Hey Doc - b-ref.com Similarity Scores


Ghost sez,

BBRef comp lists generally make no sense to me.  They compare players based on counting stats accrued in a given amount of playing time (and age)...they tell you little about context, skill type or player family.  BP's list at least include those phenotypes and some other info that helps narrows things a bit (plus the counting stats are better-adjusted for context).

That said...one pattern I'm noticing with both the BBRef and BP lists...most of the players on those lists evaporated or at least transitioned down to much lower production levels when they hit 30-34.  Or became increasingly injury prone...either/or.

And the original source here, has Matty's back.  Bill James politely stated that b-ref.com similarity scores don't work for him, either.  This James comment is rather important to note, since b-ref.com sells its Similarity Scores as being James'.  Excellent judgment by SABRMatt here, in my opinion.

James made a remark, at BJOL, that was a show-stopper for me.  He said that he re-does Similarity Scores "every time he wants a new comp for a player," or somesuch.

This had been exactly my own feeling, that when you want a set of comps for Dustin Ackley, you need to set down the "irreducibly complex" similarities for Ackley as an individual.  (True, you have to be objective; you can't take a 12-win, 3.49 ERA pitcher and comp him to everybody with 12 or more wins and a 3.50 or lower ERA.)

The Similarity Scores at b-ref.com are attempting to slice kiwi fruit with a very dull axe.  They're good only for the vaguest possible generalizations.

James' attitude cemented my own conviction that you should (carefully) craft a new comps list for any given player.  Certainly, that approach is a given for a chess grandmaster.


=== Prince Fielder Downside ===

That said...

One thing I do like about b-ref.com's comps - you go to this link rat cheer and you can see the "Age 28 Until End Of Career" production for 10 fairly reasonable Prince comps.  (Average 8 more seasons at an average of 124 OPS+.)

You can scan the list of 10 players and decide which players are in fact reasonable comps ... essentially none. Juan Gone was a free swinger, and I ain't comp'ing lousy EYEs to Prince.  Are you comp'ing Richie Sexson to Edgar Martinez?

Right hand batters are problematic:  when they're 33 and slowing down, they have the platoon disadvantage 80% of the time.  Jim Rice is both items that are the opposite of Prince:  right handed and a free swinger...

Teixeira is an obvious comp, but he has only played 4 years since he was Prince's age now.  Those four years, he has slowed down noticeably -- that dovetails with the fact that these 10 players did not OPS+ 150 through their thirties.  :- )


There are three fat guys on the comp list, which is quite a coincidence.  Boog Powell, Kent Hrbek, and ... Greg Luzinski!  Now there's a RH comp! ... were cleanup hitters until 33, 33, and 32 respectively.  Huh.  

Cecil Fielder, not as good a player as Prince, had his last 40-homer season at age 32.  Mo Vaughn slugged .500, with 36 homers and 117 RBI, eighty walks, at age 32 and his K rate bloomed that year.

Would agree with Positive Paul that, surfing b-ref.com comps, you get the first impression that Prince has 6-7 years left as a cleanup hitter.


I don't believe that there are enough true comps for Prince to take much confidence.  David Ortiz, a huge lefty pitch-stalker with a lot of walks, had OPS+s of 154 and 137 the last two years.  At ages 34-35.

Two other guys who remind me of Prince:  Carlos Delgado and Willie Stargell -- huge, huge Take No Prisoners swings ... at age 36, Delgado had 36 homers and 117 RBI, whereas a flabby Stargell actually won the NL MVP at the age of 39.

If the Rangers signed Prince, I wouldn't hold out a lot of hope that he'd tail off anytime prior to age 33.  And it might be our bad luck that he'd be ripping along at age 35, as Ortiz is.

Honestly, whether Prince signs with the Nats or whoever, the b-ref.com list convinces me that age 33 will be the earliest year in which he fails to perform.  Everybody* similar to Prince hit great at age 32.

My $0.02,

Dr D


ghost's picture

Fielder should be fairly safe to continue being effective through age 32.  After he won't stop hitting homers either...he'll just probably start hitting .230 instead of .280.  That sort of thing happened to Frank Thomas, if you recall (he was a .330 hitter that dropped to .260).  That hurts his value but he'd still help a club for a few more years even in the downside scenario.


Decline vs Implosion
Agreed that bref comp scores are broad spectrum stabs at similarity.  That said, I like using the comp scores "through age X", which "tend" to at least group guys with similar arrival times and PT, (though injury laden players can get grouped strangely).
But, there's a significant difference between saying someone has 6-7 years left vs. saying a player has 3 years (or less) left as one of the best hitters in the game and then another 3 years as a pretty good bat.  The argument for acquiring Fielder tends to begin with the concept that he *IS* one of the top hitters in all of baseball and that his equals are few and far between.  BUT ... the question that should be asked is what is he going to be over the entire term of his contract.  There is only one team that can ever get Fielder's performance through age 27 ... Milwaukee.
Is he going to continue to be good?  Absolutely.  Is he going to continue being one of the top (5?) bats in all of baseball?  The odds of that are effectively zero.  But, nobody knows "precisely" when the decline will hit or how severe the FIRST plateau drop will be. 
The fact Ortiz had a great 2011 season doesn't undo the reality that after 4 straight seasons over .983 from age 28-31, he didn't break .900 the following three seasons.  Ortiz DID decline significantly.  His 2011 bounceback doesn't negate the decline.
Boog Powell posted 160 and 163 OPS+ seasons at age 27 and 28.  The next 4 years were: 138, 129, 126, 125.  Nobody really complains about 130 OPS+ guys.  But paying for 150 OPS+ and getting 130 OPS+ is not a good business decision. 
One trick I use because of the weakness of the bbref comp methodology is doing "one-off" comps.  Look at Fielder's list, figure out the guys that you think are the best comps and then see who THEY comp to.  Tex is one of his best comps, but hasn't finished his career.  So, go and get the Tex comps.  If handedness is critical, pick out the lefties from that list.
I see Will Clark and Fred McGriff.
Clark never had Fielder's power, (though he did hit 35 HRs at age 23?!).  Not a guy I'd say was a strong comp, but what happened to him anyway?  He had 153 and 150 OPS+ seasons at age 27 and 28, then a single season over 130 (his age 30 season), over the next 7 seasons.  He went from a 150 to a 130 hitter at age 29.
Crime Dog feels like a much better comp, (ignoring girth).  Great combo of power and patience.  Except for 1994, he was also perpetually healthy.  But after posting 150-167 OPS+ scores through age 30, at age 31 he swooned badly for 4 years: 119, 119, 106, 111.  He bounced back up to the 140s from age 35-37, (but since this is during the steroid era, one has to really hesitate at drawing conclusions about things for any player during that period).
Will Fielder be worthless for the next 6 years?  Absolutely not.  But there is a HUGE range between 160 and worthless.  I remember arguing against you about the wisdom of signing Teixeira to a long term before the Yankees snagged him, arguing that *ALL* players, (even great ones), decline sooner than most people accept. 
The error that *I* see is that most people are working too hard to find comps and not paying enough attention to the basic reality, (missing the forest for the trees).  Fielder will not decline because he is fat ... or patient ... or left-handed ... or a dog lover.  He will decline because he will get older and players, nearly all players, decline after age 27/28.
The chances are nearly 100% IMO that whoever gets Fielder for the next 6-7 years will be getting (mostly) a 125 OPS+ bat.  It is certainly possible that he could produce another 2 or 3 160+ seasons.  But, when you look throughout the history of baseball, (and make some allowances for the steroid era), the odds of him actually being a 160 hitter for a significant portion of his next contract are small.
Maybe the decline doesn't kick in until age 31.  Maybe it hits at 29.  Nobody knows the specific point in time it will happen.  But, it will happen.  And typically, the first drop in production hits sooner than most people think.


Wish I could have said it that well.
Fielder is definitely a good hitter for the next 6-7 years, barring injury.
The question is whether he is a $25M hitter for the next 6 to 7 years.
He will be for a while. That won't be for the entire 6 or 7 years.
It certainly won't be for 8-10, which is why Boras isn't getting that.  I think that's the real result of the Pujols' signing. GM's are saying, "You're going to be paying how much? When? For a 40 year old guy?"
I suppose you cold argue (fairly) that $25M won't really be $25M in 7 years (the inflationary effect)...and that's worth considering.
But, all the same, Fielder is a mahing terrific bonker for only part of his long term contract.  What you're reallly betting on, is how large that part is.
Give him 6 years, Z.


That is the M's operating premise, absolutely, that Fielder is going to be an MVP candidate for three years, and then simply a MOTO hitter for 3-4 years after that.  
That's the nature of 9-figure free agents.  Among big-TV-market teams, it's pretty much just Seattle, and Atlanta, that have refused to dance.
Teams underpay early and overpay late.  But if Fielder, at age 32, gets $24M for +35 to 40 runs -- a pedestrian cleanup hitter season -- that will be roughly market value even for the age-32 year in isolation.  The $/WAR will have increased a lot by then.
We don't get to buy 8 consecutive MVP seasons, from Fielder or anybody.  That's a given in the industry.

PositivePauly's picture

Looking back I think I overstated how strongly I trust BB-Ref's similarity scores. I agree that you really have to take them with a huge grain of salt, especially since they seem to sometimes ignore quite important elements as handedness.
But anything strongly over 900, adjusted for handedness, I still see as somewhat trustworthy. I was just trying to show an alternate "computer-generated" similarity score perspective :-)
Clearly the biggest concern with Prince going forward is how long he'll stay productive. You can try to find players that are similar, and IMHO, there certainly are a few, but I don't think it's unreasonable to think he'll be similarly productive for the next 5 years as he has been recently. Predicting injury, as Bill James said in his handbook a few years ago too, is really really tough. So, ultimately, it boils down to good scouting opinions and as trustworthy data as you can find.


James always says, the difference between knowledge and baloney, is that you can build on knowledge.  Baloney moves in circles.
BB-Ref's comps are often very enlightening, though in this case they might have missed the mark a bit.  In any case, they helped us organize our thinking.
I'll give you a nickel if you'll post at least once a day my friend :- )

Add comment

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd><p><br>
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.


  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.
  • Web page addresses and e-mail addresses turn into links automatically.