THE BOOK--Playing The Percentages In Baseball

I wonder how often there will be a difference between the best hitter and the best clutch hitter?

Looking great!  If you don’t get much Reds traffic, let me know and I’ll try to stir up some more.  I don’t know if there’s anyone with a more “unclutch” reputation than Adam Dunn, at least among local fans, so I’d think lots of folks would be ready to vote “against” him, no matter who they’re voting for.

...while this won’t really matter (no one’s going to pick him), Corey Patterson is probably going to be the Reds’ centerfielder for at least half of this season.  Might be worth adding him if that can be done without difficulty.

Also, when’s the deadline for voting?  Prior to opening day?

Cheers,
Justin

I would prefer it be opening day, but I’ll take what I can get.  I actually hadn’t realized the date until last night.

Pujols, Ortiz, and Vlad are for sure double-dippers and will be excluded.

I’d like to see the Yanks fans choose Jeter over ARod.

The player list was exactly the 40-man roster on MLB.com as of last night.  It’s too much trouble to change, plus I doubt anyone would pick Patterson over Junior or Dunn.

Already 230 ballots, and it’s been just 6 hours.  A big portion is SOSH.  I’m just waiting for USSM to hear about this, as it’s always a race between those two boards as to who are the more rabid fans in my polls.

Any way to get a link from BTF and also have Neyer give it a mention in his blog?

You/we can try and get a bunch of team blogs to put it on their sites. I bet you BP’s unfiltered would generate a ton of votes as well, if they’re willing to put it up. Try sending it to whoever is in charge of the MVN blogs.

I have a list of blogs that are friendly to my projects.  They typically post it on their own.  Otherwise, I’ll ask them on Monday. 

I already have 518 ballots cast in under 24 hours (223 from Redsox fans).  Here are the bottom teams:
fla 5
pit 5
tex 5
kc.  4
mil 4
oak 3

***

Yankee fans are crazy: they actually have Jeter, overwhelmingly over ARod.  ARod had as many “GREAT” votes as “POOR” votes if you can believe it.  Jeter is 18 to 1 in that respect.

The clutchers are making this too easy for us.

What do you mean “bottom teams?” I do not know what you are going to do with the ballots?  Can you explain? 

I thought you were going to have each person choose ONE person from each team, or as many teams as they wanted, to be their best clutch hitter and you would match that hitter with your best overall hitter and then you would compare the two in clutch situations.

But you have people being able to rate each player.  I don’t get it. Why wouldn’t Jeter, for example, rate as a great or good clutch hitter, since he is a very good hitter overall?

Also, you need to explain what metric is going to be used to rate the players in high leverage situations.

But, as I said, I have no idea how you are going to handle the data given how you set it up. I am not sure that 4 out of 10 people who go to the site can do that either (figure out what you are doing).

I re-read your instructions, and I still don’t get it.  You are going to take the consensus best clutch player from each team?  The one with the most clutch votes?  And then match them up with your 30 best hitters?

I think you should have had each person pick ONE player from each team, not be able to rate every player.  That is going to screw up the data, I think.  What if everyone thinks that Erstad is a great clutch player, but not as good in clutch situations as another player from his team.  He is going to show up as great clutch on everyone’s ballot.

You tell people you want them to pick the ONE player from each team that they would want at the plate when the game is on the line.  O.K.  Then you allow them to rate each player on a sliding scale and choose any number of players on a team.  Huh?  I don’t get it and I don’t think they will get it and you are NOT going to convince anyone that the fans chose the clutchiest player on each team, for the reasons I already mentioned.

I cannot support this project the way you have it set up.  Sorry.

An example of what I am talking about is that you are forcing people to vote for A-Rod and they are voting him as non-clutch.  Some of these people are thinking that yea, he is a great hitter, but I don’t think he is clutch (maybe he hits .900 OPS in clutch and .950 in non-clutch).  He and players like him (great hitters who are not considered clutch) will get shortchanged.  And conversely, mediocre or even poor hitters who are generally considered as clutch hitters are going to get a ton of “SUPER” votes, even though people do NOT think that they would rather have that player at bat over any other player on the team!

If you left it at choosing one player per team only, A-Rod would simply be left off a lot of Yankee ballots.  However, some people who don’t consider him to be clutch would still put him on their Yankee ballot.  Their thinking would be, “Yeah he is by far and away the best overall hitter on the Yankees, but he is not so great of a clutch hitter, buy probably still the best in clutch situations because he is so good anyway.”

By forcing those people to rate A-Rod, you are precluding him from being the pick from the Yankees.  Those same people who might still want him as their hitter of choice in a clutch situation, but don’t think he is that clutch, compared to his overall talent, are NOT going to rate him as SUPER clutch in your sliding scale.

You set this up so that you can’t possibly lose and it is not a fair fight.  You really need to change how you do this.

Your initial premise was right, since I said:

“Go through your favorite team’s roster, and select the one GREAT clutch hitter, the one guy you want at bat, when the game is on the line. One selection will do.”

I also said:
“Note: select the guy you want at bat, when the game is on the line, regardless of how good or bad he is otherwise. So, even if you believe that Darin Erstad or Willie Bloomquist can step it up a notch or two with the game on the line, they’d still be several notches below the other players on your team. Unless of course you really think otherwise.”

So, to me that’s very clear.  It’s whoever you want at bat with the game on the line.  That was your original thought as well.

What seems to have thrown you off was the sliding scale.  Since some people provided their email address, I will ask them if they understood the implication of their Jeter/ARod choices.  That even if they think that ARod loses a notch as a hitter in pressure situation, and Jeter raises a notch, do they believe that this means that they’d still rather have Jeter than ARod.

I’m thinking that the Fans are going to say “yes, I understand that, and that’s what I believe.”

I updated the instructions, as well as each team page.  Let’s see if the Yankee ballots that come in will change in their selections.

I will send out the good word on the MVN list-serv.  I might even be able to put out a plug on a StatSpeak to pull in a few more folk.

To remove any possible ambiguity, I put this on each team page:

“When the game is on the line
this hitter’s performance
relative to his teammates
will be...”

Clearly, if someone still sets Jeter to GREAT and ARod to POOR, then he believes that ARod is the worst choice on the team, regardless of how ridiculous that statement sounds to some of us.

No one would be able to interpret that as ARod compared to himself in normal situations.

6 yanks ballot since my new instructions: 5 prefer Jeter over ARod, and one has them even.

It seems clear to me that the average fan understood pretty much my original instructions and intent: they simply prefer having Jeter over ARod in clutch situations.

***

Now at 600 ballots.

Tango--

One of those ballots for Jeter I had my dad fill out (who hates anything to do with numbers). I asked the same question to him (who would you want up with the game on the line?), and he instantly responded with Jeter. I haven’t done one for myself yet, but I’ll be going with a-rod when I do (in about 5 minutes).

Yet another Yankee ballot, and once again, another Jeter/game-on-the-line-guy vote.  I don’t think there’s any ambiguity at this point.

Tango, I don’t disagree that most if not all voters/fans will say that they clearly understand the instructions.  But, I still think that the way you set it up taints the process.

Why did you not just let them choose one player per team, as you imply (more than imply) you are goign to do in the instructions/introduction? 

Actually, the way you have you have it set up is great if you are going to compare at the end of the season, and during the season, the difference between each player’s clutch and non-clutch (or overall) stats and then compare that to the average rating on that player, but you CANNOT have everyone potentially vote on every player and then use your method to evaluate the fans.  It just won’t work.  The fans can’t win.

When you do a survey, the wording is more important than what people “think” they are communicating.

Just as another example, I assume that some of the voters had “Super” for several players on one team.  So how would you know what player they would want at bat for that team?  I realize that you are aggregating the results, but if ANY fan has “Super” for more than one player on a team, which I assume some of them do, that tells you that they do NOT think that they are voting for ONE player on each team.

At the very least, even though the instructions are very clear, they don’t jive with the format.

Nope, I thought about this a lot, and I respect and admire the effort, but I cannot support it.

Yet another Yankee ballot, and once again, another Jeter/game-on-the-line-guy vote.  I don’t think there’s any ambiguity at this point.

Unless my web page is different from everyone else’s, that is not a vote for “Jeter game-on-the line.” Those fans simply said that Jeter is a “Super” clutch hitter.  That certainly is not the same thing as “voting for Jeter as your number-one guy.” How could it be when they could have also rated one or more other players on the Yankees as “Super?” Heck, I would probably say that Jeter is a super clutch hitter, but that does NOT mean that I would prefer him over A-Rod.  How could it be?  Again, how can you ask people to “vote for their number one guy on each team” (which is VERY clear) and then turn around and have them rate every player on the team AND give them the opportunity to choose “Super” for more than one player?  That makes no sense.

Is there another site/ballot that I am unaware of?  Seriously.

I mean, if I am the only one who thinks this way, I’ll be quiet (even though I still think I am right).  Your original idea was perfect (simply vote for your ONE go-to-guy on each team, or as many teams as you want).  Why did you choose this “rate each player, as many players as you want, on a sliding scale” thing?

Well, most fans selected only one GREAT player.

Worst case, you can discard those fans that selected more than one GREAT player.  My intention was to give those fans half votes if they selected 2 players or third votes if they selected 3 players.

If let’s say you have 30 yankee ballots that choose Jeter as the GREAT, and another 10 that has Jeter and Posada as double-voted for GREAT, and 5 others with AROD alone as great, then what does it matter as to your particular objection to the formatting?  It’s clear, isn’t it, that in this instance, the fans want Jeter?

Let the process play itself out, and then you can see the results as to who they ballotted.  You are presupposing too much what people are or are not doing.  Basically, any of your objections can be addressed, if you just wait for the results to come out.

Look at the heading:

When the game is on the line
this hitter’s performance
relative to his teammates
will be…

If Jeter gets 20 or so GREAT out of some 40 or so ballots, what does that mean to you?

And if ARod gets just a handful, what does that mean to you?

We’ll just have to disagree on this one as you obviously have no interest in changing the format, which it is probably too late for anyway.

None of your arguments is incorrect.  It is just that you had two choices of format.  One was exactly what you were looking for and the other was not.  You chose the latter and are now defending it.  Not defending it over the former - just defending it in and of itself, which is fine.

Again, I ask (which you have not answered), since you were very clear what you wanted the fans to do, which was to CHOOSE THE ONE PLAYER PER TEAM THEY WOULD WANT AT BAT MORE THAN ANY OTHER, did you not let them do exactly that, which would have been to list all the players on a team and have them choose one and only one?  What point would there be for someone to choose Jeter as Super, and then to rate any other player, other than to confuse the voting and the issue? 

And just the fact that you got at least one ballot with more than one player rated as “Super” proves unequivocably that not all of the voters understood the instructions, which is understandable since the instructions and the format do not jive - but I already said that.  Why don’t you send an email to everyone who voted for (rated) more than one player (which is probably everyone) and ask them, “Why did you vote/rate for more than one player when in the instructions I asked you to choose the ONE player you most want at bat?  And then to the voters who voted “Super” for more than one player (and I’m sure you will get plenty of that), why don’t you email them and ask them, “Why did you vote “Super” for more than one player.  How do I know who you would want at bat for that team?” The response from them would HAVE to be, “Well, it was your ballot.  It was not very clear what you wanted me to do, other then rate every player, I guess, which is exactly what I did!  It is your fault that in rating every player (which is exactly what I did), that I did not necessarily answer the question that you posed in the instructions/introduction!”

I have to agree with MGL on this one. When you set this up I thought we were going to choose one player we want with the game on the line… not rate how each player does in the clutch. IMO, it should be a drop-down menu where we click one player and then click submit.

I’ll reiterate that if you want to leave the ballot as is, you should simply look at throughout the season how each category of player (using the average of the fans’ ratings for each player) does in clutch minus non-clutch or clutch divided by non-clutch, whichever is supposed to be more constant (I don’t know if that has ever been settled).

It does not have the “fun” aspect of the “I’ll pit my players (the best) against yours (the clutch),” but I just don’t see how you can do that with the ballots you have now.

Doing it that way is a lot better than the original way, anyway.  I mean if the clutch/non-clutch differential is exactly the same for all categories, we can pretty much throw clutch out the window, at least as far as knowledgeable fans being able to predict clutch ability, no?

Either format works.  One may be superior to the other.  The key is if either is acceptable.  And both are.

What am I supposed to do with the Angels?  All the Angels fans would select Vlad obviously.  So, I need a second guy.  Ok, you will then say “second choice” as a dropdown, which is fine.

What if Torii Hunter is the second choice for say 7 of them, but there’s a whole bunch of them that would say “no way! never!”.  If I have 7 Torii as GOOD and 6 as POOR, and then I have Garrett Anderson as 5 GOOD and 0 POOR, then GA might be the better choice.

In my version,I have more data, and it gives me more choices.

In the alternate simpler version (1st choice, 2nd choice), it is clearer, but might deliver what i might need.

In the end, you have two things:
1. Regardless of format, you will end up with very similar results (one is simpler, the other more flexible).  It will certainly not invalidate the study.

2. As long as either format is satisfactory, we don’t have to justify one over the other.

Like I said, if you feel that the double-voted ballots are invalid, then don’t count them.  They’re handing chads.

Wait for the results, and then see if there’s an issue.  What we have here is a philosophical issue that in all practical sense, will be irrelevant in practice.

"but might deliver what i might need.”

should be:

“but might NOT deliver what i might need.’

Judging from the comments so far on BTF, people (in general) or at least a significant portion, don’t give a rat’s ass about a study which suggests something that goes against their belief system.

And of course this is not a “study” examining clutch hitting ability.  Those have already been done.  This about a “real world test” to challenge the notion that baseball is not played in your mother’s basement.  Bottom line is that if 78% of the people think the world is flat, no matter what evidence you present that it isn’t, only 3% of the 78% will EVER change their minds.

People also don’t realize that the fans are actually a great source (probably the best or at least one of the best) of baseball wisdom.  A comment on BTF suggested that the average response you get will be “silly fan-boy baseball-speak” or something like that.  A lot of people don’t realize that it is quite the opposite, as evidenced by the fan player forecasts and the fan defensive scouting reports.

If you really wanted to do a (rigorous) “study” you take the ballots and the data you have and do exactly what I said.  Look at the clutch/non-clutch differential between all categories of players (based on how the fans voted).

If you want to do it the “fun” and simple way (which was the original plan, though not necessarily the best one), you simply have each person pick ONE player from each team, with no second choices.  Then you use the consensus player from each team (the one with the most votes) according to the fans and match him up against your best player.  So what if they are the same player on ANA, STL, and BOS (I would guess).

You either still use those players for your “test” or you only use those players and teams where your players are different from theirs and see who wins among those.  That is probably the best way to do it.

MGL/27: If the responses at BTF are that, then that’s disappointing.  It’s as if they have no idea as to how well the fans who vote on my polls are serious knowledgeable fans.  As MGL said, they line up pretty well on fielding, and fantastically well on forecasting (enough that the combination of rate stats from fans, and their depth chart would make the Community Forecasts the best in the business; their lack of knowledge of minor league players is tempered by the fact that very few minor league players will make any kind of impact as a rookie).

MGL/28: I don’t really have a disagreement with that post.  “Fun” though (head-to-head) has its fun-side of course.

I didn’t have a big problem with the comments at BTF, and without needing to rehash what has already been said here, I responded:
http://www.baseballthinkfactory.org/files/newsstand/discussion/the_book_blog_tango_the_great_clutch_project/

Actually, statement B is:
“I’d rather have Junior than Dunn”
“I’d rather have Jeter than ARod”

Those are the statements that are being tested. That someone worse than someone else will somehow be able to rise above that guy (or the other guy fall; or some combination), when the game is on the line.

When I did my ballot for the Jays, I was surprised to see all of the players with the various scales to rank. However, it was clear to me to choose only one player for “great”. The rest I chose from the other options. I put a link at Battersbox to do the ballot. Someone should put one up at the Athletic’s Nation site to get them to participate. Good luck, Tom.

This is purely a SAMPLE page.  Here’s what David at Fangraphs has cooked up:
http://www.fangraphs.com/clutch.aspx

Pretty sweet.

I think it would make sense to chart each team’s highest rated clutch player against himself this year too, in the same way.  With 30 or so guys, that should be a good enough sample that you won’t get wildly unpredictable results one way or the other.

I still think you’re preaching to the choir - I don’t see anyone’s mind being changed by the results here.  But adding in a traditional clutch analysis would seem like a fun, worthwhile thing to do given that you have already collected subjective ratings of “clutch” from the fans… plus it’s a way to get the Vlad and Ortiz club members included as well.

Love the fangraphs page.

I’m pretty sure the fans are going to choose Junior over Dunn, and I know they’ve got Jeter over ARod.  I can’t wait to see the other teams. There’s no choir yet.

The comparing to their non-clutch performance is what’s in The Book, and everywhere else.

Since my revised instructions, 10 prefer Jeter over ARod, and 3 prefer ARod.

Here are the current counts, after 818 ballots, as the guys people who with the game on the line:
teamId Count Player
bos 253 David Ortiz
nyy 26 Derek Jeter
stl 26 Albert Pujols
bal 19 Nick Markakis
chc 17 Aramis Ramirez
nym 15 David Wright
sea 15 Ichiro Suzuki
cws 14 Joe Crede
ana 12 Vladimir Guerrero
atl 11 Chipper Jones
cin 6 Scott Hatteberg
tor 6 Matt Stairs
la.  6 Russell Martin
phi 5 Chase Utley
cle 5 Victor Martinez
pit 5 Jason Bay
was 5 Ryan Zimmerman
tb.  4 Carl Crawford
tex 4 Michael Young
oak 4 Jack Cust
min 4 Justin Morneau
hou 4 Lance Berkman
fla 4 Hanley Ramirez
det 4 Magglio Ordonez
col 4 Matt Holliday
mil 3 Prince Fielder
sd.  3 Adrian Gonzalez
sf.  3 Bengie Molina
kc.  3 Billy Butler
ari 2 Orlando Hudson

Half of those teams will have to go to the number 2 choices for both of us. 

Clearly, the first thing to note when someone says “who do you want with the game on the line”, half the responses will be “my best hitter!”.  It’s an obvious thing of course, and one we should expect.  But, so often, when you hear about clutch, that is ignored, not so much as if it’s so naturally implied, but as if its forgotten.

***

That list is the current leaders, but that doesn’t mean those are the clear leaders.  Reds fan for example are doing their best to select anyone other than Dunn.  Here are their totals:
6 Hatterburg
6 Javy Valentin
5 Dunn
5 Encarnacion
4 Junior
2 Phillips

***

Braves fans are going to give me trouble, and this is a reason that just having “pick one guy” or “pick one guy, and give me your 2nd choice” doesn’t work well enough.  The top 2 picks for the Braves fans are Chipper, and then Teixeira.  Those are also the top 2 for Marcel in that order.  Now, I need to go for a 3rd guy.  Without the extra informational ballots as I have them, the decision would have rested on the one Brave guy who didn’t select Chipper or Teixeira as the guy they want with the game on the line.

***

Redsox will give me a similar problem.  Virtually every single ballot selected Ortiz or Manny, in that order.  And that’s how Marcel has them too.  Once again, the “select only one guy” would force me to go with the two guys that selected Yuke or Lugo or Ellsbury, thereby throwing out the other 250 smart ballots that put Ortiz and Manny on top!

This is why the “select only one” doesn’t work.  The guys smart enough to select Ortiz gets their ballots thrown away, since they cancel out Marcel.  The next set of guys smart enough to select Manny gets their ballots thrown away, since they cancel out Marcel for #2.

Clearly, what you want is each of those people #3 picks.  That’s 250 ballots right there.

Otherwise, what I’m left with is about 10 Redsox ballots who were follish enough to not have Ortiz or Manny as #1. Asking 250 smart Redsox fans to vote, so that it forces those names away from Marcel, and have no other say in the matter, while leaving the choice to the 10 foolish fans is not my idea of getting the best possible choice here.

While I’m not saying that the “select only 1” and “select runner up” is not without its merits, clearly there are advantages to both systems.

Just noticed.  The Sox clear #3 clutch hitter is Mike Lowell.  He had 14 “1st place” votes to Manny’s 28, but has a total of 140 “1st or 2nd” place votes, to easily outpace the rest of the non-Ortiz/Manny group.  Checking quick, it looks like he’ll face off against Youkilis.

This is exactly the kind of matchup I was envisioning, that someone (Lowell) clearly worse than someone else (Yuke) will be the Fans’ pick with the game on the line.

The same situation will repeat itself with the Reds, as Dunn seems to be hated by Reds fans.

Looking at Cubs: Ramirez, inferior to Derrek Lee, has a sizable lead as the Fans’ choice.  So, that’s another one that seems to be in the bag.

A close one shaping up is with Indians, as Victor Martinez is being preferred to Travis Hafner.

Joe Crede over half the team?  That’s what Whitesox fans believe.

Tigers want Sheffield or Ordonez over Cabrera (just barely though).  Hard to blame them in this case.

Berkman cancels out with the Astros, leaving the Fans with Carlos Lee a shade ahead of Marcel’s Hunter Pence.

Phillies want Chase Utley over Ryan Howard, though it’s hard to blame them, since they are virtually tied according to Marcel.  Fans will get Utley.

I may have to disqualify the Mariners, since only one ballot came in without Ichiro as #1, and I have a 5-way tie for #2. I’ll post at Lookout Landing later on today.

With the elimination of Pujols, Cards fans are going with Molina right now!  Yowza.

Responding to a Sox fan:
http://sonsofsamhorn.net/index.php?showtopic=29339

Right now, your #1 and #2 picks are Ortiz and Manny, which are also the #1 and #2 picks of Marcel the Monkey.  So, we’re going to our #3, and that’s Mike Lowell v Kevin Youkilis.

Marcel thinks that Youkilis is a much better hitter overall.

You guys, while you may believe that Yuke is a better overall hitter than Lowell, have decided, that when the game is on the line, and you have the free choice of Yuke or Lowell, have gone for Lowell.

In order to match clutch relevant for practical purposes, Lowell needs to as good, or better than Yuke.

I am in firm agreement that clutch hitting does in fact exist.  But, it’s such a small skill, that it can’t possibly make up the gap between an average hitter (like Lowell) and a good hitter (like Yuke). 

The clutch supporters however are very clear here: they want Lowell.  Therefore, they are saying that Lowell will hit as well as Yuke.

Obviously, one team tells you nothing, which is why we are doing this for a collective 30 teams.  Reds fans are prefering practically anybody to Adam Dunn, the monkey pick.  Joe Crede is the preferred choice of Whitesox fans, incredibly enough.  Victor Martinez over Travis Hafner.  Jeter over ARod for darn sake!

These are all the choices that Fans are making, that they believe that the clutch skill is so strong, that it can make up for the sizeable gap in otherwise overall talent.

***

If on the other hand, all you are getting for this is that Lowell’s OPS goes up 10 points while Yuke stays flat, and Jeter’s OPS stays flat while ARod’s goes down 20 points, and on and on, what does that mean?  Well, that clutch exists, but it’s such a small thing, that it can’t be the determining factor, that the player’s overall talent must rule the day.

***

And that’s all I want from this process: for fans to accept that even if clutch exists, at the end of the day, a player’s overall talent level will be by far the determining factor of who you want at the plate.

***

Btw, as a sanity check and to quell any misgivings this poll may leave in some of your mouths, I will post the head-to-head players in 30 polls at Baseball-Fever (team-by-team), and see if they agree that those players are the ones that represent their choice over Marcel’s pick.

Basically, what we’ve got here are the delegate picks, and we’ll go to the convention to ratify those picks.

A Molina is also leading the Giants.

***

I just noticed that Marcel has Ibanez ahead of Ichiro.  So, that’s another good test case.

Looking at the Blue Jays, and this is another good example of having the fans put in as much information.  It’s shaping up as a 4-way tie between Matt Stairs, Vernon Wells, Alex Rios, and Frank Thomas (all with 5 or 6 votes).  However, 5 fans put Wells as their POOR player in game-on-the-line scenarios, while Matt Stairs got no such votes.  Clearly, if it’s between these two, the Fans really prefer Stairs over Wells.

Anyway, don’t take these comments as bashing mgl and those who don’t support my ballot process.  I have a good feeling as to what would work for me, even if I can’t articulate it.

***

Now at over 1000 ballots.

----"Ramirez, inferior to Lee, has a sizable lead as the fan’s choice.”
_______________________

I’m not sure how much inferior to Lee, Ramirez actually is. Lee will project with a higher OBA, and Aramis a higher SLG, I believe. Overall, there’s not much difference.

But, it’s quite clear from watching them everyday that Ramirez is geared, or gears himself up, for RBI situations, while Lee is pretty much a ‘neutral’ type batter. And also, his skill profile is more SLG vs less OBA.  And is it not the case that ‘RBI situations’ has a good correlation with ‘clutch’ situations, according to the common viewpoint?

The problem for me with the clutch thing is that not only is the ability, if it exists, so small that it takes a few thousand (or whatever) PAs to spot it--but that even then, clutch is a ‘dependent’ situation. If, given the same overall stats, you are relatively good in the clutch, it means that you are not so good in ‘setting up’ clutch situations. Statistical analysis which looks at clutch vs non-clutch in isolation fails to include enough of a temporal, or dependence, factor, IMO.

And that totality, to me, is “much ado about nothing”.

Well, I think you (Tango) are setting yourself up for massive criticism by eliminating all the ties between the clutch and the Marcel.  Why are you doing that?  Why not just accept that those are the players who the fans view as [clutch plus overall ability] and leave it at that?  So what if you have some of the same players on your teams?  You can still compare the overall performance of both teams.  People are going to excoriate the whole “study” if you leave out fan picks like Ortiz, Vlad, Pujols, etc.  Whether they are right or wrong in your eyes, they are going to invalidate the whole think because of that, and I think rightfully so.  You can’t just throw out several of the fans’ picks, even though you think you have to.  Especially players whom the fans love.  This study is for THEM and not for YOU.  As I said in one of my previous posts, why not just leave all the picks alone or simply create the teams based on only those picks that are different. You are making this WAY to complicated for the fans and as a result they are NOT going to support the results, whatever they are!  This is the last post I am going to make on this thread as I think I have made my feelings clear.  Good luck.

What would be the point of having 30 players on each team, where 20 of them are identical? 

If both teams have Pujols, Ortiz, Vlad, Chipper, David Wright, et al, then that means that neither teams have them.  Right?  Every HR hit by those guys for the fans is also a HR hit for me.  It cancels out.  They have no reason to follow their performances at all.

Might as well just make it 10 players against 10 players!  Where’s the fun in that?

***

This study is not designed to prove if clutch hitting exists or not.  That’s what The Book did already.  This study is designed to answer the baseball version of the age-old question:

Ginger or Mary Ann?

The first thing that this balloting did was this:
no matter how much a fan thinks a guy is “clutch”, what they are really saying is that they want their best hitter period, if that guy is overwhelming better than the “heart/ice” guy; at least half the selected “clutch” guys is really the team’s best hitter.  They are not selecting Molina over Pujols, even if they think Molina is super clutcheroo.

Given that, the idea of clutch is now reduced to “how much will I trade heart/ice for talent”.  That is:
Mike Lowell v Kevin Youkilis
Ichiro v Ibanez
Encarnacion v Adam Dunn
Victor Martinez v Travis Hafner

Basically, the wOBA of the Fans’ heart/ice guy will be around .350, while the wOBA of the otherwise remaining best hitter on the team will have a wOBA of .370.

This is the choice I have basically given the fans.  Do you want the really good hitter with a .370 wOBA, or do you want a not-bad hitter with .350 wOBA, but that you believe he has heart and ice in his veins and are basically in love with this hitter.

Once I present the final list, it will be a fair representation of who the fans really prefer among the head-to-head.  I can go to the Redsox message board and tell them: “Head-to-head, you prefer Lowell over Yuke… you serious?” And they will answer an emphatic Yes!  The same for the Reds, Indians, and Mariners for the above players.

Utley v Howard is no big test obviously.  Both great hitters, with little separating them.  Howard is slightly better.

But, that’s what I’m going to end up with: a group of ice players, who are some 5 to 50 wOBA points behind the better hitters (for an average of likely 20 or so wOBA points), who the Fans prefer.

That’s it.  That’s what the test is designed for.  All I need is a fair representation of players from the Fans side to compete against.  And this is what I believe I am getting.

***

This is similar to “chemistry”.  Bonds is worth say 15MM for his talent.  But, teams aren’t paying that.  But, they would surely pay him 1MM right?  There’s a cost to how much they are valuing his antics.  Same for say Albert Belle or Dickie Allen or Milton Bradley.  There’s a price that you are willing to go, at which you will give some discount for chemistry, but not a whole lot.

The argument for Ichiro and Martinez and Lowell is how much of a premium do fans give them?  It sounds like they are going to give them some 20 or so wOBA points of premium. 

THAT is what fans are really saying when they talk about “clutch”.

I’m the Mariner fan who didn’t pick Ichiro!. I wanted Bloomquist, but you talked me out of it. (Yeah, that is definitely sarcasm.) I love the guy’s hitting ability, actually bought shirts with his name on them for the kids, etc. But you gave me an out by letting me rank the team. I absolutely believe Ichiro! is the best bet to get on base whether it’s clutch or not.

But in the “close and late and important” definition of clutch, I don’t expect power out of him and if he gets on, someone else has to bring him home (or he has to bring home someone else, that in all likelihood being a slow runner), which is the lack of depth issue for the Mariners. So I opted out of giving anyone a great, and chose two or three goods (can’t remember exactly) who I thought would give the team the best chance of winning. Ichiro! was certainly one of them.

Which is a long way around of saying that my mind works differently than most people, and can make something simple into something really complicated. I didn’t look up any numbers, but this was as close as I could come to a gut instinct. It also probably reflects my disdain for the “one great bat” theory of baseball.

To me, the problem with the study is that by having to reject Ortiz, Vlad, etc., you’re actually testing people that fans don’t consider the most clutch. I’m assuming even people who believe clutchiness exists can’t think it’s very common. I don’t know how to get around that, though.

I think it’s an interesting take on fans’ outlook, though, and I’m looking forward to the results through the season.

Considering that Jays fans haven’t seen a healthy Wells in a year and a half, there would be quite a few who would be negative about him. However, after offseason surgery his batting results could be quite improved (being healthy again and performing at an all-star level). Things like this will heavily influence the way fans will see some players I think.

To me, the problem with the study is that by having to reject Ortiz, Vlad, etc., you’re actually testing people that fans don’t consider the most clutch.

That is one of the problems I foresee and pointed out.  Fans will complain, “Well sure I lost, but you left out my best, most clutch hitters.”

Obviously the argument does not have much merit, but that is what they will say.

The nice thing about the way you set up the balloting is that you can see if there is ANY difference between the clutch and non-clutch performances of the players who the fans think are poor or good clutch hitters.  Sample size caveats will I’m sure be in effect though, since we would be looking for presumably small differences in a not so gigantic sample of PA.

I do love the Fangraph web page!  Very nice work there.  How is that STATS copyrights the page but the data is from their competitor, BIS?

Right, it doesn’t have to be the “most clutch”, as long as in the head-to-head, you are selecting the guy who is MORE clutch.

I don’t think any reasonable baseball fan will disagree that ALL fans prefer Ortiz, Vlad and Pujols.  For a baseball fan to say that I left out the MOST clutch players, I am not speaking to those baseball fans.  I am speaking only to those fans who confronted with Lowell v Yuke, that they agree that they want Lowell, and it is on the basis of being MORE clutch.  He doesn’t have to be the MOST clutch on the team, just MORE clutch than the other guy.

48 Tango-I get what you’re doing. I just have had the unfortunate duty to closely follow teams in the last few years who clutch wouldn’t really have helped. When you’re wallowing under, or even around, .500, odds are your offensive isn’t overly helpful anyway. I do think the Lowell v Yuke example is a good one, because it’s two players people care about on a team where they might actually have some memories of each coming through. Be interesting to see how much interest there is from KC, Pittsburgh, et al.

Yeah, KC and Royals are tough.  But they are tough in all my polls.  The blogdom is simply not there for those fans.  I don’t know what it is about Seattle and Baltimore that would necessarily vault them into the Redsox / Yankees blogdom, and not the Royals and Pirates.

I’m at 1400 ballots and still counting.  Love the blogs.

***

If I discard all the 570 ballots prior to the clarifications on my site (about the overachieving Jeter v substandard ARod; and how the team page specifically notes the performance relative to his teammates), the results are almost exactly the same, as to who the fans want with the game on the line.

Specifically for Jeter/ARod, Jeter has 27 votes and ARod has 19, if I ONLY look at the ballots since the updated instructions.  So, not as egregious as the pre-MGL objections, but it still stands as Jeter being preferred.

I should say thanks to MGL for making his objections, as it made me update the two pages to remove the ambiguity.

For the Whitesox however, there is a change.  Joe Crede is no longer the guy they want with the game on the line (yet), but it’s pretty close still.  I’ll reach out to the Whitesox blogs to get more votes in.

For the Reds, Dunn falls even further, and Javy Valentin takes a slight lead over the other 4 Reds players.

For Philly, Jimmy Rollins now takes the lead, ever so slightly ahead of the other 2.  Certainly not what I’d expect with the revised instructions!  I’ll reach out to more Phillies blogs, as this race is really close, and has a big impact.  I mean I can’t believe the fans would prefer Rollins v Howard over Utley v Howard.

For the Giants, Bengie Molina has an ENORMOUS advantage over the rest of his teammates.  I guess this is expected since the Giants fans have never seen Rowand play for them.

***

So, my conclusion is that the ambiguous instructions ended up impacting only one team: the Whitesox.  And I’ll do my best to make sure the Whitesox fans speak their mind as to who they really want.

I’m curious about something. From watching many many Cubs games, I believe that A Ramirez gears himself up for RBI situations. Maybe he believes that’s his main job, or the best use of his particular skills, or the best way to make more money. Whatever. Maybe a detailed analysis of his PAs would show that despite what I casually perceive, he does not in fact do better in that subset.

But, I am not sure at all if he also increasingly gears up for clutch factors beyond RBI situations (such as inning, score). My impression is that he doesn’t, or can’t, adjust his approach to those factors. And that makes sense, because the ideal goal of a ‘clutch’ PA (defined in some LI term) may be less clear to a batter than simply whether there are ducks on the pond, and the number of outs, etc.

So, maybe the way clutch is defined for batters should be not some ‘correct’ (as shown by LI) combination of the runners configuaration, outs, score, and inning--but instead some weighted combination of those factors which reflects how realistically batters are able to adjust to each component of LI. IOW, each of those would have its own regression factor.

Or, maybe just keep it simple and look at RBI %, relative to overall stats?

I think you are going to run into another criticism by the fans who voted.  Many of them will not like your definition of clutch and when they lose, they will say, “Well, that was NOT what I meant by clutch. I meant X.” Many fans do not appreciate the fact that if clutch exists it will show up in any reasonable definition.

I think you need to do at least one thing, which I think I already recommended.  That is, you need to let them know beforehand how you are going to define clutch.  That way, they can tailor their players to your definition if they like, or if they just don’t like your definition, they can not play.

The other thing is to let them define it.  Of course, you pretty much need to tell them that you are going to use a metric of production like OPS, lwts, RC, etc. and that they can pick the “situational” criteria.  Obviously you can’t let them choose RBI or something like that, since that changes the value of players, favoring hits and HR over walks, for example.  You could of course, but then you would have to choose your players based on some other valuing system besides a traditional, overall one (which assumes a neutral context).

As DS discusses above, I have always been an advocate of defining clutch (in clutch studies) as how a typical batter would perceive it not how it SHOULD be perceived based on the LI or importance of the game situation.  As David says, what is the point of that, if what you are trying to capture is how the batter himself might bring his game up or down a notch or two, based upon how HE perceives the importance of the situation?  But that doesn’t really have much to do with this game.

For this game, it doesn’t really matter how you are going to measure clutch, but I do think you should tell the voters up front how you are going to do it (another thing is that they may think you are defining it to stack the deck in your favor) and perhaps even let them choose the context. 
t

Given the resources available, it does not seem like it would be a problem to look at multiple ‘clutch’ definitions simultaneously.

They are aware of the definition.  I said in the article it’s at least double the leverage of an average situation, and I have a link to the chart of leverage index right in the article.

(The short form of that is: late & close, or early, close and two runners on base.  But, like I said, it’s all in the chart. At LI of 2 or greater, that’s 10% of PA.

***

I certainly can’t ask 2000 voters for their definition of clutch, and then try to come up with a consensus view of those wide-ranging views.

2000 ballots and still counting, with 4 days to go until (the 2nd) Opening Day.

Still too close to call in a few teams, so if you haven’t, please vote:

Angels
Marlins
Padres
Pirates
Reds
Rockies
Whitesox

Following up on mgl’s post, I re-read Tango’s intro of “which batter would you prefer with the game on the line”, and then I’m sure I read about Tango’s definition of a 2+ LI. But when I filled out the form for the Cubs, I simply went with my own instinctive impression of clutch, and that’s probably what most fans did.

Thinking about it a bit more, I kind of think the typical fan/sportswriter has a rather narrow perception of what is a clutch situation. I think it’s pretty much when the tying or go-ahead run is in scoring position in late/close situations. If you get a solid hit and drive in that runner, you have succeeded. If you don’t (even if you draw a BB), you’ve failed. I certainly don’t perceive bases loaded, no outs, top of 1st (LI=2.0) as a “clutch” situation.

And what type of hitter is most likely to deliver in my clutch situation? A high BIP rate, and a high BAvg. Maybe that’s why the fans are (IIRC) voting for Jeter over ARod--ARod batting .300 vs Jeter batting .320 with more BIP. (I am also aware that there is additional clutch reputation baggage both of these players).

Also, I don’t really think that the typical fan envisions clutch as a relative measure--as in “he hits better in the clutch than he usually does”. I think it’s more an ‘absolute’ thing, as in “he gets base hits with the tying/go-ahead runner in SP in late/close situations"--even if this batter’s stats are no better than his overall norm. This (if I’m right) implies that the typical fan’s concept of a clutch situation favors certain batter profiles or skill packages, instead of some suggested concept of intestinal fortitude (although the fans may incorrectly explain it in the latter way).

I forgot to include this.

Given my narrow fan’s concept of clutch, what about when a batter hits a HR with score tied, bases empty, and 2 outs in the 9th, to win the game? I liken that to the concept in ZR of a fielder being held responsible for BIZ, but also getting added credit for fielding BOZ. This type of situation is not considered clutch because the chances of success (hitting a walk-off HR) are small. So, if you fail, it’s not held against you, but if you actually do it, you get added clutch points…

I gotta think that any batter who comes up with bases loaded no outs even in the first innings realizes that this is a more pressure situation than nobody on, 2 outs, first inning, or something like that.

Tango, I know that you did tell the voters/readers that you were going to use any 2.0 or greater LI, but I don’t recall you telling them what metric you were going to use, which, as I said, is critical to you and them, as it determines the value of a player, either overall or in clutch situations.

Speaking of that, does a high leverage situation suggest more runners on base, or 2 outs more often than normal, etc., such that it really isn’t fair to use a context-neutral metric (like lwts) in clutch situations?

I didn’t mention a particular metric, because I will show the entire batting line.  I’m happy if the results may remain ambiguous.  But, if across the board the results favor one side, then it wouldn’t really matter what the metric of choice is.

More specifically, even if I say the metric of choice is OPS, someone else can say, “yeah, but look how many more RBIs I got than you”.  I want the results to be one-sided, either way.

Fangraphs is tracking it all, RBIs, R, OPS, WPA, WPA/LI, enough to satisfy any alphabet soup lover.

OK.  Certainly, then, it would be possible for the fans to beat your best players in things like RBI, since I assume your best players are being chosen for being best overall (wOBA, OPS, lwts, etc.).  In fact, the fans will probably lean towards favoring “RBI guys” as opposed to “OBP guys.”

Don’t look now, but the Fans are now TROUNCING my team after nearly 700 PA.  Their WPA/LI (Situational Wins) is now +1.5 wins ahead, even though they started off some 1.0 win behind (if we presumed no clutch hitting existed).  The RBI and OBP are a match, and close in batting average on balls in play, but my team’s hitters are getting killed with all those Ks. 

(ARod is 0-9.)

Simply put: the Fans are showing that their gut instincts is more than holding their own against the better hitters.

Anyone who wants to believe in Clutch hitting, I think that so far, this is your best argument.

The season is still young though…

Any idea what the SD of the difference is by chance, at this point?  That is important in order to put the results in context.  If the difference is 1 SD, then I would put little stock in it, other than perhaps they got lucky.  If the difference is 2.5 SD, then that is a different story.

Well, a .350 wOBA, on 670 PA is 1 SD = .020

Divide by 1.15 to get that into runs, and multiply by 670 to get total runs, and divide by 10 for wins, and you get: 1.2 wins.

Remembering that they started off 1.2 wins behind to start with (the clutch players were overall .020 wOBA points behind the better hitters), and they are now +1.5 wins ahead, that’s a 2.7 win gap, putting the gap at 2.25 SD.

***

Now, we are only look at the high-leverage situations.  It’s possible that the clutch players are having career years overall.  I didn’t check.  Regardless, the fans may have the ability to either find clutch players (guys who excel with the game on the line) or they can find players who can finally put it together (overall).  In either case, the gut instinct of the fan is being rewarded.