April 21, 2006

Baseball By The Dumbers: Part I: What happens when people with too much time try to invent statistics. Overcoming a lack of math and reasoning skills, Sportsfilter member BullpenPro gives it a shot.

posted by justgary to baseball at 10:51 PM - 22 comments

As always, good stuff from BP. It reminds me slightly of the old GWRBI stat, the much maligned (since Bill James essentially does not want to see any stat about "clutch" hitting) and equally flawed stat. It was flawed because it was the RBI that put a team ahead, not necessarily the one that was the one they needed to win the game. Maybe "clutch" factor could be defined as, the GWRBI, made in the 7th inning or later, that was made with RISP. Although this would be flawed as well, because it discounts the GWHR. But, isn't that more often a pitcher's mistake rather than a hitter's "clutchness"? Ah, the pain in my head just from trying to define what it should be, forget about going back and actually trying to figure out what would actually qualify.

posted by elovrich at 11:20 PM on April 21, 2006

Good job, there... I remember also when the GWRBI was the en vogue statistic, I didn't know Sabermetricians had something to do with its demise. What about incorporating an RBI, no matter the score of the game, that is delivered with two outs? Surely that's a clutch RBI as it's a team's last chance. For that matter any RBI in a game where a team is behind could be clutch, as it is contributing to a team coming back?

posted by vito90 at 02:12 AM on April 22, 2006

see how hard it is to define clutch? while lyihng in bed awake last night pondering this, it occured to me that it could involve also, any RBI with RISP if your team is within 2 runs, last 3 innings, or any extra-base hit, to put yourself in scoring position, with less than 2 out, late innings, and score within 2 runs.

posted by elovrich at 09:12 AM on April 22, 2006

Sabermetricians may have hated the GWRBI, but they did not cause its demise. That was the result of the public's general indifference to it and sportswriters' incessant complaints that an RBI in the first in a 10-8 game could just as easily be the GWRBI as a RBI in the ninth. There have been lots of definitions of clutch. But nobody agrees with any other definition. And clutch statistics usually come in such small sample sizes that they are meaningless (He's gone 3 for 6 with the bases loaded!) Plus the fact that no one can find any evidence that players who are better in the clutch (however it's defined) in one year are more than 50% likely to be better in the clutch the next year. elvorich - Bill James is by no means insistent that there's no such thing as clutch hitting or that we shouldn't count it. In fact, he once invented a stat called Victory Important RBI. It was Bill's audience much more than Bill that pushed that stat into oblivion.

posted by spira at 10:27 AM on April 22, 2006

since a team is really in a do or die situation when it has 2 outs, a clutch RBI should any RBI delivered when there are 2 outs, in a close game (+/- 3 runs), an it should not matter how it was accomplished. I mean there should be no difference between a Home Run or a bases loaded walk. The bottom line is that when there are 2 outs, the team is living on the edge. If that player makes out, the entire inning is over and an entire new rally has to begin the next time. Coming through under those circumstances is clutch.

posted by joecab at 11:22 AM on April 22, 2006

It's an interesting idea, but it leaves FAR too many games out of the picture. Any game that is won by two runs could possibly exclude all the single RBI producers? So if four guys get an RBI and the team wins 4-2, no one was a "Game Maker"? At least three of them combined are a "Game Maker", so some sort of credit should probably be assigned. Another sign something is slightly flawed is the fact that your stat seems to indicate Tony Womack was the most valuable player on the Yankees last year. Obviously this could not possibly be the case as almost every other measurement out there listed him as the worst player in the American League.


If you are interested in a statistic that measures discrete "clutch" performances, then I'd recommend reading up on "Win Probability". In short form: it measures how a player/pitcher does in every single at-bat. It compares the teams probability of winning before the at-bat to their probability of winning after the at-bat. The probabilities are based on how MLB teams have actually done in these situations (culled from almost 30 years of play-by-play data). Example: Player A is batting in the top of the 9th. Nobody out, nobody on base. Score is tied. Before his at-bat, his team (visitors) have about a 48.3% chance of winning (based on historical data). If he hits a single, then his team's probability of winning goes to 56.8%, meaning he added 8.5% to his team's chances of winning. This is listed as +08.5%. If he strikes out (or flies out), then his team's probability of winning goes to 42.5%, meaning he subtracted 5.8% from his team's chances of winning. This is listed as -05.8%. For the pitcher involved, it's opposite result for him (negatives for the batter are positives for the pitcher). So a lead-off HR in the bottom of the first inning to take a 1-0 lead (+10.0%) is "worth" less than a lead-off HR in the bottom of the 5th inning to take a 1-0 lead (+14.7%). The idea is you sum up all the different occurances and you can get a measurement of how much each player/pitcher contributed to their team's chances of winning in that particular game. For an example, here is the chart and final percentages for the players in last night's Toronto/Boston game. Note: even though Russ Adams had a 2-run home run (and the team won by a single run), he was the worst contributor to the Blue Jays win. That's because other than the 2-run HR (which only changed the score from 6-2 to 6-4, an increase of win probability of only +5.6%), he was terrible at the plate (going 0-5 the other at-bats, for a combined -22.0%).

posted by grum@work at 11:49 AM on April 22, 2006

I was at that game. Great game. Burnett is busted goods, though. Not good.

posted by WeedyMcSmokey at 02:00 PM on April 22, 2006

As i've said before most fans today know little about the intracacies of the game but are rather frustrated statisticians

posted by THE DOME at 02:05 PM on April 22, 2006

As i've said before most fans today know little about the intracacies of the game but are rather frustrated statisticians Actually, most fans know very little about the statistics of the game, and are quite happy to just watch and cheer for their team. And the "intricacies" of the game are usually quite obvious to anyone that is watching the game. Changing pitchers, shifting defences, pitch selection and base running strategies are always visible and simply paying attention allows most people to learn about them.

posted by grum@work at 03:22 PM on April 22, 2006

As i've said before most fans today know little about the intracacies of the game but are rather frustrated statisticians Ever stop to consider why you have to keep repeating this delicious bon mot?

posted by yerfatma at 05:14 PM on April 22, 2006

The one thing I love about baseball is statistics but even I have to admit statistics get pretty wierd on a nationally televised game I mean I don't care what someone's Batting Average is with two outs two men on and he's go three balls and two strikes on him let the game decide

posted by luther70 at 05:38 PM on April 22, 2006

I noticed that when I stopped paying attention to stats I started enjoying the game more. But I drink a lot, so you can factor that in as well.

posted by mr_crash_davis at 07:07 PM on April 22, 2006

if we want a simple version of "clutch" hitting, that doesn't eliminate the HR or any other type of RBI or hit for that matter, let me propose a new statistic. or rather a class of statistic, that certainly could be in the running for what we are tlooking for in this instance. BA, OBP, RBI, RISP while the opposite team is in a save situation. this would eliminate the early and middle inning dilemna, sure a hit could be clutch in the 1st or 5th inning, but delineating when it is and is not is just too subjective. If we take the definition of a save opportunity, and compile batting records for the trainling team at the same time, we would, by extension, have an analogous situation to when a save is a save. Just food for thought, now all are welcome to dine freely at my trough.

posted by elovrich at 08:06 PM on April 22, 2006

Sorry bullpenpro but these people don't seem to agree with you. Of course, the article itself gives nothing on how exactly this guy figured it out so for all we know, the study is absolute bullshit. I am interested to see if they make the study public though.

posted by Ying Yang Mafia at 11:59 AM on April 23, 2006

There are 3 types of lies. Lies, damn lies and statistics. Why would anyone want to be thought of as a clutch hitter? One of the most notable clutch hitters was Reggie Jackson. Granted, there were very few players I'd rather see at the plate than Reggie when the game was on the line. But I'll be damned if I can figure out why this clutch ability is praised as something wonderful. In clutch situations, Reggie was one of the best. Yet his career batting average was 262. As he was such a good hitter in "clutch" situations, he must have been a sub - 250 hitter at other times. Since he could hit so well when he concentrated and gave an at bat 100% effort, why praise him for not giving 100% for all his at bats? The opposite of a "clutch" hitter would be someone like Ted Williams. He treated every at bat like the game was on the line. He knew only how to give 100% all the time. He hit at a 344 clip day in and day out. I'll take that consistant effort every time. I use Reggie and Ted as extreme examples. There are plenty of both types. It's simply my belief that anyone that can hit in clutch situations should be able to hit equally as well in other situations. Being a clutch hitter should be viewed as a negative.

posted by drevl at 01:11 PM on April 23, 2006

drevl; No one has said that to be a clutch hitter means that you only do well in pressure situations, or that you do not hit well ealy in the game or in a blow out. Being clutch should not, and to me does not, mean that you do poorly at other times, it means that you do not let the pressure get to and therefore perform WORSE than you usually do.

posted by elovrich at 02:01 PM on April 23, 2006

Not performing worse than usual is the most liberal description of a clutch hitter as I have ever seen. I guess most players are clutch hitters. There must be a different word to describe Reggie's magical ability to excell at critical times. Clutch (as described as not being worse than usual) is way too mild.

posted by drevl at 03:45 PM on April 23, 2006

I'm really pleased the column generated a discussion this good. I really expected to come back to find three responses, two of them telling me how much Jeter and/or the Yankees suck.. To be honest, looking at this fresh, I'm not sure the Game Maker really measures "clutch" so much as it does overall player value to the team (you may argue they are one and the same -- that's not my definition). The Game Maker basically says if Player X was not in the game, the team would not have won because his RBIs were the difference between winning and not winning. I would argue that the principle difference between the Game Maker and the GWRBI is that the latter is unfairly bestowed upon one hitter in a 4-3 game while a Game Maker is given to every player who drove in a run for the winning side. So, I like it more for that reason. Grum, my gut reaction to the "Win Probability" is that I wouldn't like it, because I have a hard time giving Russ Adams a negative score for a game in which he hit what I would call a pretty critical 2-run homer. I don't think he deserves a huge plus-number, but such a big negative seems wrong. Your reaction to Womack, though, is right on -- I agree that his standing raises serious questions about the viability of the stat. I guess it has to be put in context in the same way a batting average does -- a .375 average from a guy with 8 ABs is not nearly as impressive as a .310 average from someone with 500 ABs. Drevl, I'm not convinced the chief goal of a ballplayer is to become most like a robot who can generate 100% performance with every at bat regardless of the situation. No human should be expected to, or be able to, ignore the varying degrees of importance of their at bats (or play in the field -- defense is clutch, too). Just as even the best closers are practically expected to let down when they are brought into a 9-3 game, you can't expect a hitter to feel the white-hot spotlight of urgency during a 4th inning at bat in Tampa in May. You can't dun Reggie for not being as robotic (or as good) as Teddy Ballgame (though you CAN dun him for being a total ass). From a team-success perspective (as opposed to a personal rooting perspective) I would rather have a "clutch" hitter like Reggie than a guy notably in reverse, like, say, A-Rod. Crap. I just wrote another column. Sorry.

posted by BullpenPro at 11:26 AM on April 24, 2006

I would rather have a "clutch" hitter like Reggie than a guy notably in reverse, like, say, A-Rod. Very interesting. Does that mean you would rather have Reggie than ARod on your team? 162 game average over their careers: .............Reggie..........ARod Hits............148..........193.... ARod +45 HR................32............44.....ARod +12 RBI...............98...........125....ARod +27 BA..............262...........307 OBP...........356...........385 SLG............490...........577 Nothing in the pure statistics shows whether or not all those extra hits, HRs, and RBI for ARod were clutch or not, but you can bet they sure helped his teams win games. I think that matters most of all. In a clutch at bat, I too would rather see Reggie. In the far more frequent non-clutch at bats, I'd rather see ARod.

posted by drevl at 03:53 PM on April 24, 2006

PS to BullpenPro - A Manchester guy shouldn't come down too hard on a Rockville (OK, make that Vernon) native. You know how dumb we can be.

posted by drevl at 03:57 PM on April 24, 2006

Bullpenpro the Yankees suck and Jeter is not the best shortstop on his team. By the way, I hear of a new labtop called the toughbook that just may be able to withstand being thrown out a window.

posted by Ying Yang Mafia at 06:40 PM on April 24, 2006

Drevl: Having to choose between Reggie and A-Rod is like wondering in which eye I'd rather have the ice pick. As has been said many times, when your team is up 9-3 there is no more dangerous hitter in the world than A-Rod. Oh, and don't go back to Rockville. Bullpenpro the Yankees suck and Jeter is not the best shortstop on his team. He IS the best swimmer on the team, but he thinks swimming is for sissies ("Basically, it's actively not drowning...") and thinks it would be much harder to excel at the sport of drumline. Oh yeah, I'm reading you...

posted by BullpenPro at 07:43 AM on April 25, 2006

You're not logged in. Please log in or register.