Value of Defense at Catcher

[Guest Rotblatt]

Attempting to reduce a complicated concept like measuring a player's impact in a game by looking ONLY at W-L record is frankly silly.

It's like looking at Reyes' 99 runs last year and concluding that he must have reached base a lot.

I'd suggest looking at Win Shares, VORP, etc.--whatever metric floats your boat.

[Guest KC]

Nice pics JD, you can lead a stallion to a gold platted trough of purified
spring water but you can't make him partake in the pleasure.

[Guest old original jb]

Predictive models based on the sum of the predictive statistics of individual players calculated using league averages will work best to predict average performance and will [u:202a443da8]underpredict[/u:202a443da8] extremes in either direction.

The estimates of how likely it is that making or not making a defensive play will prevent/lead to a run surrendered are based on the average of the situation accross all teams. But if you are on a very good defensive team, it is likely that your mistakes will have less impact and if you are on a bad defensive team they will have more; conversely, on a bad defensive team, a star defender may save more runs than on a great defensive team. The worst combination would be more than one well below average defender plus below average pitching.

You would expect adding a poor defensive player on such a time to have a much worse impact, and that such teams would do much worse than predicted on the basis of the sum of the impact statistics of their players.

Therefore, the balance of offense/defense value is profoundly affected by how far below or above average someone is and how far below or above average his teammates are. Adding a slightly below average player to a good defensive team would have minimal impact. Adding a well below average player to a below average team could be the straw that breaks the camel's back.

[Guest Johnny Dickshot]

My guess would be the record is more or less the same, a few points up or down, and agree that it wouldn't say much nor show cause and effect.

[Guest Bret Sabermetric]

Piazza aside, don't you agree then that the impact of individual players on a team's performance over the course of a season is overstated, usually in proportion to the evaluator's sophistication?

Somewhere James makes this comment: If a team loses its big star for two weeks, that team tends to get by surprisingly well while he's out. That's one of the bases for my argument to trade stars having good years: If trading Star X actually costs you only 2 or 3 games over the course of the season (after you've gotten far behind in the pennant race anyway) and trading him lets you get out of an unfavorable LT contract, get younger, get more versatile, look at your own players in Star X's role a little bit, etc., then it may be worth doing. Fans typically respond "NOOOO!!!!! It will kill the team to deal off Star X for a lot of garbage, and we still haven't mathematically been eliminated until the fat lady has sung yyybbb." Individually, this might be true, but when a team persists in hanging onto veterans year after year to make pennant runs that either don't happen at all or fail, well, I have a problem with that, as you may have noticed.

[Guest Johnny Dickshot]

Bret Sabermetric wrote: Piazza aside, don't you agree then that the impact of individual players on a team's performance over the course of a season is overstated, usually in proportion to the evaluator's sophistication?

Well, I prolly agree with the first part though every situation is different.

I noticed your other problems.

[Guest Rotblatt]

="Bret Sabermetric"]Piazza aside, don't you agree then that the impact of individual players on a team's performance over the course of a season is overstated, usually in proportion to the evaluator's sophistication?

Sure, but that's why we have metrics--to look objectively at the contributions of each of our players. I can look at the 2005 WARP3 scores, and verify that, in fact, Wright was our most valuable hitter, contributing 8.5 wins over a replacement 3B.

="Bret Sabermetric"]Somewhere James makes this comment: If a team loses its big star for two weeks, that team tends to get by surprisingly well while he's out.

Yes, that's true. The point, however, isn't that the big star isn't valuable, but that his value is spread out across the entire season. The difference on a day-to-day basis isn't meaningful--it has to be taken in the context of the entire season.

I mean, Wright created 0.053 wins per game last year. Matsui created 0.016. Wright would have to sit out for 27 games in order for us to "lose" one single game because Matsui's playing instead of him.

Now, over the course of the season, that difference spreads out to 6 full wins, but on a game-by-game or even week-by-week basis, the impact is negligible.

[Guest old original jb]

And I'm saying that the predictive value of the metrics can depend in complex ways on how far from average the rest of the team is. Especially true for defensive (see my argument above), but could also be true for offense.

If I am not mistaken, win shares are based on the impact of what a player contributes calculated using data from all teams and all games. But the impact of a hit, a run, a catch, an error etc. may be very different depending upon the expected performance of the other players.

For offense the relationships may be even more convoluted than on defense and may in fact have a different relationship to overall team skills-- that is weaker effects when teams are at either extreme. If an offense is very weak, I think that the loss of the star hitter will mainly result in losing by slightly bigger margins than usual, but may not be a big factor since he will have few opportunities to drive in or be driven in by his teammates. If an offense is great, than one player going down may also not have a big impact because others will pick up the slack. It will be a bigger impact than on the bad offensive team, though, because on a good offensive team the players increase one another's impact.

I would predict that the biggest impact for loss of a slugger would be on teams with average to somewhat above average offense where that player provides a large chunk of the production, but the other players are on base or driving him in enough to give a multiplier effect to his individual production.

If I ever find the time to find a way to prove this with math or with real raw data and statistics, I will do so. If someone can draft a grant proposal for this work, I'll send it to NIH and see if they'll fund me to do it as my day job

[Frayed Knot]

]I can look at the 2005 WARP3 scores, and verify that, in fact, Wright was our most valuable hitter

well, by one method of measurement is does.
But let's not go mistaking whatever metric you want to use for the scientifically-stamped final word.

[Guest Rotblatt]

Frayed Knot wrote: well, by one method of measurement is does. But let's not go mistaking whatever metric you want to use for the scientifically-stamped final word.

Yeah yeah yeah. I'm willing to be that most metrics (including the ubiquitous CPF SPoY) would show David to be our most valuable player last year, but your point is well taken. Metrics are to be used in moderation and taken with a grain of salt.

[Guest Bret Sabermetric]

Rotblatt wrote: [The point, however, isn't that the big star isn't valuable, but that his value is spread out across the entire season. The difference on a day-to-day basis isn't meaningful--it has to be taken in the context of the entire season. I mean, Wright created 0.053 wins per game last year. Matsui created 0.016. Wright would have to sit out for 27 games in order for us to "lose" one single game because Matsui's playing instead of him. Now, over the course of the season, that difference spreads out to 6 full wins, but on a game-by-game or even week-by-week basis, the impact is negligible.

Say the difference between playing Matsui 162 games at 3b and playing wright 162 games at 3b is 6 games (M=5-11, W=11-5). If Wright were to cripple himself today, and Matsui were to take his place on the roster for the rest of the season, Mets fans would go absolutely nuts. They'd say that we just lost the season. If you suggested that difference was the difference between going 85-77 and going 79-83, they'd say you were nuts ,the difference is more like 90-72 and 70-92 because they tend to minimize the contributions of a Matsui and exaggerate the contributions of a Wright.(and oversstate the strength of the team generally). That makes it very hard to move a star player, because if you did move a star, and if contrary to expectations the move failed, you will get vilified. The risk of that worst-case scenario is much greater than the possible gains of moving the star, at the least from a GM's perspective. In reality, there is usually immediate value coming back to the other way, which, coupled with the surprisingly low impact of just losing the star player even without compensation, makes each such deal much safer than you'd think.


For example, I don't think the Mets got anything like real value in the Cameron/Nady deal. I hated it. But I'll concede that Nady is no worse than a 5-7 player this season, and that Cameron is no better than 7-5, and I'll also concede that maybe I'm wrong in my evaluation of these players. It's a big deal in the context of the Mets making several other foolish deals, and the total is maybe 10 or 15 lost games, but the Nady/Cameron deal will probably cost the Mets under 2 net games even if the deal shakes out the way I think it will.

i don;t recommend making stupid trades on General Principles, but I think each one's downside is limited.