THE BOOK--Playing The Percentages In Baseball

"Everyone else is out of this discussion.”

I’m surprised to see you put this much emphasis on WPA.  Like Studes (I think I’m citing him correctly), I’d only consider it on the margins to make a tough call.  Just looking at the Fangraphs “OPS wins” as quick measure of offensive production, these players are all ahead of Jeter:  Giambi, Dye, Thome, Manny, Hafner, and Mauer.  And I’m guessing Santana also outperformed him on a runs-saved basis. There’s certainly a serious case for Jeter, but I sure wouldn’t rely on WPA to settle the issue.

“A-Rod has 121 RBIs.  In 10 years, no one will remember how empty those RBIs were.”

Let’s hope that’s right.

And in 10 years, the Yanks’ 97 wins will be attached more to ARod than Jeter.  Any Yankee fan still alive would never believe it.

In terms of “backward-looking attribution of performance to team wins”, I look at WPA, fielding, and repl-levels only.  (Ideally, we’d have fielding with LI.)

Tango, I’m confused by your post.  Your clutchiness list is different than Fangraphs’.  How are you defining it?

Guy, you’re citing me correctly.  Tango is a bigger advocate of WPA as a primary stat for evaluating players’ contributions than I am.

There’s different ways to look at “clutchiness”.  Is it simply a guy who performs better when the pressure is on, than overall?  Or, is it a guy who is a good hitter to begin with, and performs even better in key situations?

For example, if some crappy hitter did what Jeter did in the clutch, and went from -3 wins in context-neutral, to -1 when considering leverage, is that being “clutchy”?  So, my requirements for being “clutchy” is a good hitter, that hit better in crucial situations.  Pujols, Jeter, and Ortiz are all great hitters, having great seasons, and performed unbelievably in key situations.  Clutchiness.

I’ll remind everyone to remember Stephen Colbert’s truthiness.

Question for Tango on WPA and MVP:

If 2 batters had the exact same WPA, but one guy hit .300/.400/.500, while the other guy only hit .280/.350/.440 but did better in the clutch--would you consider them a tie for MVP?

Same WPA in the same number of games (and fielding is a wash) = tie.

Question for you: if two teams win 97 games, but one team scores 163 more runs than it allows, while the other is 103 runs (and their strength of schedule was equivalent), did they both win 97 games?

Or, if one team scores +44 more runs than they allow, and another is at +51, does the first team with their 93 wins deserve to make the playoffs ahead of the 80-win team?

Tango, that’s fine.  But I would point out that you’ve just given a word a different definition than has been used elsewhere.  You may want to use your own labels for things, to avoid confusion.

Without rehashing the whole WPA thread we did a while back, I think Tango is missing a key distinction.  WPA does not = “real wins.” It tells us how much a player’s hitting increased his team’s CHANCES of winning AT THE TIME he hit.  If you look at Fangraphs, you’ll see games there nearly every day where the WPA totals do a poor job of assigning credit for that game’s final outcome, given complete after-the-fact knowledge. 

I have no problem with considering a player’s total actual performance, including his good luck, in selecting an MVP.  But WPA is only one tool for assessing that, and not always a good one.  I’d nominate Jason Bay, whose WPA (1.64) is lower than Freddy Sanchez’s because he apparently got his hits at the “wrong” time, as the 2006 don’t-get-carried-away-with-WPA poster boy.

Actually, I started using the word “clutchy” (or was it clutchiest-clutch?), and then Dan Smith coined the term “clutchiness”.  My plea to remember Stephen Colbert’s truthiness word seems to have been missed.  Truthiness is whatever truth you know is in your guts, not books.  The entire idea of clutchiness is very tongue-in-cheek, it’s the very difference between the intelligent Stephen Colbert, and the character Stephen Colbert.  To paraphrase the character Stephen Colbert, to debate the definition of clutchiness is to miss the concept of clutchiness.

***

Guy, right, “at that moment in time”.  That’s all I care about.  That’s all fans care about when they watch the game.  Armchair quarterbacks look at the game after-the-fact.  Vlad’s grand slam in the playoffs a few years ago against the Redsox had, to me, at the moment I was watching it, the exact same value, the same pulse rate change, the same blood pressure change, the same emotion, whether the Angels lost the game or not.  After-the-fact, that grandslam might as well have been an out.  After-the-fact, the Dent and Fisk HR live forever, because their teams won that game.

"Guy, right, “at that moment in time”.  That’s all I care about.  That’s all fans care about when they watch the game.”

I have to disagree.  I think fans care about winning.  Exciting is great, but crushing your opponents is also fine.  Winning is what matters. 

When they establish an MEP award, for Most Exciting Player, I’ll gladly give it to that year’s top WPA player.

Right, the Rangers +51 means little to the A’s +43 in run differential.  WPA is the only one that actually cares about winning.

Fans caring about winning means that they care about Jeter more than A-Rod.

As for caring, are you say Angel fans would have cared more, for Vlad’s grandslam, if the Angels won?  The only way to reconcile your position is to only count performances in team wins, and discard all performances in team losses. 

I’m not against that viewpoint.

But, I am against the viewpoint to give ARod credit for Yankee wins in blowouts and in losses, when the leverage was much lower, and to discount Jeter’s performance in close games.

The only way to reconcile your position is to only count performances in team wins, and discard all performances in team losses.

I’m not against that viewpoint.

***

Really? Then what happens when Mariano Rivera ends up being the most valuable player every year for the past decade? You’d have to disregard all starters and closers, because they almost always pitch in wins (well, the good starters anyways), and players from good teams would still end up with an advantage over players from bad teams. Such a system would be un-manageable…

Anyways, Tom, I think you’re not selling WPA correctly. It is not an “excitement” statistic; it is a statistic that acknowledges that a player can only control how he performs in a given situation—and not what comes after.

I’ve sold it every which way but loose.  If they ain’t buying, they ain’t buying.

The WPA graphs shows exactly what happens during a game.  Each change is given to the player.  Those graphs coincide with a fan’s emotions and appreciation of the game as the events unfold.  Nothing complicated about it.  It’s about as simple as simple.  You either drink the WPA juice, or you don’t.

According to VORP, Millwood was slightly better than F Rod this year. Millwood had a 0.87 WPA and F Rod had a 5.39 WPA. I’m not sure what replacement level WPA would be for these two pitchers, but does anybody think these two had an equally good year?

As a rule of thumb, add +2 wins for a starting pitcher, +2 for nonpitcher, and 0 wins for a reliever.

Remember, the key here is leverage. Rodriguez had a 2.12 LI, so his non-leveraged WPA would be 5.39/2.12 = 2.54. Millwood’s LI was .93, so his non-leveraged WPA was .87/.93 = .94. Using Tango’s handy rule of thumb, that would make Millwood’s 2.94 wins above replacement and Rodriguez 2.54. That’s a difference of about four runs. VORP shows a difference of about one run. Close enough.

0.4 runs, of course.

As far as I knew, VORP doesn’t take leverage into account when evaluating pitchers. So shouldn’t evaluating by WPA lead to a different result?

About the only thing that I am fairly clear about is that c ontext neutral stat is a poor measure of MVP performance.  After that, you can make a case for just about any context non-neutral metric you want, including R/RBI.

The only problem with completely discouning context-neutral stats is that the voters are NOT going to remember all of those clutch performances.  They WILL, however (many of them at least), look at a player’s stats at the end of the year and some of those stats they will be looking at are context-neutral or at least somewhat context-neutral (like BA).

"0.4 runs, of course.”

.4 wins = 4 runs.

“As far as I knew, VORP doesn’t take leverage into account when evaluating pitchers. So shouldn’t evaluating by WPA lead to a different result?”

And it does. Using Tango’s rule of thumb, Rodriguez is worth 5.39 wins above replacement, and Millwood is worth 2.87. If we remove leverage from the equation, their worth turns out to be about equal. That’s what I was demonstrating.

The last closer to win a Cy Young was Eric Gagne in 2003. According to VORP he was 33rd in the majors and F Rod this year was 49th in the majors.

With the discussions lately about replacement level pitchers, chaining, etc. it would seem that closers should never even be considered for the Cy Young.

I guess who deserves the Cy Young turns on the question of whether leverage should be a factor when evaluating pitchers.

"About the only thing that I am fairly clear about is that c ontext neutral stat is a poor measure of MVP performance.  After that, you can make a case for just about any context non-neutral metric you want, including R/RBI.”

It seems to me there are at least three different kinds of “context” to think about:
1) the quality of a player’s teammates
2) the relative importance of the player’s PA/IP (measured as leverage in WPA)
3) how well the player performed in important situations (vs. less important situations).

I don’t think any of us believes #1 should affect our MVP vote.  And to some extent, stats like R and RBI are affected by that.  So you do want to strip out some context.  I think Tango and DSG would say we do want to capture both 2 and 3.  I could go either way on #2 and agree that #3 should be included.  But I don’t share their confidence that WPA measures those things correctly, in a post-hoc world.  And I’m not just talking about the fact that one gets WPA credit in losses.  If A-Rod hits a 3-run homer to “pad” the Yanks’ lead to 7 runs in the fifth he gets relatively little credit, even if the Yanks end up winning by one run.  A player who hits a no-out triple but doesn’t score gets more credit than one who hits a 2-out triple and does score (same inning/score).  A starter will get vastly different ratings for two 7 IP/0 ER outings, because of how many runs, and WHEN, his teammates scored (my sense is that WPA is least useful when rating starting pitchers).  Etc.

Personally, I’d be more interested in using run advancement (as Ruane did) than win advancement to judge MVPs.  It picks up a lot of what Tango wants in terms of “clutch” performance, but leaves out the game state impact which—to me --creates most of the problems.  It would be great if THT or Fangraphs provided run advancement stats.

There is certainly some merit in treating the outcome of the game as a variable.  That is, if a game ends up 7-6, after starting 7-0, then “every run is important”.

You do have issues to work out though.  Imagine the game is 6-6, and a guy hits a grand slam in the 9th, to win 10-6.  Or the game starts 10-0, and ends up 10-6.  What then?  That 6-6 game in the 9th might as well have ended 7-6, since all the extra runs on the HR meant zero.

Then you have the issue of closers.  Again, the reason that Smoltz, Papelbon, Rivera, et al have value as relievers is because managers can leverage their skills to maximize wins.  What then?  Do you just treat their 80 IP as if they were randomly distributed?

The great thing about WPA is the clarity of what it is measuring (same for R and RBI), and how it goes about it.  (WPA is clearer than ER for a pitcher!).

Maybe David can stop by and explain AWP.  I remember it was a good alternative.

Actually, I started using the word “clutchy” (or was it clutchiest-clutch?), and then Dan Smith coined the term “clutchiness”.  My plea to remember Stephen Colbert’s truthiness word seems to have been missed.

Good grief, Tango.  Not missed at all.  You’re using a parody of a labeling concept when I’m arguing for a consistent application of a label.  Following Colbert’s example in the extreme undermines good, consistent communication and, in the end, undermines getting concepts like WPA accepted by anything close to the mainstream.

The title of the blog is called “Clutchy, Chokey”.  I don’t think anyone’s used those words.  In the blog entry, however, I did use: “The Clutchiness Hitters of the Year, The Chokiness Hitters of the Year”.  I suppose I should have used “The Clutchy Hitters...”.  In any case, even Dan Smith, coiner of Clutchiness, has posted the different kinds of Clutchiness leaders, depending on whether he was above average or not.  His site even highlights such players, since he’s obviously not comfortable giving someone a “Clutchiness” award, if he’s an overall below average hitter.

There’s no reason for Clutchiness to make it to the mainstream, since most of Clutch is not even real.  That’s why I have as much fun with it as I do, and I really don’t care how Clutchiness is perceived.  That’s why I like the word clutchiness!  It’s a “I can say whatever I want, and it’s true” word.  It’s great!

I don’t see WPA breaking into the mainstream or not, conditional on how I define or use the word Clutchiness, Clutchy, or Clutch.  When all is said and done, the career Clutchiness scores of most hitters will be near zero.

For example, one of this year’s Tango’s Clutchy Award winner, Albert Pujols, had a Clutchiness score of -3.69 wins from 2002-2005.  His astounding +2.67 Clutchiness wins still brings him to -1.02 wins.  Jeter was zero in 2002-2005 coming into 2006.  Ortiz was a minus from 2002-2004, leading into the unreal 2005-06.  A-Fraud was a plus leading into 2006.  Jason Bay on the other hand is now -4.14 wins.  Manny may be the Chokey hitter of 2002-2006.

All to say, I’m not interested in taking any of this Clutchy stuff seriously.  It’s pure fun.