The older players whose careers fell off the charts due to injury are poor comparables; because of preventative care, surgery, and rehabilitation we are in a different world now. If those players had been born in 1980 their career trajectories would have been entirely different.

Ergo, A-rod's chance of breaking Bonds' record are substantially higher than 30%.

RustyJay

That's crazy talk, Nate. PECOTA has to break down for players of A-Rod's stature, because it can't account for his extraordinary motivation to break the record, or for the fact that he'll be able to get deals even if he does break down (see Ken Griffey Jr).

I hope you're right - I'm a Red Sox fan - but you're not.

Eh, I'm just cranky because you told me to take Mickey Rourke in my Oscar pool.

Ooh! Is LDL back as a regular feature, Nate?

The election is over, the inauguration is over. More BP, less 538 Nate. Please?

(a) Nate, didn't you do a career HR projection for A-Rod a few years ago (was it on BP, ESPN, or SI)? Can you compare the actual HR producdtion to your projection then?

(b) Also here's an interesting PECOTA exercise, which I suggested on last week's article by Steve Goldman, in response to a reader who wanted to look at "average" players:

Here's another exercise that could be done that might yield similar results to what you're interested in: which player at each position shows up most frequently as a PECOTA comparable? Presumably those individual should be the most "typical" or average at their positions historically (in Nate's PECOTA database).

Please consider this.

Give us a 'Comparable All Stars, 2009' (maybe top 3 to 5 at each position?)." Presumably there would be a bit of a longevity bias -- longer-career "average players" have a greater chance to become comparable all-stars. But mediocre longevity deserves recognition!!!

Thanks for descending from your cloud to give us this, Mr. Big Shot.

Just kidding. But seriously, I hope that you have time for LDL more often - I love your work here.

"Rodriguez already has all the money that he'll need for life"

I hate this argument. You don't know what he wants to do with that money. Maybe he wants to start a baseball league in Puerto Rico. Maybe he wants to buy the Florida Marlins. Maybe he wants to run for President.

Can PECOTA really create meaningful comparisons for atypical players like A-Rod? There aren't really a ton of middle infielders that have hit 50 HR and 20 SB in a season. Sammy Sosa didn't turn into a decent player until 1996 or so, while A-Rod started producing pretty much out of the gate as a rookie. Nor did Sosa immediately get the plate discipline to walk a whole bunch.

I can't believe we're only a few years away from ARod's somber, strange, and soulless quest to break Barry Bond's lifetime home-run record.

Should be a fun!

One more small factor in A-Rod breaking Bonds's record: What if Bonds gets more HRs himself? I'd bet he's going to beat his rap and he can still DH and pinch hit. Will anyone sign him?

Ah, PECOTA, ever the pessimist. I'll gladly take the over on 730. Hell, I'd take the over on 800.

The PED effect is the variable that can't be judged here. I know BP is sacred 'put our fingers in our ears/heads in the sand' territory regarding PEDs, but the reality is it is confirmed that A-Rod attempted to enhance his performance with PEDs and the circumstantial evidence is that this continued beyond the years to which he has admitted. How can this be figured in? It can't, of course. We don't know the positive effects of PEDs; we don't know their injury recovery/avoidance effects; we don't know precisely when he took them in the past; and we don't even know if he might continue to take them given, of course, that he is a mini-corporation who can easily afford his own medical dept to keep him ahead of the testing curve (one could argue that his incentive to do PEDs will be particularly strong this year, in order to show previous years' performances were not PED-impacted). But one thing we do know: it's a variable for which it is difficult to take account.

It is, of course, easier to live in la-la land and say this is a non-issue. But since this is a Nate Silver piece I'll be a bit political and steal a phrase from the last campaign: i.e., I'm all for joining the reality-based community. This variable has to be recognized even if it's effects can't be measured and makes Pecota's projections all the more uncertain.

Im always surprised when Bill James' formula is not used or at least referenced when projecting stats. I do not know if there is any research into the accuracy of it but the method is better than averaging 20 top comparables in my opinion. There is a free program you could once download which I currently can not find but the formula is available all over the net as is this nice little tool on ESPN:

http://sports.espn.go.com/mlb/stats/assessments

As others have commented, PECOTA seems to be less accurate projecting careers of extreme players, including players like A-Rod, so good at playing baseball that he is among the very few best ever to have played the game. It's tough choosing 20 comparable players to A-Rod because there aren't twenty other position players as good as A-Rod at his age. Picking comparables means minimizing ways that other players are inferior.

That said, some of the players listed as A-Rod's best comparables aren't reasonable choices because they're slow-moving sluggers prone to swift decline, not excellent shortstops who moved to third base to make room for another HOF-caliber player. Greg Luzinski is, to me, the most laughable comp: he had earned -132 FRAA over a career at left field and first base by age 29, and he played his last four years as a DH. Saying that a 32-year-old Greg Luzinski is the twelfth-best comparable to a 32-year-old A-Rod is, to my mind, absurd. Yes, Luzinski played just one more year and retired after a disappointing season at age 33. But he'd only played 1.2 games on defense in the last four years before retiring. That's not comparable to what A-Rod has done at third base his past few years.

There are other among A-Rod's comps who were declining slow-moving sluggers. Albert Belle fits into this group: while his career ended swiftly, lots of slow-moving corner outfielders and corner infielders see their careers end swiftly in their early thirties. Jimmie Foxx, mentioned in the article, fits that mold as well.

The trouble with A-Rod is that his defensive comparables were lesser hitters. Still, there are four of twenty comparable players who were reasonably similar to A-Rod in the field through their careers: Brett, Grich, Sandberg and DeCinces. (Caminiti was clearly a lesser fielder.) Of those four, Sandberg oddly retired after a bad start in 1994 at age 34 before returning for two years at age 36, Grich and DeCinces had reasonable decline curves, and Brett was still a 7.1 WARP player at age 37 and a 19-HR hitter at age 40, when he retired. It's tough calling Sandberg's retirement choices "comparable," but he, DeCinces and Grich all ultimately retired at ages 36-37 because they needed to do so. Brett, probably the hitter of this group most like A-Rod, was still an MLB-caliber player when he chose to retire at age 40, and he was hitting home runs at better than half his peak rate from his prime. This article has A-Rod hitting just eight HR at age 40: discounting the lesser fielders (Luzinski, Belle) and the lesser hitters (Sandberg, Grich, DeCinces), and maybe discounting the known juicers (Caminiti), A-Rod's odds look better.

I'd caution, though, after all of this writing intended to bolster A-Rod's case, there's one thing that still troubles me. If you check A-Rod's best comparables at Baseball Reference, there's one from the top ten who's better than any BP comp in many ways: Rogers Hornsby. Both Hornsby and A-Rod were probably the second-best hitters of their era, each surpassed only by a corner outfielder who defied the standards of their times.

That said, Hornsby had just one more MVP-caliber season left at A-Rod's age, his 14.9 WARP 1929 season at age 33. He injured his foot badly in 1930, ruining the season. He came back for a 7.6 WARP 1931, but then he accrued just 2.8 WARP and just six home runs over the next seven years before retiring. His last MLB-caliber year was at age 35, just one year earlier than Grich, DeCinces and Sandberg all found themselves nearing retirement.

While I've questioned the validity of PECOTA comps for a player as good as A-Rod, maybe the answer of "probably not" remains valid for this question all the same.

Even Griffey, with his diminished skillset, has not been projected to hit less than 20 home runs. I don't see how A-Rod can be projected to hit less than 20 from ages 36-43. Just the sheer cost of his contract means he could miss a season from rehab and still be allowed to play out the rest of it. People just don't retire from injuries as much as they used to.

I was struck by A-Rod's comps when I read the annual. Caminiti? Yeesh.

Great article Nate.

Saw this on ESPN.com this morning and reading the comments from the people the absorb everything Joe Morgan and every other terrible analyst says was probably the highlight of my day (as sad as that is).

Anyone with insider access that wants a laugh should check it out.

What's the steepest part of a pitcher's age curve?

I don't know if PECOTA deals with this effectively or not, but my socks don't match.

Do we have enough historical data to back test Pecota projections of first ballot hall of famers? I'd like to know if Pecota would have forcast those extraordinary careers. For example, what would Pecota have forecast for the Babe's last ten years?

Baseball is now a mature enterprise. That means that players, trainers, coaches are all more tightly bunched at the top of the learning curve. There should be much less variance between all of those individuals. It also means that prediction systems like Pecota will become more and more accurate. (A hundred years from now we won't even have to actually play the games!)

It seems that it would be useful to compare the average rate at which players have declined over the last 25 years with the rate at which they declined 50, 75, and 100 years ago. That would give us some idea about how valuable those old comparables are for predicting a modern player's last few years. The guess here is that there will be a significant difference in the rate of decline.

One hundred years ago the incentive to play late into a career was not nearly as great as it is today. Players were not making the kind of money they make today-even adjusted for inflation. The travel was much more brutal; and probably took a greater toll on a player's home life. That's without even getting into the medical, coaching, diet and training issues.

PECOTA's home run total for the rest of his career seems ridiculously low. Will there be a downturn, sure, that's very likely, but come on. He's suddenly only going to hit 8 home runs (and then 4!) at the end of his Yankee contract? That doesn't look like it's predicting a natural decline... considering A-Rod's ability, that's predicting a complete and total implosion where he suddenly loses all of his strength. Unless PECOTA can see the future and knows he's going to come down with a debilitating disease that saps all of his strength (or is predicting an injury 9 years from now, which seems quite out there), I don't see any chance of that.

I also don't see this "home run decline" over the past two years to be legit. In 07, he had 54 less plate appearances in the second half, and he also was having his best season ever. If anything is the anomaly, it's 30 home runs in the first half, not the 24 in the second half. In 08, he missed time with the injury and still hit 35 home runs...while he certainly wasn't on pace to repeat his 2007 season, I'd hardly call that a huge dropoff, especially considering in 06 he hit 35 home runs in a full season (and he hit 36 in 2004 for that matter). It's not impossible it was due to a decline, but that certainly wouldn't be my first explanation, not considering the injury and the fact that he's not consistently hitting 50+ home runs every single year.

I, for one, hope he doesn't even crack 700.

The list of ARod comparables show why steroids make PECOTA a HOUSE OF CARDS.

Since SOSA is Arod's closest comp, do we have to assume that Arod needs to continue taking steroids to match Sosa'a numbers?

Or do we assume that Sammy had the greater benefit of steroids up to this age in his career and his steroid-inflated numbers really aren't a comp for Arod (if you discount any benefit for arod thus far from them).

The point is that all these comps are USELESS unless you know how the numbers were generated with respect to steroids.

Is a 'clean' Dave Winfield or a 'dirty' Sammy Sosa or Ken Caminiti a better comp? Simply comparing numbers is PURE FOLLY.