Half of those were from Jones.

Part 5: Conclusion

Based on this study of superstars most similar to Pujols, it is not realistic to expect many more seasons of production out of Pujols equal to his established level. The study suggests we can expect one more. If he gets three more, that will be better than anyone on the list. In fact, the study suggests that one more season at or above his established level and two more between 75% and 99% of that level would be beating the odds.

If I had to guess, I’d expect him to beat the 60% average over the first 4 years of a deal and produce at a level equal to 75% to 80% of his established level. But players like Helton and Guerrero provide a warning of the risk that even a superstar can quickly lose his value between the ages of 32 and 36. And this whole study is a sobering reminder of how quickly time erodes the skills of even elite talents.

Part 6: Extras

In the discussion of my previous study, someone asked about the production of players before reaching age 27.

So in response, for this pool of players I looked at how players’ age 24, 25, and 26 seasons compared to the "established level of production" in my study (from ages 27-30). The age 24 seasons were at 75%, age 25 seasons at 95%, and the age 26 seasons at 117%. For Pujols, the numbers were 88%, 79%, an 86% respectively. As a group, the players in this study peaked much earlier than Pujols did. I do not know what implications this has for his aging.