Thursday, April 17, 2008

Guest Post: Everett vs. Tolbert

Nick's Note: Good game last night, but I just don't have time write about it, so I'm going to present you with this guest post submitted by frequent reader and occasional contributor Corey Ettinger. I will note that I'm a little skeptical of the numbers he presents, because I think that a) Everett is not nearly the defender he was in Houston anymore, which may be because of age or because of a shoulder ailment; and b) I think Ettinger is probably underrating Tolbert by putting him on Brendan Harris' level defensively. Nevertheless, it's a well-written article, so I'll leave you to formulate your own opinions based on the evidence provided. Thanks Corey!

***

Matt Tolbert
is playing out of his mind right now, and if you're the Twins, this is clearly a good thing. His play is even causing a groundswell among the fan base to make him the starter at shortstop and I think this is something that needs to be addressed. In one camp, you have people who like the fact that Tolbert should be a better offensive player over the course of the season. The other camp believes that Adam Everett can save as many or more runs with his glove. Let's break it down:
NameLifetime AVGLifetime OBPLifetime SLGAVG HAVG BB AVG TB AVG 2BAVG 3B AVG HR Runs Created
Everett.246.298.355135
30.41
195
25.80
3.54
8.86
55.14
Tolbert.280.345.404154
46.44222
29.57
7.70
7.70
74.49

So over the course of a season, Tolbert is good for an additional 19.35 runs. That's pretty significant. If you consider that the Twins averaged 4.43 runs per game last year, that's an additional 4.37 wins that Tolbert bring to the team. Of course, that also presumes that Tolbert can translate those minor-league numbers into major-league numbers. While it is far from impossible for him to do so, I think many would say the odds are against that given the traditional growth rates of players transition from the minor leagues to the major leagues.

Now is where things get a bit tricky and I'm prepared to admit that defensive statistics are not my forte, but using some pretty simple logic-based calculations, I think we can fairly easily translate Everett's effectiveness into a sort of reverse runs created. Let's say, runs prevented. Also, as a caveat to anyone who is interested in really understanding defensive metrics, PLEASE look into buying a copy of The Fielding Bible; its written by some of the greatest minds in sabermetrics and is a MUST for anyone who is interested in understanding precisely how defense impacts a game. Furthermore, even if you aren't going to buy the book, I strongly recommend reading the excerpts posted on their website.

The system they use is called +/-, and essentially, after charting every single ball put in play, they determine based on ball speed, location, vectoring, etc... how many more (or less) balls a player converts into an out than the average. Players who get to balls below the average receive negative ratings. Adam Everett led all shortstops in the major leagues from 2005-2007 with a +92 rating (Jason Bartlett was second with a +45.) While I'd love to use these stats, it's not possible since Matt Tolbert's numbers aren't available. So we need to find another way of doing this. Getting back to setting up our own system...

First, I'll be using The Hardball Times fielding stats. A basic fielding stats primer: ZR= Zone Rating BIZ= Balls in zone. Plays = # of balls in a players zone fielded (Plays/BIZ=UZR) OOZ= Outs made on balls out of a fielders zone. Since shortstops don't really have any effect on triple and homeruns, and only have a very small degree of effect doubles, I'm going to presume that every out that a fielder doesn't make, counts for a single. In order to ensure the closest approximations possible I'm going to use the data available from Everett's past 5 seasons. In Tolbert's case, I only have a sample size of nineteen innings. Clearly this presents a major statistical dilemma, but I can only work with the information available to me.

As a precursor, I'll be using an average total balls in zone for the shortstop position over a 162 game season of 420. That will effect a players AVG Plays as well as thier AVG OOZ plays. Obviously the last shortstop to play 162 games in a season was Miguel Tejada (and before him Cal Ripken Jr) but guys who do are rare. And neither Everett nor Tolbert will do so. But it gives us an important baseline to work from.

NameAVG BIZ AVG Plays (420*ZR)AVG OOZTotal
Everett420
369.60 (.880)
77.90
447.50 (448)
Tolbert
420
326.76 (.778)
¹37.00
363.76 (364)

¹: Brendan Harris' 2007, a player I think Tolbert compares well to. I had to make an arbitrary decision since Tolbert has yet to record an out from outside of his zone. Harris created one out of zone out per 20.29 innings last year. Tolbert having played 19 innings and not yet made an OOZ play was the nearest comparison given Tolbert HAVING made an OOZ play.

This shows us that Everett creates 84 more outs over the course of a season than Tolbert. Given that there are 27 outs per game you COULD say that means Everett's glove wins 3.11 more games per year than Tolbert's.

But now we need to plug this into 2007 hitting data. Last year the average major-league hitter batted .268/.336/.422. Over a 550 at-bat season that comes to 147 hits, 185 times on base (H+BB). Therefore, the average batter would've had ~38 BB. A .422 SLG works out to 232 total bases. If Everett prevents 84 more outs per season, that means he prevents 84/9 (9.33) outs (# of additional out/batters = Individual batters share of additional outs created) per batter.

Plugged into the same basic RC formula we used above, this "standard" hitter who had been averaging a RC of 72.99 would now be averaging an RC of 66.75. Now, since Everett's defense is in play over 9 batter over the course of a season, that negative variation of 6.24 becomes multiplied by 9 to equate to 56.16.

Everett has a 56.16 Runs Prevented rating over Matt Tolbert. That would equate to 12.54 wins given the Twins average allowed runs of 4.48 per game.

Of course so much of this is speculative. We don't know if Tolbert can duplicate his minor league numbers and my study of the fielding stats is wildly simplistic. But I think the study is fairly accurate, and I'd be happy to discuss it with anyone who might be interested. The final results show that Tolbert's 4.37 additional wins due to his bat (over Everett's) do not make up for Everett's additional 6.24 wins from his glove over Tolbert's.

11 comments:

Anonymous said...

what if you factor in everett's bum shoulder?

Dave Thompson said...

I was a big fan of Everett's when the season started, but his bad shoulder means he will probably be on the DL for a while. I sure hope it's Tolbert taking his place, not Nick "no hit" Punto.

SL__72 said...
This comment has been removed by the author.
SL__72 said...

I spent a lot of time reading/thinking about run/win math this off-season. Using the following projections for Everett I came to the conclusion that he represents roughly 2.5 wins over a replacement level player:
0.245/0.292/0.343
+31 runs over AVERAGE defensively at SS.
I might be under adjusting for position, but since I'm comparing him here to another SS that doesn't matter.

Using this projection for Tolbert:
0.250/0.308/0.370
8 runs below average defensively at SS (The projection I used for Harris)
I have Tolbert as basically the definition of a replacement level player.

So while my method is quite different, I came to the same conclusion: that a healthy Everett is worth about 2.3 wins more than Tolbert at SS.

None of this really matters if Everett isn't playing as well defensively due to injury or otherwise...

What about this idea I suggested on my blog yesterday:
Start Tolbert when you have an extreme fly ball pitcher going (Slowey, Baker) and Everett when you have one who gets more ground balls (Blackburn, Liriano, Livan, Bonser).

A whole different kind of platooning...

Nick N. said...

I don't doubt that Everett is a reasonably valuable player when healthy and playing top-notch defense. The problem is that I've seen nothing from him so far to indicate that this is the case, which renders the point someone moot.

If Everett is not playing good defense, his unspeakably awful performance at the plate turns him into a fairly awful player, and in that event I'm sure the Twins could get more production from a Punto/Tolbert combo at SS. With that said, I still haven't heard anything about the results of Everett's MRI.

Nick N. said...

Also, I apologize that some of Corey's tables in this post got cut off (at least on my browser). I didn't really know how to fix that and didn't have time last night to figure it out.

SL__72 said...

I completely agree. I was just curious how Corey's analysis would look next to mine since he did it in a completely different method then I do.

Anyway, story on the results of the MRI here.

Oh, one more note. In his article Corey says "Plays/BIZ=UZR" That is RZR, not UZR.

UZR is more like the +/- system he also talks about.

Corey Ettinger said...

Apologies for the Ultimate v Revised mix up. Quite unintentional.

And yes, if Everett's defense were to slip for any reason, his value as a player is completely negated.

If the Twins are to compensate for Everett's lost value however, they must be playing Tolbert over Punto which was not the case last night. Punto's defense as has been pointed out numerous times is wildly over-rated. Combining his poor bat with his average at best defense at that position is the wrong move.

What must be surprising and pleasing to Twins fans would be Brendan Harris' defensive play at second where his RZR is second in the AL with a .885. Regrettably he's still as range limited as ever, having made no plays out of his zone. But at least he's doing a far better job of fielding the ones he should.

The lack of range combined with his poor turn abilities on the DP are going to hurt the team on days when we have a GB pitcher going though. And a combination of Harris with either Punto or Tolbert would leave our MI defense as probably the most range limited in the league.

SL__72 said...

Tolbert is getting the start there tonight according to LENIII.

Anonymous said...

over the course of a season, Tolbert is good for an additional 19.35 runs. That's pretty significant. If you consider that the Twins averaged 4.43 runs per game last year, that's an additional 4.37 wins that Tolbert bring to the team.

honest question: how do you come up with the claim that Tolbert's 19.35 additional runs would lead to an additional 4.37 runs, given the Twins' avg run production last year?

Off the top of my head, it seems that most of those additional runs would come in games in which the Twins already won (in which case, zero additional wins) or lost by more than a run (in which case zero additional wins), or lost by one run (in which case the add'l run generates a tie and xtra innings, thereby something like a coin flip).

so you'd need to know the distribution of observed margins to guesstimate an effect on wins from these marginal runs, right?

just ak-sing.

Nick N. said...

If the Twins are to compensate for Everett's lost value however, they must be playing Tolbert over Punto which was not the case last night. Punto's defense as has been pointed out numerous times is wildly over-rated. Combining his poor bat with his average at best defense at that position is the wrong move.

I'm not sure if I'd agree with that assessment. If anything, I'd say you're wildly underrating Punto's defense by tabbing it "average at best." I think Punto is a considerably better defender than most defensive metrics imply, and I also think he's a substantially better defender than Tolbert. He's also probably not as far behind Tolbert offensively as one might think.

Everyone wants to bash Punto at every possible opportunity after his miserable 2007 campaign, but the reality is that he won't repeat those terrible offensive numbers, and he really is a strong defender at any infield position. If Everett is out for a while, I like the idea of playing Tolbert on days where a fly ball pitcher is on the hill, with Punto starting behind guys like Liriano and Blackburn.