THE BOOK--Playing The Percentages In Baseball

There may be no direct value from having Superman behind the plate, but there would certainly be indirect value - pitchers could abandon the stretch and always pitch from the wind-up being the main one. 

I think we’re on shaky ground if we try to make an argument that completely stopping the running game has no value.  It might not be an effect that can easily be measured, but that’s a lot different from claiming its not there.

I’m not sure I understand the discouragement-of baserunners-has-no-value argument here. If there is an average number (or if you prefer, replacement level number) of runners who would advance against your team via the stolen base, but if instead of an average (or replacemtn level) catcher imagine your team has a scary-good catcher whose reputation causes fewer than the average (or fewer than the replacement), number of runners to advance.  Isn’t that a calculable value for the team, using the linear weight value of a successful stolen base multiplied by the number of stolen bases below average (or below replacement)?

I did a couple articles at FanGraphs last year on catcher’s defense, and I remember Carter being near the top of the list, much as you describe. Bench in his first few years was also dominant, but the second half of his career not as good as he had been.

If Superman was catching an no one attempted, that does not make him average. Compare the actual SB & CS allowed to the expected totals, given the playing time, pitchers and runners, for example. Let’s say expected was 60 SB, 25 CS

With the approximate numbers you gave, -0.3 runs for a SB not allowed, +0.6 for a CS.

           SB   CS   SBr  CSr  Tot
Bench    -250    0  +75     0  +75
Carter      0  160    0   +96  +96
Superman  -60   25  +18   +15  +33

Being Superman behind the plate so that no one runs, yields the highest value for a catcher IF AND ONLY IF teams stole bases somewhat optimally.  If runners attempt to steal bases sub-optimally, i.e. they run too much, then a catcher who prevents runners from ever running are doing them a favor.

It is as simple as that. There is no magic in Tango’s statement.

As long as base runners are running even a little too much, a great catcher should actually back off a little in order to encourage runners to run a little.

Now, we maybe assigning wrong values to a SB and CS, because of the hit and run situation and because we don’t quite know the value of a runner in motion when a ball is in play or the value of a SB attempt on the batter, but that is a different story.

The value of a catcher’s arm is simply the value of the successful SB, plus the (negative) value of the CS, plus the value of a runner on motion when the ball is put in play, plus any effect a SB attempt has on batters. Nothing more and nothing less.

Now, depending upon whether the average team runs optimally or not (more or less) against each and every catcher, given those values, determines the value of the catcher against whom no one runs, as compared to an average catcher or any other catcher.

There is no controversy here.

Back to the article - I don’t get it.  First of all, I don’t really get the article, but without knowing all of Rally’s adjustments, I can’t comment on it.  First of all, what is a “catcher adjustment” beyond a positional adjustment?

Of course catchers will lag behind all other positions because they get less playing time.  If you want to adjust for that, such that a catcher who plays 130 games is equivalent to another player who plays 150 games, that is fine by me.  But I don’t really see any reason to do that.  WAR is WAR.  If a player doesn’t play, whether it is his “fault” or not, he creates no value.  Why not adjust for the fact that a starting pitchers only can pitch every 5th day?

So if you don’t make a 130 game catcher equivalent to a 150 game position player, then catchers will simply not have as many career WAR as other players. So what?  Again, if you don’t like that, then do the adjustment.  If that is what Rally’s “catcher adjustment is,” again, that is fine by me.  Is it enough?  I don’t know.  I don’t know what it is.  It should be an amount equal to the difference between how many games an average starting catcher plays per year compared to an average position player other than a catcher.

As far the positional adjustment being not enough, I don’t know what Rally uses, but for a catcher, since we have no idea how that compares to other positions, you should basically use the difference between an average batter and an average catcher as batter. Does Rally use that?

Finally, as far as replacement level.  I am not a big believer in using different replacement levels for different positions, although I suppose you can.  If truly somehow it is harder to find a -20 catcher for free than a -20 LF or CF or 2B, then go ahead and use whatever number you think works. 

Now, if after all those adjustments, the best catchers STILL come out lagging behind other positions historically, it must mean that there simply have not been as much of a spread in talent at the catching position, at least on the plus side, as other positions, which is always possible.

Right, I’m taking the position that runners are running sub-optimally, such that the overall success rate is barely above the breakeven point, when it should be comfortably above the breakeven point on average.

As MGL said very well, if all runners ran optimally, then Superman would be valuable.

***

Brian: don’t use .3/.6 . It’s closer to .2/.45

Tango, the more I think about the value of the runner on motion on a batted ball, the more I think that the overall value of the SB and CS is closer to .3 and .6 than to .2 and .45…

Tango, those were for illustration, I pulled the values out of my head where they probably had resided since the publication of ‘The Hidden Game’. It does not mean those are the values I actually use in my code, I did some lookups on linear weights for those.

I’m pretty sure what the article calls “a generic adjustment for Catchers” is just catcher defense (i.e. rating for SB/CS and PB/WP like Tango is talking about in the post), which is not really a generic adjustment for catchers, but another component of defensive evaluation.  The generic adjustment for catchers is just the position adjustment.  From the context, it would appear that the article is either misunderstanding or mis-communicating what Rally’s Catch column means. 

The author is also adding TZ into catcher ratings, when the TZ numbers are all from other positions.  That drops 10 runs off of the figure he uses for Bench’s catcher defense, just by unknowingly throwing his poor range at other positions into the mix.  So the difference is actually smaller than it is given in that article.  Given that Carter played a few hundred more games at catcher than Bench and was a damn good catcher, I agree that I don’t see what is wrong with Carter coming out ahead of Bench on defense.

Sean had an article on THT a while back looking at replacement level, and it looked at the offensive projections for the nth best player (including minor leaguers-n being however deep you would have to go to get to replacement level) at each position or something like that to find a replacement level for each position.  So once you get past all the starters and backups at the position in the Majors, and maybe controlled prospects in the minors who aren’t going to be available, however far that is down the list, you get a replacement offensive level for each position.  I think that’s where Sean’s catcher adjustment came from; the replacement catcher was something like 10 runs worse on offense than the overall replacement level, so the position adjustment was set to make up the difference.  That’s a bit different from the difference in offense between an average catcher and an average hitter since it is comparing replacement level instead of average, but it’s still based on the offensive difference.  This is all from memory, though, so take that explanation with a grain of salt.

"Sean had an article on THT a while back looking at replacement level, and it looked at the offensive projections for the nth best player (including minor leaguers-n being however deep you would have to go to get to replacement level) at each position or something like that to find a replacement level for each position.  So once you get past all the starters and backups at the position in the Majors, and maybe controlled prospects in the minors who aren’t going to be available, however far that is down the list, you get a replacement offensive level for each position.  I think that’s where Sean’s catcher adjustment came from..”

That’s fine, although that makes an average catcher pretty darn expensive, does it not?  Another 4.5 mm?  So an average catcher is 3 wins above replacement and gets paid 14 mm?  I don’t think so!

Anyway, that is nothing to do with the defensive adjustment. Defensive adjustment and replacement level are two completely different things.

So if Rally is NOT adjusting for the fact that catchers only play 130 games a year (which is correct I think) why is the author surprised that the best catchers are far down the list in career WAR?  Did he look at career WAR per game?  If he does, I suppose that all positions will be roughly equally represented at all levels.

On the other hand, if Rally’s replacement level is so low for catchers (or is it high?) than the best catchers should have a lot MORE WAR.  OK, maybe Rally’s replacement level for catchers is HIGHER than other positions.  Kinkaid, you say 10 runs worse - I think you mean 10 runs better. IOW, a replacement catcher is around 10 runs worse on offense than an average catcher whereas a replacement level at other positions is around 20 runs worse. Thant makes more sense.  That is because even average catchers are such poor hitters, that by the time you get to replacement level, they can’t be THAT bad in hitting…

I never understood the issue with catcher being less well represented than other positions on top-player lists like this.  Like MGL said, the lower playing time is a real thing that cuts into a player’s value.  If you have 130 games of a stud catcher and 32 games of a backup, or you just have 130 games of a replacement level catcher and 32 games of the same backup, the difference is just the 130 games of production over replacement.  It’s not the 130 games prorated to 150 games or however many you want to consider full-time for other positions.  Additionally, I would expect that catchers have more difficulty at maintaining long, productive careers at catcher, so there could also be the problem of less playing time by having shorter careers.  If you have a shorter effective career, that’s less career value you are able to provide.

So I agree with MGL that because of playing time issues, catchers probably do lag behind other positions in how much value they can provide.  On top of that, because of those playing time issues, teams probably tend to put great hitters who can play both catcher and non-catcher positions at non-catcher positions to keep their bat in the lineup more and to prolong their careers.  You see this a lot with good-hitting catchers getting moved to other positions at some point in their careers.  Also, because of the grind of the position, I would think that a lot of players who can play catcher and another position will choose the other position on their own.  I expect this happens quite a bit at the amateur level so that many of the players who could be very good catchers but have just as good a shot at going pro at another position never play catcher.  This would create a selection bias, both from teams/organizations keeping their great hitters from the deteriorating qualities of catcher and from players choosing to avoid those same qualities on their own, that leads to fewer great players playing catcher than other positions.

That’s fine, although that makes an average catcher pretty darn expensive, does it not?  Another 4.5 mm?  So an average catcher is 3 wins above replacement and gets paid 14 mm?  I don’t think so!

If the average hitting catcher hits just as well as the overall average hitter, yes.  Otherwise, no.  An average catcher is however many runs above or below average on offense (on average, this will be negative for catchers), plus 10 runs (prorated to playing time, so probably less than that for a catcher, since they are playing fewer games), plus defense.

Replacement level was just the baseline Sean used to derive the position adjustment (if I am remembering correctly).  It is the same as looking at how many runs worse the average catcher hits than the overall average, except instead of comparing to 0 runs, you are comparing to -20.  You said that you should take the offensive difference between the average catcher and the average hitter to get the position adjustment, so that if the average catcher is -10 (ten runs below average), then the position adjustment is +10.  You basically assume that average catcher overall = average player overall (over the same amount of playing time, anyway), so if the average catcher is -10 on offense, he must be +10 on defense to make him average overall.

That’s exactly what the article was doing, but instead of seeing that the average catcher was -10 on offense compared to a baseline of 0, it was seeing that the replacement catcher was -30 on offense compared to a baseline of -20.  Instead of assuming an average catcher is equally valuable to an average player, you assume that a replacement level catcher is equally valuable to a replacement level player.  You know a replacement level player is -20, so if the replacement level offensive level for catcher is -10, then he must be +10 on defense to make him replacement level overall.

Obviously, since this derivation assumes that replacement level for catcher is the same as replacement level for other positions, that means that replacement level is not lower for catchers.  It’s the same.

"You know a replacement level player is -20, so if the replacement level offensive level for catcher is -10, then he must be +10 on defense to make him replacement level overall.”

should be “...if the replacement offensive level for a catcher is -30, then he must be +10 on defense to make him replacement level overall.”

By the way, the -30 is below the average hitter overall, just like Sean would present his offensive projections as r150 figures, not below the average catcher.

On baserunning, I suggested in another recent catcher thread that catcher fielding metrics should try to incorporate the impact of runners going on the pitch.  My guess is that with Bench behind the plate, the Reds gained a fair number of GDPs, and that runners went 1st-3rd and 1st-home less frequently (and the same would be true even more dramatically for Superman).  And this was a bigger factor in the 70s than it would be today, as runners went much more often.  So if we’re going to debit Bench in a sense for the poor success rate of opposing runners, we really need to credit him with the benefits of holding runners at 1B.  I doubt this could close the entire gap with Carter, but it might erase half of more of it.

* *

On catcher position adjustment, I agree that this author overstates his case.  But I think Tango’s research has shown that catchers have a “hitting penalty” of 9-10 runs, that presumably results from the wear and tear of playing the position.  That’s different from their defensive value.  We can’t just compare FT catchers’ offense to that of backup catchers, because backups aren’t playing 130 games and so may not be paying the full catcher hitter penalty.  We have to compare catchers’ to what a replacement catcher does when forced to play 130 games.  It may be that WAR doesn’t fully capture that.

Brian (#3), Superman should be -60/-25 - nobody is running on him, so he’ll catch 25 fewer runners than average, not 25 more.  Using Tango’s figures of .2 and -.45, that puts him within a run of being exactly average.

Brian,

In #3 you have a misplaced sign.  Superman saves 60 SB for 18 runs, but he has 25 fewer CS than average, costing his team -15 runs, for a net of 3 runs.  Close enough to average, and closer still using the proper run weights.  But as Guy has mentioned and we discussed a few weeks ago, this is incomplete since we don’t know enough about the value of having runners in motion.

Kincaid, you are batting pretty close to 1.000 in explaining WAR this thread.  Thanks.

An average catcher in my system is one who is a -10 hitter and average fielder.  With the position adjustment, he will rate by WAR the same as an average player at other positions, at least on a per game basis.

There may be some sentiment on player rankings to give a catcher a little extra credit because the demands of the position mean he plays 130 games instead of 150.  If I wanted to give that credit I could just multiply catcher WAR by 1.15. 

Problem is that for the position as a whole I’d now by giving average catchers 2.3 WAR per team while 1B get 2.0.  As Guy mentions, backups being better rested and don’t get the same penalty.  If I only apply that 1.15 to the starting catchers then I’m still too high on position WAR, at 2.25.

Also, my calculations get really messy.  How much extra credit does a catcher get for 60 games? 90 games?

I will keep WAR what it is:  How many games you would expect to lose if that player had been hurt and needed to be replaced.  If you think other adjustments are needed to make it a “greatest players of alltime” list then feel free to make them.

We should know the value of having runners in motion: look at the move rates with 2 outs (or better yet, with 2 outs and full count).

Anyone?

Tango:  Wouldn’t you want to compare the change in RE when runner is going vs. not going, for situations where runner on 1B, 2B open, and batter puts BIP?

If anything I am suprised that there are catchers valued as high as there are. If you’re an elite hitter (or even just a great hitter) you’re probably giving up something like 3000 at-bats over your career by playing the catcher position.

So why would anyone chose to do that? And even if we don’t have any good way to calculate how the somewhat specific physical abilities needed to be a catcher translates to other position I think we can safely say that almost all catchers could’ve (if they had switched early enough) played some other position adequatly.

For most catchers that other position would have to be first base.  Few of them have the speed to play anywhere else, at least play well.

The vast majority of catchers don’t hit enough to get consideration at first base.  Catcher is their best route to the big leagues.

Guys like Wieters or Mauer certainly have the bats to play elsewhere.  Wieters as a 1B would probably be called the ‘next Mark Teixiera’, switch hitter, same college and everything.  Mauer could probably have played 3B or the outfield.

Why do they catch?  Probably because they are good at it, their organizations think having them catch gives them the best chance to win, and they are more focused on the goal of winning than maximizing their career longevity.  I doubt a 23 year old has any concept of the toll catching will take on his body when he’s 35 anyway.

You guys keep missing the solution: steals are per runner, caught stealing are per steal attempt. Supermaskman is 0/0 when a normal catcher is 60/25, and this rates him at +12—he didn’t get a chance for a caught stealing, so he’s really not short any.

Great post, Tango, and I’m sorry I just got to it now. Just want to say that I agree, although I didn’t know quite how to say it (certainly not as well and as well-researched as you have done), as I ran into that with my crude catcher defense valuation for 2009. People always want to add “reputation runs” for catchers who are “feared” and thus don’t have to deal with as many attempted steals.

Might I suggest a rough analogy? We know that Barry Bonds (and others, but let’s stick with Bonds for the simplicity), in the 00s, especially, got “pitched around” a lot. Not just iBBs, but a lot of “intentional unintentional walks.” If he hadn’t been treated that way, if he had been pitched “normally,” we would have good reason to think he would have hit more home runs. Does that mean we need to credit him with additional “reputation” linear weights beyond what we already do for walks? I say no, of course, but if you follow the “reputation” logic with catchers, shouldn’t you follow it with dominant hitters?

Here’s another analogy.  Let’s presume that Mariano Rivera is the greatest relief pitcher ever.  But, suppose that in baseball, it’s the opposing manager that gets to decide when to bring in Mariano Rivera.  Isn’t he likely to bring him in only when there’s a blowout? And so, he can neutralize Mariano?

That’s what happens with catchers.  If the catcher is so good that the only runners that take a chance on him are the great runners (and thereby have no chance at the guppies at the poker table), then he won’t be as effective.

And the poker analogy also works too.  You can’t be so effective a poker player that no one is willing to bet against you.  I had this problem in school once, where the guys knew that my draft system was so good, that no one wanted to play against me.

If I remember correctly that’s why draw poker died out right?  The fish got beat up too much and the good players had no ones money to take.

Superman has a wicked snap throw.  There’s got to be some extra value in having all runners start on the bag like it’s softball!

One commenter made a good point, that pitchers wouldn’t have to worry about a runner, well, running, and therefore could concentrate solely on making his pitches.  Wouldn’t that matter to some extent in this argument?  If a pitcher could worry less about the man on first when Bench is catching?

I think it’s a fair enough point that since no one runs on IRod, the pitchers might be able to focus better on the batter.

Then again, since the runner disrupts the batter, the batter is going to be able to focus on the pitcher too!

One point and one question:
1) Just looking at move rates with full count and 2 outs will not tell the whole picture of the value of having runners in motion, because in that situation the fielders hold their positions, unlike on a typical steal attempt.

2) Are we statistically confident that a pitcher’s ability to hold runners is not affecting these results in a meaningful way?  With someone like Pudge I doubt it would be a problem, given that he’s played for so many teams and caught so many different pitchers, but for Bench, who caught for one team, for a shorter period of time, in a time when players changed teams less frequently, and starting pitchers threw more innings, it seems like a couple of very good or very bad pitchers (at holding runners) could have some effect.  Just imagine catching Chris Young for a few years of your career!

Preston, in my WOWY, I take care of that.  It may SEEM that there is not enough non-Bench to compare to, but that’s not the case.

Tango/26:  Not sure which thread it was, but I’m pretty sure we discovered that the baserunners-disrupt-batters thesis from The Book was mistaken, or at least the evidence is far from conclusive (the only significant error I know of).  Maybe you included batters who struck out on SBAs (but of course not those who draw a walk or make contact)?  Or something else.  In any case, I’m pretty sure MGL, at least, agreed that there isn’t strong evidence in favor of the disruption theory. 

*

Couldn’t WOWY be used to compare runner advancement and GDPs by catcher?

Tom:

You assert that “[t]he value comes not in preventing runners from stealing, but in actually removing them from the basepaths.” While obtaining an out through a CS is certainly valuable, your analysis misses other aspects of the value of a catcher who, by reputation, causes fewer runners to attempt a steal.  First, by keeping the runner on first, a catcher creates more opportunities to obtain outs through GDPs and force-outs at second.  Second, by keeping a runner out of scoring position, a catcher decreases the opportunities for the opposing team to score runs by driving in that runner.  Your analysis also punishes catchers who happen to be catching for pitchers with good pick-off moves who keep a runner from attempting a steal.

Thanks,

Brad

In my research, I came up with a compete wash, but I don’t think it has to do with “disruption” of the batter.  I think it has to do with batters occasionally taking pitches that they would otherwise swing at.  I think that the evidence suggests that on the average, the batter is neither helped nor hurt with a good runner on first, but there are two caveats:  One, the evidence is not conclusive, I don’t think. Two, it REALLY depends on the batter.  Remember what Tango found. The difference between an experienced batter and non-experienced one was enormous, so there appears to be a potential for the runner greatly affecting the pitcher (both in disruption and the number of fastballs thrown), and it all depends on how the batter handles the situation. 

Speaking of, with all he great pitch f/x research, no one has looked at how a base stealer on first base affects the quality and type of pitches thrown?  Come on guys!

Speaking of, with all he great pitch f/x research, no one has looked at how a base stealer on first base affects the quality and type of pitches thrown?  Come on guys!

I’m pretty sure someone has looked at this.  I vaguely remember seeing it.  John Walsh, maybe?

Rally, you are right, I frakked up Superman’s analysis.

There are so many minor countervailing aspects to this, so many little advantages to each side. Maybe a decent subject for a wisdom of crowds survey of the bright minds who haunt this site. If you have 2 catchers playing 130 games, but catcher A is Jhonny Superbench (a Native American with a golden arm) who nobody attempts to steal on, and catcher B is Joe Average, who throws out 30 would-be base stealers out of 100 attempts. Both players have essentially produced zero lwts runs on steal attempts.

Which catcher would you rather have? My instinct is to go with catcher A, having read all of the pros and cons. The reason is that he has more throwing talent, and it’s probably better to assume that the more talented player is having a bigger net impact on the game, even if it is difficult to quantify at present.

The first time a runner tried stealing on Superman, he was safe, as the throw came with such force that it ripped the shortstop’s arm off and carried through the centerfield wall.  After that Superman stopped throwing, he just ran to 2B and tagged the runner out himself.

"You guys keep missing the solution: steals are per runner, caught stealing are per steal attempt. Supermaskman is 0/0 when a normal catcher is 60/25, and this rates him at +12—he didn’t get a chance for a caught stealing, so he’s really not short any.”

You don’t really believe that, do you?  Would you evaluate a 60/25 basestealer as 12 runs better than a runner who never makes an attempt?  If not, what is the difference?

"You guys keep missing the solution: steals are per runner, caught stealing are per steal attempt. Supermaskman is 0/0 when a normal catcher is 60/25, and this rates him at +12—he didn’t get a chance for a caught stealing, so he’s really not short any.”

Solution to what?  Yes, it is simple.  It depends on what the average basestealer/catcher does AND the values you assign to a SB or CS (and the otehr attempts).  Bottom line is that if runners are running too much, the 0/0 guy has less value than other guys.  If they are running way too much, the 0/0 guy has a lot less value and he should back off a little.  If they are running optimally or less, then the 0/0 guy has the most value of anyone.

You have to remember that the only thing that counts is the value of the catcher compared to the average catcher (which is the same thing as adjusting for the average runner).  I don’t really understand Charlie’s post, but if it is wrong, it it because he is not comparing the 0/0 to the average catcher, and of course you have to stipulate what the average catcher does in order to set a value for the 0/0 guy.

A 60/25 (I am not sure what those numbers mean - I assume 60 SB and 25 CS) catcher means nothing unless you know what the average catcher is (and of course you know the value of the SB and CS).

If we use .2 and .45 then the 0/0 guy is .75 runs better than the 60/25 guy, but that doesn’t mean that the 0/0 is above average.  Depends what the average is.

So I don’t know where Charlie gets 12 runs from and I don’t know what his point is. 

A catcher’s absolute value is simply his SB allowed times the average value of each of those SB minus the average value of each of his CS times his CS.

The relative value of a catcher (which is mostly what we care about) is an average catcher’s value given the same number of innings or base stealing opps (or whatever you want to use) minus that value.

I don’t see the controversy here.

I think it would be obvious that catchers wouldn’t show up too high on the list.  Not only do they only play 130 games per year, they end up wearing down and get moved to different positions quite often.  One of the bad things about it, is even if you count their contributions at other positions, they usually get moved to 1b or DH, so they end up becoming much less valuable.  It isn’t like a shortstop getting older and moving to second, costing him 5 runs per year.  These guys lose like 30 runs when they move, and that’s tough to make up.

Like MGL said, total WAR should be a lot lower for catchers than most, but WAR/game should be about the same.