this is kind of mean but funny

The Yankees will outbid them for Lackey, so the Mets will sign Chone Figgins to play LF at 4/50M.

I don't understand what the point of this is. I grabbed every starting pitcher in the expansion era who, through age 30, had an ERA+ 124>x>110 (Lackey's +/- 7), with a K/BB of greater than 2, more than 85 wins, and more than 1200 IP. I deleted a couple of guys who actually aren't 30 yet, like Beckett. I get 42 names.



Provided by Baseball-Reference.com: View Play Index Tool Used
Generated 11/21/2009.

Nice list, Jeff. A lot of guys who were great in their 30s, too. "A lot of pitchers get hurt" isn't a newsflash.

So how many of those aren't reasonable comps for Lackey? Jenkins isn't. Jimmy Key has a slightly better ERA+ in almost the exact same innings while walking 3/4 of Lackey's total but striking out 2/3, and winning 1 more game while losing 2 less. By the criteria Marchman used, he every bit belongs on the list, he's a better comp than Fernandez. Maloney isn't. But there's easily 30 names here that are perfectly valid comps when looking at ERA+, wins and losses, IP, and K/BB. Why are Drabek, Morris, Fernandez, and McDowell more instructive than, say, Tanana/Perry/Carlton/Drysdale?

(EDIT) That's actually 38 names, I didn't notice it didn't re-rank after I deleted the few guys.

Some of these guys pitched a lot more innings. That's one big reason I don't see Carlton and Drysdale as great comps; they had a longer track record of pitching at a high level for full seasons. Lackey's thrown around 1500 innings; it might be instructive to throw out the guys over 1800 or 2000 innings.

Alternatively, you could look at ages 25-30, rather than career totals through 30. Of course, there would still be big differences in innings totals, especially across eras.

Here's the same table, with the differentials from Lackey:

Some of these guys pitched a lot more innings. That's one big reason I don't see Carlton and Drysdale as great comps; they had a longer track record of pitching at a high level for full seasons. Lackey's thrown around 1500 innings; it might be instructive to throw out the guys over 1800 or 2000 innings.

Except Marchman's point seems to be that Lackey's likely to tire out/get injured. I don't see how guys who threw considerably more innings but stayed healthy are anything but perfectly valid comps.

OK, I used ages 25-30, and since Lackey's ERA+ is 121 over this time, used the range of 114 to 128. When I sort by strikeouts, it sure seems that the guys who struck out more guys in their late 20s had more success in their 30s.


Provided by Baseball-Reference.com: View Play Index Tool Used
Generated 11/21/2009.

Also, Marchman mentions Smoltz could be put into the group. Smoltz is one of the highest differentials at 559 more IP. Only Carlton, Buerhle, Chance, Gooden, Tanana, Kaat, Eckersley, Viola, and Blue get thrown out if that's the standard. But, here's the list minus guys who are more than 285 IP away from Lackey:



Seems to me the best comps on that list are Key, Cone, Morris, Millwood, Garcia, El Sid, and then Nolan/Perry.

Lackey's injuries have been sore forearms that were babied by the team through spring training specifically to keep him able bodied down the stretch. They were quite minor and the team took the most conservative approach possible.

Wouldn't the way the angels deal with lackey tell you a lot about his medical situation? If they make a push for him I'd think they view his injuries as minor. They would know best.

One last one. The list from #10, with their ERA+ and IP from age 31 on (* is active):

So while the list overall in #13 (sorry for splitting stuff, but I don't like to test fate with BBCode's formatting of tables too much) looks a little grim, the guys I noted as the best through-30 comps are:

2 guys who remain excellent over many IP, from each extreme of the spectrum in Vinay's list sorted by Ks (Cone and Key)
1 guy who sucked in about 2 years IP (Morris)
3 guys who were slightly above average for starter ERA+, two done after few innings (Garcia, El Sid) and one still going strong (Millwood)

and then Gaylord Perry and Gary Nolan, who are the complete extremes.

Given we're talking about pitchers over the age of 31, that is not a bad ####### list of comps.

By the way, I somewhat like Vinay's list in #9 the best of the bigger lists wrt Lackey comps. That's a list of guys that I completely see as both good comps and as representing the range of what he has a chance to finish up as. The only one he doesn't realistically stand a chance of being at the end of his career is probably Carlton.

While I think these tables provide an interesting glimpse into possible career paths for Lackey, I'd be more interested in who some of the historians on this site see as strong physical parallels to Lackey. (In the not too distant future, we'll be able to run stuff comparisons in addition to tables like these thanks to Pitch F/x and then we'll really be able to pick out who comparable players are.)

You say historians, and while I agree that it's important to look at the physical side, I would say that with the changes in training, rehab, surgery techniques, etc. that even given close physical similarity I wouldn't look at a physical comp past 20 years or so. However much a pitcher's innate physical profile affects his ability to pitch, pitch for a long time, and do so well, the same physical profile now is able to do completely different things than its counterpart 30 years ago.

I'd be more interested in who some of the historians on this site see as strong physical parallels to Lackey.

Cy Young was a huge mouthbreather, so I think Lackey will be fine.

Ignoring the issue of just how appropriate the comps are (because I'm lazy), the list in #13 shows the other piece of risk too. It's not just the chance that Lackey will get injured. In that list (#13) there are a number of guys who gave you a reasonable number of innings but not, say, $15 M performance -- Colon, Millwood, Drabek, Nagy and Neagle. Those 5 guys plus the 7 total wipeouts is a lot of risk. Based purely on that list, you'd guesstimate a team would have about a 25% chance of Lackey continuing to pitch at his established levels. So, give or take:

25% -- established performance
25% -- average or worse performance
50% -- disaster

Marchman might have cherry-picked but the gist seems to be correct.

Note, that doesn't necessarily mean Lackey would be a bad signing at, say, 4/$60. At his peak, fangraphs had him at worth $20-23 (and that was in older dollars) and last year he was still worth $17.6. (I'm not a particularly big fan of these numbers but they at least give an independent valuation). So you could repeat the above table as something like

25% -- $70-80 M value
25% -- $40 M value
50% -- $10 M value

Still pretty ugly as that would give you an expected value of about $35 M, not a good expected return on a $60 M contract.

By the way, is there an easy way to take a P-I table and then look at the performance of those players going forward? I've been wanting that forever. Or did you have to calculate that from each player's record.

Matt Morris lost two mph off his fastball just from his age-28 season to his age-29 season, and then had shoulder surgery at the end of the year. Statistically he had a nice bounce-back at age 30, but his stuff did not come back with his numbers. He's not that useful a comp for a pitcher with a healthy shoulder.

While I still stand by my conviction that it is in the Mets' best interests to sign Lackey, I must confess that this statement:

Lackey's injuries have been sore forearms that were babied by the team through spring training specifically to keep him able bodied down the stretch. They were quite minor and the team took the most conservative approach possible.

read with the knowledge of how the Mets handle injuries, gave me pause.

DB

[19] I agree with your scenario-based approach to looking at the 5-year return on the possible $60 - $80 M investment into a 4-5 year guaranteed contract for Lackey.

On bb-ref, you can take a look at comps (by similarity score) for a given age, and then look at the performance of those players going forward. On looking at Lackey's comps, I'd say the percentages are more like:

15% -- established performance
55% -- average or worse performance
30% -- disaster

Using your weights, this would come out to a similar expected return of $37 M on his contract. It's definitely not worth it to the Mets on a 4- or 5-year basis. It's really only worth it for the Mets if you believe that Lackey will help them win in 2010 or 2011 while Wright, Beltran, Reyes, Santana, and Rodriguez remain at or near peak.

When you look at the age 31-35 performance of the comps, it is unimpressive. One, Bunning, the one Hall of Famer, significantly improved versus his pre age 30 seasons. Flanagan improved slightly (99 ERA+ to 104 ERA+). The rest arguably declined and delivered league-average performance or worse and ate (or in the case of Millwood will eat) fewer than 1000 IP.

Lackey's closest comps through age 30 have scores from 949-977. They are, in descending order, Kevin Millwood (ERA+ of 114 through age 30), Mike Flanagan (99), Bartolo Colon (121), Freddy Garcia (113), John Smiley (103), Matt Morris (116), Jim Bunning (118), Bill Gullickson (101), Brad Penny (104), and Pat Hentgen (113).

Five of the seven retired pitchers pitched through their age 35 seasons; Smiley pitched two years and Morris only three. (Penny's age 31 season was 173 IP at 94 ERA+. Millwood (754 IP at 100 ERA+) and Garcia (345 IP at 99 ERA+) just concluded their age 34 seasons.)

Based purely on that list, you'd guesstimate a team would have about a 25% chance of Lackey continuing to pitch at his established levels. So, give or take:

25% -- established performance
25% -- average or worse performance
50% -- disaster

Marchman might have cherry-picked but the gist seems to be correct.

And this is one reason why 1-year fliers, even though they don't solve the hole in your rotation long-term, are attractive.

On bb-ref, you can take a look at comps (by similarity score) for a given age,

Unless things have changed, those are not era-adjusted and as such, especially for a guy like Lackey, the comps are next to useless.

Oh, and Walt, you know better than that. There's virtually zero upside in nearly every major free-agent acquisition with regards to the salary they're making and any value they're reasonably likely to provide. Teams almost always pay at the highest level, and often overpay even that.

The only reason I am not more pessimistic about the Mets signing Lackey is that Lackey's K/9 IP rate of 7.2 is 3rd on the 11-player list of him and his age 30 comps, behind Millwood and Colon (625 IP at 96 ERA+).

I don't see why Marchman chose K/BB rates instead of K/9 rates, other than the convenience of using one rate instead of two (K/9 and BB/9). On the list shown in [3], there is a weak correlation between the K/BB and K/9 data. Lackey's K/9 rate of 7.2 puts him 9th of 38, with a range from 5.0 to 8.4 and a median of 6.4.

The K/9 rate shows where Marchman's comparison set for Lackey is less balanced. All 5 have a similar WHIP (Drabek-1.18, Morris-1.27, Lackey-1.31, Fernandez-1.28, McDowell-1.28). All 5 have a similar BB/9 (Drabek-2.32, Morris-2.47, Lackey-2.64, Fernandez-2.82, McDowell-2.90). However, Lackey's strikeout rate is nowhere near the others: Drabek's K/9 rate was 5.47, Matt Morris 6.44, Lackey 7.2, Alex Fernandez 6.40 and Jack McDowell 6.24.

In 2009, 78 MLB pitchers pitched 162+ innings. The median K/9 rate was about 7.0; 12 had rates of 5 or less. The median K/BB rate was 2.33, the same or worse than three of the four Marchman comps, and 33 of the 78 pitchers had K/BB rates worse than 2.16, i.e., what Marchman quoted as Jack McDowell's multiyear average.

I would suspect but cannot prove that the K/9 rate is the clearer predictor of performance decline than K/BB. (Of course, it is clearer to look at separate K/9 and BB/9 rates than K/BB.) The K/9 rate should drop as a pitcher ages and the BB/9 rate should rise but less drastically if the power/change of pace "skills" age quicker than the control "skills." Does the K/BB rate fluctuate from year to year over a typically pitcher's career? I don't have a good feel for it.

[24]:

On bb-ref, you can take a look at comps (by similarity score) for a given age,

Unless things have changed, those are not era-adjusted and as such, especially for a guy like Lackey, the comps are next to useless.

"not era-adjusted ... next to useless" ?!?

Did you even take a look at Lackey's comps? Five, including his closest, pitched in the 2000s. 7 of 10 pitched in the 1990s or 2000s. Hell, the 1st, 3rd, 4th and 5th closest comps are on your final list in [13]! 4 of 10!

By comparison, in your first list, 29 of the 38 pitchers played before the year 2000. Sure, you _tried_ to control for era using ERA+, but did that really control for all the things that have changed that affect the longevity of a pitcher? These include training regimens, nutrition, medical procedures (note Tommy John was on the list), use of video, five man rotations, etc.

In your final list in [13], 10 of 16 pitchers never pitched in the 2000s. 3 pitched in the 1960s. Gaylord Perry through a spitball. Again, does that list "control for the era" any better than Lackey's similarity score comps? 12 of these comps have a K/9 rate lower than Lackey, between 5.0 and 6.9. A different subset of 12 have a BB/9 rate higher than Lackey, from 2.7 to 3.4.

After looking at all this data, I think Millwood probably is a pretty good comp. While ERA+ trend and peak through age 30 are not exactly the same, the ERA+, WHIP, K/9 and BB/9 rates through age 30 are pretty close.

Lackey comps through age 30:

<br>
Sim Player          From To Years<br>
    John Lackey      2002-2009<br>
977 Kevin Millwood   1997-2005<br>
967 Mike Flanagan    1975-1982<br>
962 Bartolo Colon    1997-2003<br>
961 Freddy Garcia    1999-2005<br>
956 Matt Morris      1997-2005<br>
956 John Smiley      1986-1995<br>
951 * Jim Bunning    1955-1962<br>
950 Bill Gullickson  1979-1989<br>
949 Brad Penny       2000-2008<br>
949 Pat Hentgen      1991-1999<br>
<br>

Cy Young was a huge mouthbreather, so I think Lackey will be fine.

I'm a Lackey fan, and that made me laugh.

Based on no statistical evidence whatsoever, I'm generally of the opinion that players who have only just scratched the surface of greatness through their prime, whether because they only achieved greatness for a very brief time, or they only reached a level that could barely be considered great, are not good risks to maintain their performance through their decline phase. This may seem obvious, but I feel like every year there are teams that pay these types of players like superstars in the hopes that they will play at a level that they have scarcely ever achieved. This goes for both Holliday and Lackey. Both will make the Mets better in the short term. In the medium to long term, the Mets would be better off passing.

Did you even take a look at Lackey's comps? Five, including his closest, pitched in the 2000s. 7 of 10 pitched in the 1990s or 2000s. Hell, the 1st, 3rd, 4th and 5th closest comps are on your final list in [13]! 4 of 10!

By comparison, in your first list, 29 of the 38 pitchers played before the year 2000. Sure, you _tried_ to control for era using ERA+, but did that really control for all the things that have changed that affect the longevity of a pitcher? These include training regimens, nutrition, medical procedures (note Tommy John was on the list), use of video, five man rotations, etc.

In your final list in [13], 10 of 16 pitchers never pitched in the 2000s. 3 pitched in the 1960s. Gaylord Perry through a spitball. Again, does that list "control for the era" any better than Lackey's similarity score comps? 12 of these comps have a K/9 rate lower than Lackey, between 5.0 and 6.9. A different subset of 12 have a BB/9 rate higher than Lackey, from 2.7 to 3.4.

Well, I would argue the exact opposite. The fact that an inordinate number of comps on the list come from the last 10 years, when the whole point of similarity scores is that they are supposed to look at guys across all eras and say "These are the dudes a lot like that dude, and so looking at what these dudes did will be instructive to figuring out what this dude might do." (ignoring the HOF-worthiness discussion that actually led to SS), is just further proof that they aren't era-adjusted. I mean, we all know they aren't, but even if you didn't, if you looked at a list purporting to find comparables from the past and 7 of 10 guys were exact contemporaries or damn close, you'd say "These aren't era-adjusted."

Because what that kind of extreme overrepresentation of contemporaries shows you is that you're essentially seeing era differences overwhelm actual player differences. And that's pretty true. The difference between Tanana and Lackey's raw numbers is more due to the fact they pitched in different eras than it is due to their personal/talent differences. Which is, of course, why we adjust for era. The issue with SS not doing so is that you have huge era effects (the increase in strikeouts being a major one here, along with the decrease in IP per year) that shape the careers/numbers of every player that played in the era, and so if you don't adjust them out, you end up with the only guys having a chance at being similar being those who were contemporaries.

And so yes, 10 of 16 pitchers (when looking at all pitchers in the expansion era) not pitching in the 2000s *is* adjusted for era, especially moreso than similarity scores that are explicitly *not adjusted for era*. We can reasonably say that "pitched in the 2000s" covers a period of about 15-20 debut years. So of the ~50 year expansion era, ~15-20 of those would result in guys who pitched in the 2000s. This actually overstates the number because of the qualifications Marchman used (kind of.) For a guy right now to have pitched in the 2000s and yet be around long enough to compile those baselines, he'd have had to debut some time between 1984/1985 and 2003 or so, but those are the edges. The vast bulk of the representative population probably debuted between 1990 and 2001, inclusive. Anyway, 15-20 of 50 means a properly balanced for era list of 16 should have between 4 and 5.33 pitchers who pitched in the 2000s.

As for Gaylord Perry through a spitball

So?

12 of these comps have a K/9 rate lower than Lackey, between 5.0 and 6.9. A different subset of 12 have a BB/9 rate higher than Lackey, from 2.7 to 3.4.

What, are you asking for a list where exactly 8 are above and exactly 8 below in each and every category?

What, are you asking for a list where exactly 8 are above and exactly 8 below in each and every category?

No, what I want is a set of starting pitchers who are similar to Lackey through age 30. (I'd rather have a couple good comps than many lousy comps.)

By similar, I mean pitchers who have achieved similar results, had a similar workload in similar environments who have achieved their results through similar skills. ERA+ controls for similarity of rate results across different run scoring environments. IP controls for the quantity of those results and somewhat for workload. SO/BB tells you next to nothing about whether the set of pitchers has similar skills, i.e. high strikeout power pitcher, low walk control pitcher, ground ball or fly ball pitcher. Neither IP nor ERA+ addresses changes in training, medicine, days of rest, pitch counts, etc., and bullpen usage across eras.

The difference between Tanana and Lackey's raw numbers is more due to the fact they pitched in different eras than it is due to their personal/talent differences. Which is, of course, why we adjust for era.

No. The significant differences between Tanana and Lackey in assessing longevity are:
(a) Tanana's first full major league was at age 20 in 1974, but Lackey came up at age 23 in 2002.
(b) Tanana pitched 2,350 innings through age 30, but Lackey has pitched only 1,500 innings through age 30;
(c) Tanana pitched about 7.20 innings per start through age 30, Lackey only 6.44; and
(d) Tanana completed 116 games through age 30, Lackey only 14.

The issue with SS not doing so is that you have huge era effects (the increase in strikeouts being a major one here, along with the decrease in IP per year) that shape the careers/numbers of every player that played in the era, and so if you don't adjust them out, you end up with the only guys having a chance at being similar being those who were contemporaries.

It's not the use of ERA+ that shows that these two men are quite different from an actuarial perspective. Tanana is 35th on the career list for innings pitched (at 4,188 innings of 106 ERA+ pitching.) Does Lackey's IP history indicate that he is or will be the workhorse that Tanana was? That's not a bet I would take. (As you probably know, "similarity scores" are calculated to adjust for differences in innings pitched.) To your credit, your final list narrowed down the list of comps using IP.

As for Gaylord Perry through a spitball

So?

1. I should have written Perry "threw" a spitball. I was tired.
2. Written correctly, it goes to my point about similar skills.

Gaylord Perry is 6th on the career leaderboard for innings pitched at 5,350. The reputation and ability to throw a spitball undoubtedly contributed significantly to his longevity. Lackey doesn't throw a spitball. Therefore, they do not have similar skills. Therefore, Perry is a poor comp for Lackey.

Similar skills can be far better assessed through component statistics--such as strikeout rate, walk rate, and groundball rate--than through ERA+. The skills are the means to the end and pitchers with difference skills will obviously age differently, even starting from the same or similar levels of effectiveness as gauged by ERA+.

the guys I noted as the best through-30 comps are:

2 guys who remain excellent over many IP, from each extreme of the spectrum in Vinay's list sorted by Ks (Cone and Key)
1 guy who sucked in about 2 years IP (Morris)
3 guys who were slightly above average for starter ERA+, two done after few innings (Garcia, El Sid) and one still going strong (Millwood)

and then Gaylord Perry and Gary Nolan, who are the complete extremes.

David Cone's K/9 rate through age 30 was 8.4 and his BB/9 rate through age 30 was 3.4. For Jimmy Key, it was 5.0 and 2.1, respectively. Lackey is 7.2 and 2.6. To roughly adjust for era, Cone was more of (or more effective) a strikeout pitcher than Lackey than the raw numbers show, as shown by his more frequent appearances on annual leaderboards. Similarly, Key was a more effective control pitcher, i.e., at limiting walks.

As [20] noted, Morris sucked in 2 years because he was injured at age 29. So his pre age-30 health history makes him a poor comp for Lackey.

Sid Fernandez has similar K/9 and BB/9 rates through age 30 as his contemporary David Cone: 8.3 and 3.4. However, he was an extreme flyball pitcher, far more so than Lackey is, for example.

That brings you to Garcia (6.7 and 3.0) and Millwood (7.4 and 2.7), who, by the way, are the third and first most similar pitchers to Lackey through age 30, with sim scores of 961 and 977, respectively. Interesting. ;)

On sim scores, let me referee ...

1. There's little evidence that "similar players" are useful for projecting future performance. Systems like ZiPS and PECOTA (and maybe others now) make some use of sims (based on the last 3-4 years of performance, not career to date) in projecting future performance but they have a minimal effect. In short, sims are a fun toy, not a serious too. As such, quit getting worked up about them.

2. Sim scores don't adjust for era at all and, as a tool, are close to useless. (As a toy ...) Sometimes they come up with reasonable comps, sometimes they don't. P-I will almost certainly come up with a better set of comps. Still, for these sorts of questions, it's probably better to look at sims over a limited age range (and possibly limited era range) rather than whole career.

3. But P-I is far from perfect. If we really wanted to era-adjust, we'd probably have to look at Lackey's K-rate vs. league average and compare that to others (for example). You can't do that in P-I.

4. Both Jeff and Bob are in some sense correct about the effects of era adjustment. You don't really want a lot of older comps because so many of the underlying factors of the game have changed since then you should feel uncomfortable making the assumption that aging patterns weren't dramatically affected. That is, even if Gary Nolan (in context) was similar to Lackey through a given age, it's still a huge assumption to think that means Nolan's aging is indicative of Lackey's future.

The ideal would be to have players from the same era because whatever era effects there are will be zeroed out. The problems there of course are that you necessarily will have a small sample size from which to draw your comps and, being recent players, you will have little/no future performance information on the comps. In short, you're screwed both ways and the best you can do (within the context of people bloviating on the web) is try to balance the two in your argument.

5. If we make a mistake, it's in insisting the comps be too similar. 1-2 good comps, and nothing else, do you absolutely no good. If you fine-tune your comps enough, you'll end up with nobody. It's not at all clear why you'd expect (for example) a 2B with a career OPS+ of 114 through age 30 to age substantially differently than one with a 100 OPS+. You don't look at your tailored set of comps and say "the average future performance was X, my guy is going to be X." You look at a broader set of comps and say "these guys tended to decline by about 10% in OPS+ and 15% in playing time." There would be reasons to think the 114 OPS+ 2B won't see as big a decline in playing time (because he'll likely still be an average or better hitter for his position) but the pool of "2B who started from ages 28-30" is likely to be more useful than "the 2 guys who are a good match based on a combination of OPS+, HR rate, BB rate and K rate."

5a. Actually my preferred approach is similar to what Jeff did. Start with the broad set, pick up the general pattern. Narrow the focus to see if the pattern shifts substantially. Narrow again (or in a different way) and see if something different emerges. If you keep coming up with fairly similar results, you have more confidence.

6. Tanana is a lousy comp for ANYBODY except maybe Doc Gooden. Tanana was looking like a future HoFer before his arm came up lame at 24 and fell off at 25. He managed to come back as more of a finesse innings-eating lefty ... who oddly got a nice boost to his K/9 rate in his early 30s. Anyway, Tanana 19-23 had a 131 ERA+, 7.8 K/9 and 3.2 K/BB. Tanana 24-29 had a 100 ERA+ with 5 K/9 and a 2.0 K/BB. Completely different pitcher.

6a. Now, if you have a 29-year-old finesse lefty and you're curious as to how he might do from ages 30-33, then Tanana (ages 25-29) might be a perfectly good comp for that guy.

No. The significant differences between Tanana and Lackey in assessing longevity are:
(a) Tanana's first full major league was at age 20 in 1974, but Lackey came up at age 23 in 2002.
(b) Tanana pitched 2,350 innings through age 30, but Lackey has pitched only 1,500 innings through age 30;
(c) Tanana pitched about 7.20 innings per start through age 30, Lackey only 6.44; and
(d) Tanana completed 116 games through age 30, Lackey only 14.

Other than the fact that Tanana came up at age 20 vs. 23 for Lackey (and possibly even that), every single one of these goes exactly to my point. The biggest difference in the raw numbers between Lackey and Tanana is that Tanana pitched in a time where starters went deeper in games, got more decisions, and threw more complete games. Those factors, the extra innings worked, combine to create a huge differential in the raw numbers, where looking at the rate stats shows much more similar pitchers. The ERA+ is 2 points apart. The K/BB ratio is .19 apart. The K/9 ratios are 7.2/6.3 and the BB/9 ratios are 2.9/2.5 (Lackey/Tanana.)

Similar skills can be far better assessed through component statistics--such as strikeout rate, walk rate, and groundball rate--than through ERA+. The skills are the means to the end and pitchers with difference skills will obviously age differently, even starting from the same or similar levels of effectiveness as gauged by ERA+.

I don't at all disagree. I didn't pick ERA+, Marchman did. But I don't see how this applies to your point about era-adjustments. I'm saying that the components of similarity scores should be era-adjusted, you're saying that it's better that they're not because, from what I understand, you would rather deal with a list comprised of worse comps (or less comps) and not have to deal with era adjusting than have a bigger pool and better comps but have to do the admittedly-imperfect era adjusting. I don't think your stance is ridiculous or anything, but I do think it's inferior.

Interesting discussion. For a serious projection rather than something tossed off to make a snarky point of course you want to normalize to league rates. For a quick and dirty check, though, I think a search like the one I ran (RHP in the last 20 years with ERA+ marks within five points of Lackey's and K/BB rates above 2) does just fine. (Of course sub in whatever strikes you as more useful.) The list you get from that is Smoltz, Colon, Schilling, Drabek, Cone, Wood, Morris, Fernandez, McDowell, Millwood, Garcia, Nagy and Burnett. That sounds about right, and there's nothing about Lackey that makes me think he'll come in at the high end there. Does he have a great fastball that will age well like Schilling's? Not really. Is he especially durable? No, he's pitched an average of 170 innings the last two years. I wouldn't be totally shocked if he aged well, but in the absence of some compelling reason to do so I wouldn't want to be the one gambling Burnett money on him.

[33]

Jeff:

Thanks for a good discussion.

I'm not advocating for similarity scores. Their implementation on bb-ref is very nice because you can compare the 10 most "similar" players to your subject at age X, and then with one click you can see the totals for the same players from age X+1 through the end of their careers (or to the present for active players). I wouldn't dismiss any data source out of hand, as it may be useful in generating or refining one's hypothesis. For example, had you looked at the sim score data, you'd see that the IP to age 30 for the 10 comps ranged from 1377 IP -1644 IP, in other words, +/- 120 to 140 IP around Lackey's 1501. You may have approached your first IP constraint differently from merely setting IP > 1200.

I agree. ERA+ is well suited to assess results across different run-scoring environments.

However, I repeat, that to assess skills, not results, one needs to at component statistics, and then adjust those for the era in which a pitcher played, in order to get a sense how Lackey's skills compare to the pool of comps. (You could eyeball it using the annual leaderboards, or use P-I to generate the data for starting pitchers on a year-by-year basis and calculate the mean and standard deviations manually.)

I don't think Lackey will age as well as Cone or Key did, but they are not an unreasonable ceiling for your expectations of Lackey. Among comps, I still think Millwood represents the most likely trajectory for Lackey: solid but uneven #2 performance, declining over time, without much "ace" type upside.

FYI - Here's the Transaction Oracle post from Millwood's 12/2005 signing

Do you guys think the Mets should go after Lackey so they can be serious contenders before the Beltran window closes?

For the record, the top 10 ZiPS comps for Lackey are:

John Smiley
Frank Viola
Harvey Haddix
Bruce Hurst
Bobby Ojeda
Mickey Lolich
Billy Pierce
Jon Matlack
Mike Flanagan
Charlie Leibrandt

I'd have to pass on ... all of their 30s. Geez.

[36] That's not unreasonable, as I wrote in [22]. The big issue is, if the Mets commit $15M+ to Lackey for 2010, how much $ is left in the Wilpon budget to fill the other holes at corner OF, SP, 1B, and C?

For the record, the top 10 ZiPS comps for Lackey are:

Dan, do you have all LH and RH ZIPS comps or something?

Do you guys think the Mets should go after Lackey so they can be serious contenders before the Beltran window closes?

This is all, of course, just one fan's opinion, but I think the Mets should go after Lackey because it makes them a better team. I don't think Lackey alone makes the Mets serious contenders, but I do think that Lackey combined with a serious upgrade at one of the corner outfield positions (i.e. Matt Holliday) does. I think the Beltran window should be a factor in motivating the Mets front office to do all that it can to make the team into a contender for 2010.

DB

Have we given up on calling it the "Beltran-Reyes" window?

It depends on the money, of course, but at something like 6/120, which seems high to me, I'd rather have Holliday than Lackey at 5/80. The former seems much more likely to help in 2010 and 2011 without the fairly serious risk of punting 2012-2014.

I hate it the way that the rich teams in the National League keep trying to buy up all the AL stars. There oughta be a law.