You are here > Home > Primate Studies > Discussion
 
Primate Studies — Where BTF's Members Investigate the Grand Old Game Monday, June 04, 2001Can a Really Great Glove Make Up for a Really Weak Bat?Can guys like Rey Ordonez and Pookie really save enough runs with their glove to make them valuable? Time and time again, you will hear someone say, “Rey Ordonez saves hundreds of runs with his defense. It’s okay that he can’t hit.” Is that true? In one sense, it is true that Ordonez saves hundreds of runs, but that is compared to no one playing. Every shortstop saves hundreds of runs. Does Ordonez save more runs than your average shortstop? And is the number of runs he prevents as many as he costs the Mets at the plate by being such an anemic hitter? Today’s sabermetric tools allow us to analyze baseball events in terms of run values. I won’t pretend to understand the calculations behind VORP (www.stathead.com) or EQA/EQR (www.baseballprospectus.com), but I have a good grasp of Pete Palmer’s Linear Weights, Bill James’ Runs Created and Jim Furtado’s Extrapolated Runs. You can take any batter’s hitting line and convert it to run values. Since Rey missed most of 2000, I’ll be using another defensive whiz/weak bat player: Pokey Reese. Reese has reportedly been one of the holdups in several trades for the Reds. Jim Bowden simply won’t part with Pokey. Is Pokey worth it? Does his glove make up for his bat? The work I’m about to perform isn’t new today. I first looked at this in 1998, and developed a player rating system from it. Like all player rating systems (yes, all), this is fallible and has weaknesses. There are flaws in the systems I use, but none too severe to make the rating system “bad”. My rating system is the only one that incorporates defense in the player ratings. Also the run value accuracies aren’t pinpoint; they’re estimates. They are good estimates, and the finiteness is expressed because that’s what the formulas yield. As a general rule (and this applies to all rating systems), players that are close in run value can be described as equal. Try to set a mental range of 58 runs when looking at a player’s numbers. Offense can be calculated with a variety of systems. I use Jim Furtado’s Extrapolated Runs. For a really indepth explanation, see http://www.baseballstuff.com/btf/scholars/furtado/articles/IntroducingXR.htm. xRuns  Extrapolated Runs: = (.50 x 1B) + (.72 x 2B) + (1.04 x 3B) + (1.44 x HR) + (.34 x (HP+TBBIBB)) +(.25 x IBB)+ (.18 x SB) + (.32 x CS) + (.090 x (AB  H  K)) + (.098 x K)+ (.37 x GIDP) + (.37 x SF) + (.04 x SH) Here’s Pokey’s line from 2000:
If you care to do the math, that’s 70 xRuns. I prefer to let Excel do my work. Before we can really use those 70 xRuns we have to account for the conditions under which they were generated. First, the runs have to be parkadjusted. In 2000, Cinergy Field was a runscoring haven, second only to Coors Field in terms of park effect. This is unusual, as Riverfront has largely been a neutral park (Park Factor, PF, of 100102) for a long time. For my system, I use the present season PF, averaged with the 3year PF from STATS, Inc. From 199799, Cinergy had a PF of 102. For 2000, the PF was 116. This gives Cinergy a working PF of 105, and that is what we will adjust Pokey’s xRuns with. This adjustment gives Pokey 67 xRunsp+. We also have to adjust the xRuns total by the number of chances Pokey had to generate them. As “outs” are baseball’s currency, I use xRuns/out. Actually, for a season, I extrapolate to a full season’s worth of outs. A full season’s worth of outs for an average player is 465 outs (about). This number is not exact, but it is a good starting point, and I do judge all players from the same baseline, which is the important part. Pokey only used 405 outs (518132+3+3+5+8). This adjusts his xRuns upwards to 76 xRunsa+. The average NL second baseman generated 102 xRuns in 465 outs; 104 excluding Pokey. Jeff Kent, the NL MVP, generated 153 xRunsa+. Edgardo Alfonzo generated 149 xRunsa+. That means Pokey was 28 runs worse than the average of the other regular NL second baseman at the plate. Can his glove be that valuable? Note: Running these comparisons with Linear Weights or Runs Created will result in almost identical numbers. As I outlined in a previous article, http://www.baseballthinkfactory.org/articles/cdial_20010306_0.shtml, defense isn’t easy to quantify. Actually, it’s not so hard to quantify once you have a system, but it is hard to convince skeptics that your system makes sense and that it is a good estimator of defensive value. Some of you will be convinced by the following paragraphs; some of you won’t. STATS tracks all balls in play using Zone Ratings (ZR). For more on how ZR is kept and works, see http://www.bigbadbaseball.com/glossary/zone.html. I use ZR as the basis for my defensive rating system. Taking the average number of balls per inning hit into a fielder’s zone and multiplying by the average number of innings in a season, I get a number that represents the number of plays a fielder would get (at an average rate) if he played every inning of every game. This serves to normalize the number of chances a player would get and yields a theoretical number of runs a player might save/prevent. Here’s what that means: A defensive game (DG) is 8.75 innings (not quite 9). 8.75 innings times 162 games is 1418 innings. In 2000, Pokey Reese played 1129 innings, or 129 DG. He had 405 ground ball chances. That’s an average of 3.14 chances per DG. The NL average for chances was 3.05. A full season at 2B with an average number of chances would get Pokey 494 chances. Pokey’s ZR was 0.874. Had Pokey gotten 494 chances, he would have converted 432 of them. In comparison, NL MVP Jeff Kent had a ZR of 0.796. Kent would have converted 393 of those 494 chances. In equal chances, Reese would get 39 more outs than Kent. Well, what does 39 plays mean in runs? I think this converts to runs readily, and almost obviously: any play not made results in a baserunner. Each out has a value: if it takes away a single, that’s worth 0.47 (hit) + 0.28 (out). Each subsequent base is worth 0.31 runs (per Linear Wts). Here: SS/2B: (0.987*plays made*0.75) + (0.013*pm*1.06) That formula says that 98.7% of balls hit through the SS/2B zones are singles (when not outs), and 1.3% are doubles. There aren’t enough triples or home runs to count. These percentages were arrived at through analysis of actual balls hit through these zones. So Reese would save 326 runs (see, hundreds!). Kent would save 297 runs. Reese would save 29 runs (RS) more than Kent. The average NL second baseman would save 310 runs. Looking at offense and defense together: Player xRuns RS Total So a full season for both players would have Kent being worth 47 more runs than Pokey. Studies say that every 1011 runs is worth a team win. That means Kent was worth 4 more wins than Pokey. That’s not really fair to Pokey. Kent was the best overall second baseman in the NL in 2000. How does Pokey compare to the average second baseman? That will tell you if he is worth hanging on to. He was 9 runs below the average second baseman. That is pretty close. The error in my metric only allows me to say he was slightly below average. The value of 9 has some imprecision to it. It could be 5 runs worse; it could be 15 runs worse. Before you start asking about an average player versus a replacement player, I have little choice but to use average for defense. As long as I am comparing these players to each other, the baseline doesn’t particularly matter. The answer is, no, Pokey’s glove does not make up for his bat. Is it close? Yes, and I believe this same exercise for 1999 actually will have Reese above average overall. Is he a superstar? No. He just doesn’t hit enough. You can also see how much a good hitting player can be well above average even with some defensive deficiencies. For more on my methodology, see http://www.baseballstuff.com/fraser/articles/dpi.html.

BookmarksYou must be logged in to view your Bookmarks. Hot TopicsLoser Scores 2014
(8  3:36pm, Nov 15) Last: willcarrolldoesnotsuk Winning Pitcher: Bumgarner....er, Affeldt (43  9:29am, Nov 05) Last: Jolly Old St. Nick Is A Jolly Old St. Crip What do you do with Deacon White? (17  1:12pm, Dec 23) Last: Alex King Loser Scores (15  12:05am, Oct 18) Last: mkt42 Nine (Year) Men Out: Free El Duque! (67  10:46am, May 09) Last: DanG Who is Shyam Das? (4  8:52pm, Feb 23) Last: RoyalsRetro (AG#1F) Greg Spira, RIP (45  10:22pm, Jan 09) Last: Jonathan Spira Northern California Symposium on Statistics and Operations Research in Sports, October 16, 2010 (5  12:50am, Sep 18) Last: balamar Mike Morgan, the Nexus of the Baseball Universe? (37  12:33pm, Jun 23) Last: The Keith Law Blog Blah Blah (battlekow) Sabermetrics, Scouting, and the Science of Baseball – May 21 and 22, 2011 (2  8:03pm, May 16) Last: Diamond Research Retrosheet SemiAnnual Site Update! (4  4:07pm, Nov 18) Last: Sweatpants What Might Work in the World Series, 2010 Edition (5  3:27pm, Nov 12) Last: fra paolo Predicting the 2010 Playoffs (11  5:21pm, Oct 20) Last: TomH SABR 40: Impressions of a FirstTime Attendee (5  11:12pm, Aug 19) Last: Joe Bivens, Minor Genius St. Louis Cardinals Midseason Report (12  12:42am, Aug 10) Last: bjhanke 

Page rendered in 0.2978 seconds 
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
1. Darren Posted: June 04, 2001 at 12:07 AM (#603869)You are right about the double play. It is a significant factor  the difference between man on 1st and 1 out and none on and 2 out is enormous.
Reese would have to be evaluated in two skills:
The most important one is the pivot  so make a count of all grounders to 3B or SS in DP situations and see what pecentage are turned.
Then, balls hit to Reese  frequency of started double plays per opportunities.
Your second point about pitchers feeling comfortable: as with all psychological effects, the problem is that they are so small that it's hard to quantify them at all. Are you saying that a pitcher is less likely to walk batters or allow homers or more likely to strikeout batters with Ordonez at short ?
Before making such claims, it is necessary to first check and see how the pitchers performed with and without Ordonez at Short over his whole career (to get maximum sample size). If there was a very large difference, then you could start making a list of possible explanations for the split in the data.
Your explanation wouldn't be the first one to pop into my head  there could be many other plausible reasons.
However, I don*t think you even reached the tip of the iceberg on the issue. As others have said, reaching and catching balls is a large part of defense, but so is turning double plays... On the same theme, defense is a large part of what is often ignored, but there are other *little things* that should be quantified. For instance, base running (not stolen bases, baserunning) may not have the impact of hitting the ball, but a good baserunner, perhaps Pokey, could surely spawn an extra 1011 runs for his club over the season by avoiding double plays and advancing on basehits.
Fly balls  99.9% of infield popups are caught. There is a negligible difference among major league infielders in catching them. They don't reflect the skill of the IF, just the pitching staff's propensity to get them and the IF propensity to call off his teammates. I do think there are loopers that some 2B and SS go get better than others. Again, this represents a small percentage of plays made by the IF, and I don't have access to the data (chances versus plays made). Line drives are luck. The ball almost has to be hit right at you. Check out the line drive hit charts on ESPN.com
As for baserunning, getting caught once trying to take the extra base costs your team twice as much as it gains. Reese would have to take nearly 40 extra bases more than average to increase his value by 1011 runs. Remember these are run values, not how many times he scores from first on a double.
And for OF ratings, I do include OF assists.
Not so. Mike Gimbel's "Playball" software, which continues from his "Baseball Player and Team Ratings" books from the early '90's, takes defense into account, and I think he's got the best defensive system extant, because it does two things others do not:
1. it takes data by park, removing park effects
2. it takes the run value of plays not made, not just a "play not made." For example, a ball that an outfielder doesn't catch that turns into a double counts as a double. It's like Extrapolated runs against the defense, except his system is calls "RPA" (runs per plate appearance)
Cheers,
Alan Shank
Adjusting for park means he has data I don't (and a budget I don't). I've talked with him *very* briefly about his system, and from what I heard, I didn't agree with his park effect treatment.
Is his work published and accessible?
Regarding (a) the combination of offensive runs created and defensive runs prevented, and (b) just how good Rey Ordonez is, check out Matt Welch's article on SportsJones, "Why the Angels Are Better Without Mo Vaughn" (http://www.sportsjones.com/sj/444.shtml).
Yeah, I know we're talking about middle infielders and I'm bringing up a first baseman.
Anyway, in the article Welch points out how many infield throwing errors are committed with Vaughn at first vs. with other first basemen. (To illustrate: a throw in the dirt that is not caught is almost always ruled an error for the thrower; yet the first baseman's skill at scooping such a throw is a big influence on the outcome.) For comparison, he also does the same with John Olerud. The latter comparison is relevant when you consider the Mets' infield defensive stats with Olerud at first vs. with other first basemen.
The net upshot of the Vaughn analysis is that, if you adjust downward his hitting stats based on how much his glove helps or hurts the team, the result is roughly a 100point hurt to his BA, OBP, and SLG. Of that 100point difference, about 70 points are from his own "measureable" defensive stats, and 30 points are from his impact on the rest of the infielders.
The numbers look too drastic to be true, but even after arbitrarily adjusting them downward they appear significant enough to be part of the evaluation of Ordonez/Reese/etc. And perhaps part of Welch's methodology  if you want to call it that  could be used to adjust for botched doubleplays a la Offerman.
Is his work published and accessible"
He sells a CD called "Playball" software. It has data from several years back, plus the explanations that were in his books. It costs about $50. His email address is:
caiman@ptdprolog.net
"Adjusting for park means he has data I don't (and a budget I don't). I've talked with him *very* briefly about his system, and from what I heard, I didn't agree with his park effect treatment. "
Well, he compares the average "RPA" of balls hit into a fielder's area for the home player to that of all the visiting players, then does the same thing for this home player on the road, vs. all the players in their parks. With the software, you can actually see the comparisons.
"One real basic question: I notice that I'm having trouble grokking the difference, and the relative advantages, between counting things "per out" versus "per plate appearance (basically outs plus reaches). Help?"
I prefer to measure things "per out," because that's how time is measured in baseball. Let's say a ballpark has a certain effect on home runs hit per ball put in play (if we could actually know this). The actual effect on the number of homers hit is going to depend on this plus the number of balls put in play, which depend on the number of plate appearances. The number of plate appearances is strongly affected by the OBP, which is in tun affected by the park's effect on batting average. Let's say, further, that the old Oakland Coliseum and Coors Field have the same effect on homers per ball in play, with their other characteristics remaining as they are. Since Coors increases batting average drastically and the Coliseum reduced it considerably, there are going to be a lot more plate appearances and hence balls in play in Coors than in Oakland, so there will be more homers, despite the fact that they affect homer/balls in play equally.
This is why in my park factors and in my analyses of teams I use "opposition innings pitched," i.e. team A's offensive stats per team B's innings pitched and vice versa. The numbers of innings pitched in the different parks are going to be much closer than the numbers of plate appearances.
For individual players, since I usually use the reduced XR formula, I don't have the number of outs a player uses; you need SF and GIDP to get close, and you're still estimating by using AB  H, so I usually just use per PA. But I still think per out is more meaningful.
"One sweeping question: "How good is Pokey?" is not nearly as interesting to me as "How much relative extra value is there between the best defensive players and the average defensive players (and the worst)  typically/generally, for each position?" "
Interestingly, in Mike Gimbel's system it's the outfielders who have the largest variations from "average." I assume this is because, when an outfielder doesn't reach a ball or misses it, it's likely to turn into a double or triple, whereas most balls hit in the areas of middle infielders and not converted into out are singles. Here's an example: for Geoff Jenkins in 2000:
home 179 balls in zone, average RPA .097
visitor average RPA was .149
away 179 balls in zone, average RPA .100
home average RPA .165
That's a very large difference. For comparison, here's data for Rey Sanchez, whose "runs saved" rating was 18, compared to Jenkins' 44 for 2000:
home 286 balls in zone, average RPA .065
visitor average RPA .060
road 266 balls in zone, average RPA .056
home average RPA .076
Cheers,
Alan Shank
Cheers,
Al
You must be Registered and Logged In to post comments.
<< Back to main