THE BOOK--Playing The Percentages In Baseball

Here’s what really bugs me: Almost every broadcaster/talking head has played professionally. The ones who haven’t probably played at least for a while as kids and into young adulthood. What’s more, nearly every baseball fan has played at some level.

I cannot remember ever feeling “hot” or “cold.” I remember feeling overmatched against certain pitchers. I remember feeling uncomfortable in one game or one at bat (but as an amateur, I’m convinced most of my truly awful games were because I don’t take care of my body the way a professional should). And on the other side, I remember times where I knew I could hit a pitcher or whatever. But I have never felt “hot” or “cold.” I might have talked about being in a slump, but only in the sense that going 1 for 15 IS a small slump (and not indicative of anything in the future).

So the real question isn’t why is Granderson one of the only guys who gets it. The real question should ask why all these millions of people who have played the game believe in something they never actually experienced when they played?

I think the problem is that players have heard of the concept of hot/cold, so when they’re in a slump, it’s an easy way to explain what the problem is, especially if they can’t find a mechanical reason like a change in their swing. So when they magically start hitting well again they assume it’s because they’re not cold anymore. If you were to question a player about being cold, and really push that it might just be randomness (this is something nearly any ESPN or Fox reporter would do), I bet they would come to terms with it.

@1 - I’m not sure I agree.  When I pitched in high school, I definitely had days when I was on, and days where I was off.

Hockey too - I’d jump over the boards for my first shift, deck a couple of guys, get a good scoring chance, and I could tell the game was going to go well.  There were definitely hot and cold days.

I would hope that professional athletes do a better job of mental preparation than I did, so that they get their optimal performance each time they play.  But amateur sports might lead you to conclude that hot and cold exist.

Hawerchuck, I’ve played a lot of sports in my life, and I too have felt invincible or like crap on any given day, but when the day is over, the next day I feel like I can perform as well as any other day, whether I had been performing poorly or exceptionally well on previous says.  I doubt that Swisher feels like he is not going perform well just because he has gone 4 for his last 40.  A new day, a new beginning.  Most athletes feel that way.  Somehow though when they retire and get into the booth, or even start to coach or manager, they morph into babbling idiots…

I’ve read a lot about this idea that there is no such thing as being hot or cold and statisticians that claim that they have studied the issue and found no evidence of it.

I never played baseball, but I did play a lot of basketball. And I definitely had nights when I could just look the ball in the basket and others when I couldn’t make one to save my life. The reason that ex-players believe in it is because they have experienced it themselves.

Hitters can be doing something wrong mechanically and go into a slump or lose their timing for whatever reason for a period of time or just feel uncomfortable at the plate and be a split second slow for a time. Pitchers can get their mechanics screwed up and get bombed for a few outings. Some of sports is “feel” or being “in the zone.”

I really don’t believe in this concept that there is no such thing as good and bad streaks. I think if you examine it mathematically, it may very well appear to be random. And in the scheme of things, it may be random because these periods come and go and balance out for the most part.

I think the only reason that it might not have been a good idea to bench Swisher is that you never know when he might break out of his funk and playing Hairston Jr. is never a particularly good idea.

I think these episodes can be a simple run of bad luck, but I don’t think there is any doubt that there is something behind the idea. Scientists used to say that there was no way that a pitcher could throw a curveball. It was just an optical illusion. They found out they were wrong.

MGL, I definitely agree with that assessment.  I played through college, and there are absolutely times when you are at the plate and say “How in the heck have a I have ever gotten a hit before?” (I once went 1-for-25; it was miserable).  Other days, the game feels like it’s happening in slow motion, and both of those feelings have lasted for more than just a few days at a time.  Still, there were times I got hits on those crappy days and went 0-for-4 with multiple strikeouts on the good days.  Baseball is hard, and most of the time the pitcher wins.  I know there has been research (and I’m probably going to completely screw up the results here) saying there is no such thing as a hot or cold streak, but I think that what the research is really saying is that there is no way to predict how long a hot or cold streak will last.  A player could snap out of it (or simply have his luck turn around) at any given moment and there is no way for observers to know when that will happen.  I’m glad that Granderson is competent enough to articulate these feelings; we need more people like him commenting on the game.

Mgl - I just think that when you aren’t singularly focused on playing a sport - ie, it’s not your day job - your training regime or lack thereof can cause large fluctuations in your peak performance on any given day or over the course of a week.  Certainly this would be the case with children - what if a kid does something stupid like have a burger and a shake 45 minutes before a hockey game? 

Pro athletes should get enough reps and have enough focus on the task at hand to have a much smaller range of performance on any given day. 

Can we say with certainty that this phenomenon does not exist?  If we set up an experiment, would the R^2 be exactly 0?

I don’t know why ex-pro athletes would think this particular issue plays such a large role in projected short-term performance.  But I can certainly see how if you’ve only competed at an amateur level, you could convince yourself that it’s an issue.

All sports are hard as hell to play. And the human brain does not function at the same level at all periods of time. When you are dealing with such fine levels of skill, the slightest alteration is going to get you totally screwed up. The brain has to find its way back to that “zone.”

It is really a ridiculous concept that all of these players and pretty much everyone that plays any sport knows about being “in the zone” or just out of whack and the mathematicians think they are the only ones that know the real deal.

There are guys that get to the point that they literally cannot throw a strike or throw the ball back to the pitcher. To say you cannot have that type of period to a lesser degree swinging a bat is ridiculous.

HH - You’re missing the point.  It’s not that hot and cold streaks do not exist, it’s that just the presence of such streaks offer no predictive value going forward. 

For example, if a player is 20/30, he is undeniably hot.  However, he will be expected to hit like he always has during his next 5 or so games.

I don’t think I’m missing the point at all. In fact, that’s exactly what I said. Dibble’s comment about going 0-for-8 is ridiculous as is pretty much everything that Dibble says.

But MGL is claiming that players are babbling idiots for believing in hot or cold streaks. Or at least that’s my interpretation. I’d say that’s totally offbase.

If you have to choose between a guy who is 20/30 and a guy that is 4/40, who are you going to choose or are you going to count on idea of random performance. One guy might might could use the day off. He might very well be “pressing.”

If someone wants to argue that you never know when a slump will end so you might as well keep a guy in the lineup, that’s one thing. If you want to say that someone is an idiot for thinking that slumps are a reality for any number of reasons when they are actually just random events of bad luck and largely imaginary, that’s another.

If you have to choose between a guy who is 20/30 and a guy that is 4/40, who are you going to choose or are you going to count on idea of random performance. One guy might might could use the day off. He might very well be “pressing.”

That is exactly the thing we are arguing against.  MGL and Tango studied, and found ZERO effect on the population.  Simply put, knowing that someone has been on a hot or cold streak offers no indication of how they will play during the next game.  So it doesn’t matter AT ALL that a player has gone 4/40 or 40/40, he is expected to perform equally the next game. 

If someone wants to argue that you never know when a slump will end so you might as well keep a guy in the lineup, that’s one thing.

Yes, that is exactly what we are arguing.

If you want to say that someone is an idiot for thinking that slumps are a reality for any number of reasons when they are actually just random events of bad luck and largely imaginary, that’s another.

Slumps can occur for many reasons that are independent of the player - that batter could have seen really tough pitches and played in pitchers ball parks - and many reasons that are dependent on the player - he may not be seeing the ball well, or whatever. 

However, the point is that simply knowing a player hasn’t been seeing the ball well doesn’t mean that he will continue to not see the ball well. 

I don’t think that MGL or anyone else has ever said that slumps aren’t “real”, just that they are not predictive.

Amateur gamblers feel “hot” and “cold” all the time at the blackjack tables.  As much or more so than these professional athletes.  I was lucky enough to win $80 one time in about 10 minutes at low stakes blackjack one new orleans night, and my friends, highly educated elite college graduates, insisted that i need to keep gambling cause i was “hot”.  The feeling of power when rolling sevens at a craps table with a crowd cheering you on, even if you are wining only $5 a hand, is near indescribable.  After rolling a few in a row, you the dice just look like beach balls coming at ya, baby.

Your short-term performance is influenced by your being “hot” or “cold.” If I’m shooting free throws and in a funk, my shots might be a 50-50 proposition. When I’m in the middle of a cold spell, I’d be able to tell you how likely I am to hit that next shot.

On the other hand, if I’ve made several in a row and am feeling it, I’d also be able to tell you that I’m more likely to hit the next shot.

That’s not the same thing as random luck with blackjack because making a basket is not a random event.

If you were to measure my performance during those spells, you’d find that there was something to the idea because my shooting would be worse when my shot was off. And better when it was on.

If you don’t know when those windows are then you have no way to measure and everything appears random.

Of course, the manager doesn’t know when those windows are either. But if you’ve got two 90% free throw shooters and one guy isn’t shooting well at the time and the other is, you might be smart to pick the other guy to shoot your technical with the game on the line.

In any case, this is a circular argument. From what I’ve read, the mathemeticians have claimed that there is no such thing as being “in the zone.” It’s all just random. I don’t believe it.

Of course, the manager doesn’t know when those windows are either. But if you’ve got two 90% free throw shooters and one guy isn’t shooting well at the time and the other is, you might be smart to pick the other guy to shoot your technical with the game on the line.

And it’s been proven, with baseball, that it doesn’t make a difference whether a player has been hot or not in terms of projection how well he will do in his next game.

In any case, this is a circular argument. From what I’ve read, the mathemeticians have claimed that there is no such thing as being “in the zone.” It’s all just random. I don’t believe it.

Nobody is saying hot streaks don’t exist.  HOT AND COLD STREAKS DO EXIST, AND CAN BE PARTIALLY ATTRIBUTED TO TEMPORARY IMPROVEMENT IN A PLAYERS ABILITY.  However, just because a player has been in a hot streak, doesn’t mean he will continue to be.  In fact, there is zero evidence to support that.

Yes, I’m aware of that. It could be predicted but only by the player. If the manager consults with the player, he might have a good basis for sitting a guy.

If a manager is in a World Series, he might be justified in sitting a guy because he might really be in an extended slump. It worked out for Girardi.

Maybe he was onto something and maybe he wasn’t. It wouldn’t be wise to keep betting on Hairston though. That’s a game the house is going to win eventually.

Why could it be predicted by the player?  Say the manager talks to the player before the game, and finds out that the player thinks his batspeed is below the norm. 

How will that change his projection?  .10 points of wOBA, .40 points (which is about what it need to be to justify taking out Swisher)?  Besides, I highly doubt that Girardi took out Swisher because he told him he wasn’t feeling well.  He made the decision before the game day, and almost certainly made it because Swisher was 4/40.

Why wouldn’t it? Do you think that a major league manager or NBA coaches have no ability whatsoever to determine when a player is likely to perform at a higher level?

Do you think that maybe Girardi can look at the way Swisher is swinging the bat in batting practice and decide that until he sees something better, he is not going to put him in the lineup?

Do you think that an NBA coach might not be able to recognize when it is not a player’s night, but the guy off the bench is feeling it and keep the hot guy in the game in the fourth quarter of a close game?

Do you think that I can shoot a shot and know pretty well that it isn’t going to go in the basket because something is just off while other times know full well it is going in?

It is clear from this thread that there are people that believe that players are a steady state system that perform at the same level at all times and the variation in performance is random. I’m not sure you believe that Nick, but there do seem to be some that do. And I think a lot of mathematical types think it as well.

Some managers are better than others. There are some that can recognize certain things better than others. Some managers are worth a few wins a year perhaps just for this reason. All things are not equal.

Do you think that maybe Girardi can look at the way Swisher is swinging the bat in batting practice and decide that until he sees something better, he is not going to put him in the lineup?

There are couple of issues here.  Say Swisher’s 4/40 was caused, in part, by him not seeing the ball well or another internal factor.  He notices that in the cage before the game, those problems seem to still be there, so he benches him,

Here are the issues with that:

1) That 4/40 isn’t what should be expected of Swisher if those internal problems persisted.  The 4/40 is most likely a combination of him seeing tougher pitchers, bad luck on balls in play, random variation and the internal factor(s) (like he isn’t seeing the ball well).  So you wouldn’t just assume that Swisher would continue to hit at a 4/40 pace just because you notice there are some issues, you would have to come up with a new projection which will almost certainly be a lot closer to his true talent level.  How much do you want to bet that 99% of managers would do the first one?

2) If managers did indeed remove players because their was something identifiable wrong with their swing or approach, then why don’t we EVER see players who have had good numbers in spite of that be removed?  If Swisher was 13/40 or even 10/40, I guarantee you he wouldn’t have been benched.  That implies to me that managers are making decisions off of recent stats, which I think we both agree is wrong, and NOT on actual observations. 

It is clear from this thread that there are people that believe that players are a steady state system that perform at the same level at all times and the variation in performance is random. I’m not sure you believe that Nick, but there do seem to be some that do. And I think a lot of mathematical types think it as well.

I don’t believe that players are robots, no.  I don’t think that anyone else here does either.  Some players will obviously have poor days because they are feeling sick, or didn’t get enough sleep and they will have good days because they had a good breakfast and had a date with Megan Fox the night before. 

I think that expecting a manager to be able to pick up on when a player is truly hot or not and whether or not he will continue to be is pretty ludicrous.  You obviously give a lot more credit to managers than anyone else, and I don’t see how that is justified based on the fact that they seemingly make all of their decisions off of stats in small sample sizes.

Nick, you’re beating an extremely dead horse.  You cannot argue (intelligently and productively) with someone who offers an opinion on a factual matter but has exactly zero evidence to support that opinion.

I suppose that you could convince them of the veracity of your position by showing them or telling them about the evidence that supports it, if you thought that might have some value, but if they still stick to their position, with still zero evidence to support it, the discussion is over.

The reason I am so passionate about sabermetrics has nothing to do with baseball. It has to to with raising a generation (and generations after that) of children who will grow up to be adults who either attempt to solve the problems in their lives and in the world with opinions that are supported by evidence or those that aren’t.  Everyone has an opinion about what “works.” Nothing wrong with that as a start. Generally the only ones that actually work are the ones that are supported by evidence.  That is how we will solve our health care problems, terrorism problems in the world, hunger, poverty, etc.  People will offer theories and opinions about these problems and on how to solve them and then we will “test them” with evidence.  That is exactly what sabermetrics is in a nutshell.

All these things are true. But the reason that you hire a Joe Girardi instead of a math professor who is good at probability equations is to use his powers of observation. The reason you hire a scouting director to draft a high school pitcher instead of a college player is because of his powers of observation and not just rely on probability.

They may be good or they may be bad. But do those abilities not exist because a person has looked at and cannot find it mathematically?

A mathematician cannot determine whether a player is more or less likely to break out of a slump at any given point because they have no idea when the slump is on or off. If they think the slump does not exist and is just an aberration, that is an even bigger problem.

They look at numbers after 4 for 40 when they should have been looking a numbers during the 4 for 40. But they don’t know where to look because it is impossible to know.

But a baseball manager might just be able to use his powers of observation to make an educated decision on the subject that the math professor cannot arrive at with his statistical evaluations.

And he just might make a decision that leads to a guy getting on base and resulting in the winning run scoring when the other guy would have been 0-for-4. Is he going to do it every time? More times than a simple random guess. Possibly. Can you prove mathematically that he was right or wrong prior to his making the decision?

You hire Bill Parcells to bench Drew Bledsoe with a winning record and bring in Tony Romo when it makes no sense to anyone watching. That’s the reason you hired the guy. You don’t rely on any mathematical probability models and assume Parcells’ observations have no value. Sometimes they do and sometimes they don’t. But his chances are better than the math professors’.

But if you’ve got two 90% free throw shooters and one guy isn’t shooting well at the time and the other is, you might be smart to pick the other guy to shoot your technical with the game on the line.

This seems like an odd example for your argument because you are picking between two players who have the exact same projection forward (90%)

If Shaq was hot 8/8 from the field and 5/5 from the ft line and Nash was cold 0/3 from field and 1/4 on FTs would you have Shaq shoot your technical fts?

Those are the kind of decisions coaches have to make. No one second guesses the easy calls like Nash versus Shaq.

Girardi isn’t going to bench Rodriguez no matter what his slump is but Swisher is another matter. One is a lot more easily replaced than the other.

If your drafting a player and you’ve got a college prospect to choose at No. 1 who looks like a possible all star but a high school pitcher that your scouts swear is the second-coming of Steve Carlton, are you going to take the college player every time because he projects better based on historical drafting results? Or are you going to trust your scouts opinion on the guy based on observation and experience?

Isn’t the difference between some good and bad organizations that they have guys that are good enough to outperform projection over time? Couldn’t going with projection leave you with mediocrity? Why don’t math professors rule the world?

Isn’t the difference between some good and bad organizations that they have guys that are good enough to outperform projection over time? Couldn’t going with projection leave you with mediocrity? Why don’t math professors rule the world?

Wow, I see your point MGL.

But the reason that you hire a Joe Girardi instead of a math professor who is good at probability equations is to use his powers of observation. The reason you hire a scouting director to draft a high school pitcher instead of a college player is because of his powers of observation and not just rely on probability.

Exactly. Statistical analysis provides a point estimate for a player’s “true talent” - but it is only an estimate, and the more information that you can get about the factors that make up a player’s “true talent”, the better the job that you can do in identifying players who are likely to outperform that estimate, whether it be in today’s game for an established player, or five years from now from a draft prospect.

Evidence doesn’t have to be limited to the statistical. It’s harder for people outside the game to evaluate non-statistical evidence, but that doesn’t make it irrelevant. But I also would suggest that it’s not necessarily “better” evidence, either - which is why (a) you want both and (b) you want someone with an open mind to evaluate all of it.

"All these things are true. But the reason that you hire a Joe Girardi instead of a math professor who is good at probability equations is to use his powers of observation.”

No.  The reason you do that is because that is what is expected.  There is tremendous risk aversion at play in Baseball.  People lucky enough to be in the system make a ton of money and there is no way they will jeopardize that by doing something intelligent and creative if it goes against the norm.  People can fail and still have a secure multi-million dollar future as long as they fail the same way everyone else fails.  If you are going to put that at risk, you have to know you will succeed.

There is no doubt in my mind that any of the regular posters here would do the tactical job of a manager (roster and in-game moves) better than any current manager.  Something like that, or your hypothetical “math professor” doesn’t happen because the people making those decisions are risk averse and can make more money more securely by hiring some old baseball player that has been in management for a while.  That is, of course, the right move, because in a single season things completely outside the control of the manager can (and do) cause true talent .580 teams to go .420.  If that happens when you have gone against the grain (see. LAD 2005) you get fired.  If it happens when you are doing what everyone else is doing then often you don’t get fired - or if you do you find yourself back in the same position with another team fairly shortly.

“There are three things the average man thinks he can do better than anybody else…start a fire, run a hotel and manage a baseball team.”

--Rocky Bridges

I wouldn’t hire any of you.

All these things are true. But the reason that you hire a Joe Girardi instead of a math professor who is good at probability equations is to use his powers of observation.

The reason you hire Joe Girardi is because managing the club, not managing the game, is probably more important in the grand scheme of things.  And that doesn’t make his silly on-field decisions justifiable, just makes it possible for a fan to sort of accept it and turn your head.  Ideally you have a guy who is able to both manage the game properly and manage the clubhouse--I don’t think that manager exists.

The reason you hire a scouting director to draft a high school pitcher instead of a college player is because of his powers of observation and not just rely on probability.

Amateur statistics are mostly useless.  The level of competition is so vastly different that even the best HS, JuCo, CC, and 4-year College players will have to get a LOT better in order to succeed in MLB.  Amateur statistics usually can’t answer the question, “Who will get better?”, only, “Who is good?”.  While the latter question is interesting, it’s the former that a team is interested in when drafting a player.

I don’t see how this supports your point.

A question for anyone who wants to jump in on this one:

If hot and cold streaks do not exist, or are not predictive, why would a manager ever take a pitcher like Roy Halladay, Johan Santana, Tim Lincecum or C.C. Sabathia out of the game before they hit 100 pitches?

Before you answer, remember, if you believe hot and cold streaks don’t exist, you believe past performance, ESPECIALLY recent, small-sample-size past performance, is meaningless.  And performance in the very game you are making a decision in is as recent, and as small-sample-size as it gets, so this should be the *textbook* example of what you believe should be severely discounted or ignored…

Lay it on me, and while you do that I’ll go check b-ref and see who that poor, misguided soul was who took Rick Ankiel out a playoff game just because he threw 5 wild pitches in an inning.  If only he had a properly calculated projection for Ankiel, he never would have done something so stupid…

The reason you hire Joe Girardi is because managing the club, not managing the game, is probably more important in the grand scheme of things.

Perhaps, but if that’s the case then it would seem to be a knock on the team’s front office. The publicly available information has all indicated that the team was satisfied with the ability of Girardi’s predecessor to handle the people skills of the job. The supposed knock was on that manager’s unwillingness to utilize the increasingly common statistical analyses for game management and strategy.

I can’t say I disagree with you, Sodium Hydroxide.  And I don’t know if what I originally said is REALLY the case, but it seems to be so.

Here’s a question: If front offices have generally gotten player valuation down to a science, are managers financially compensated for their win values as well?  Or is a manager’s salary based purely on silly intangibles that we have no way of knowing weather or not their assumptions are true.

If the former is true, we (outsiders) probably greatly overstate a manager’s impact.  If the latter is true, well, I wouldn’t be surprised.

I can’t say I disagree with you, Sodium Hydroxide.

Perhaps my name is Nathan O’Hara. Nah, you have it correct.

If front offices have generally gotten player valuation down to a science, are managers financially compensated for their win values as well?

More statistically informed people than me — like Tango and MGL — don’t seem to think front offices have player valuation down to a science. This would suggest managers are not compensated based on win share values. Sticking with Girardi as the example, is there anything in his one year of managing the Marlins that would correlate to the Yankees having signed him at $2.5MM per year for three years?

Or is a manager’s salary based purely on silly intangibles that we have no way of knowing weather or not their assumptions are true.

Well, I’d guess that’s probably some of it. But I would also guess there are some market forces at work. Among those elements would likely be things like team economics, the general economic market of MLB manager salaries, and perceptions about how a manager may help a team’s marketing. According to Maury Brown, manager salaries this year ranged from Bob Geren at $500K to Joe Torre at $7.5MM.

Maury Link: http://bit.ly/2JjqLn

Look, I don’t think that managers have any particular ability to determine if a player is going to come out of a slump or not. As a general rule, they probably don’t.

I think there might be times when they might be able to. And I think certain managers might be able to better than others. I would not categorically say that it cannot be done. Many apparently would. I’m not sure why.

“Remember when I told you that one of the differences between a smart person and a not-so-smart one is not necessarily in their respective breadth of knowledge (or some such thing)?  That the difference is that the smart one knows when he doesn’t know something and that the dumb one thinks he knows a lot more than he does?”—MGL

Those were wise words. I know people posting here have no idea whether Girardi’s decision was wrong or right and neither does MGL. You have to know what Swisher’s true level of performance was that day. There is no way to know that unless you believe that players are a steady state system that never vary. It cannot be measured if you calculate from now to doomsday.

This is a generality and generalities are not a “factual matter.”

And Greg Rybarczyk makes a good point. Not only why would they take them out before 100 pitches, but why would they take them out after?

Why would you think you could see a guy tiring out on the mound? Why would you think the fact that he’s given up back to back homers would be any indication that he couldn’t finish off the game?

That would be absurd. He’s just as likely to complete the last 3 innings as well as he did the first three. Those five straight line shots are a statistical blip. Keep the best pitcher in the game if you want to win the damn thing.

"The reason you hire Joe Girardi is because managing the club, not managing the game, is probably more important in the grand scheme of things.”

Sure, which is why I refereed in my reply to “the tactical job of a manager...”

I can see no reason at all why these two jobs (managing the team and managing the game) cannot be separated apart from, again, risk averse FO and custom.  Additionally, I can’t see “managing the team” as requiring “old baseball guy” either.  Like any professional, I have been managing people for over a decade - whether those people are employees, co-workers or clients.  That sort of management skill is not one I developed playing baseball or football (though both baseball and football are reasonable environments in which to develop those skills).

So, again, I reject the assertion that Girardi (or any MLB manager) is really the best choice for the job or that his #1 (if not only) “value” to his potential employers is that he is an “old baseball guy.”

More statistically informed people than me — like Tango and MGL — don’t seem to think front offices have player valuation down to a science.

Well, yes.  But, if I understand their positions correctly, that’s because their assumptions are incorrect, not their model.

Well, yes.  But, if I understand their positions correctly, that’s because their assumptions are incorrect, not their model.

Your first “their” presumably refers to folks like MGL and Tango. I presume your second and third “their” also refer to them, not front offices. If so, I was under the impression that Tom and MGL weren’t using assumptions but instead were basing their opinions off of the results of their research.

HH #34:

There are stats that show that a pitcher’s effectiveness decreases after X number of pitches, so I think the “after 100” part is justifiable.

But, if one’s dedication to the belief that there are no good days and bad days, but only days, is pure, then when Felix Hernandez gives up 6 runs on 6 hits (2 HR), with only 2 K’s in the first 4 innings of a 6-5 game, your advice would have to be to send him back out there for the 5th inning, wouldn’t it?  Well, the Mariners are a well-known extremely saber-friendly org. that employs someone we all know and love as a consultant, and they pulled King Felix and replaced him with Sean White.

http://www.baseball-reference.com/boxes/MIN/MIN200905090.shtml

Now, maybe 81 pitches, which is what he threw in this game, is his effectiveness threshold, but b-ref tells us that he started another 27 games after that in 2009, and threw more than 100 pitches in all 27.  So clearly the Mariners do not regard 81 pitches as Felix’s effective limit.  They must have felt he was less effective than usual that day (for some reason or another )

With regards to Greg’s question, there is something to be said about not necessarily the pitch count, but in general the amount of times through the order. These are obviously going to be quite related, but there is definitely a decrease in effectiveness the more times a pitcher goes through the order, so even if, for example, Roy Halladay has hit the sixth inning and has somehow gone through the order three or so times, it may be wise to remove him because we can expect his projection to be worse than our relievers, especially after platoon considerations.

With Ankiel in particular, it brings up the case of observational analysis/scouting reports also being important. It’s absolutely true that players aren’t Strat-O-Matic cards and don’t perform at true level all the time. There can be observations made without use of statistics, and those observations can be very good. I don’t really think anyone here is doubting that. The problem with a move like this one is that it “seems” so based on the recent numbers, which aren’t significant to determine a future performance level, rather than a real issue, like what Ankiel had.

I presume your second and third “their” also refer to them, not front offices.

I should have clarified, but no.  The first “their” does refer to MGL, Tango, etc.. The 2nd and 3rd “their"s refer to FO people.

Thanks for the clarification, Peter. If I may re-write your sentence just so we’re all on the same page…

If I correctly understand the position of folks like MGL and Tango, the problem with front offices when it comes to player evaluation is in the assumptions, not with the models used.

I’m not qualified to discredit this, but if you (or MGL or Tango) are correct, then this seems like bad science. Assumptions shouldn’t be guiding a team’s research. Questions and doubts should be the basis for fact-finding and scrutiny.

If this presumption about teams is correct, well, then they’re doing it wrong.

Assumptions shouldn’t be guiding a team’s research. Questions and doubts should be the basis for fact-finding and scrutiny.

If this presumption about teams is correct, well, then they’re doing it wrong.

Right.  We agree.  It’s not the way it should be, but it’s the way it is (or the way I think it is).

My point was that teams probably do have an understanding as to how to value players.  The model the authors here use:

salary = playing time * (wins minus baseline) * FA_Arb_multipler

Is probably the same thing as (or theoretically similar to) the models the FO’s use.  My point was, where FO’s deviate from the truth in their valuation of players is generally in the “wins minus baseline” figure.  It’s not a systematic deviation, it’s a difference in the definition of the parameters.  At least that’s what I would like to think.

But let’s assume that’s true for a second, whether it actually is or not (we can’t really know).  Let’s assume that FO’s do use the above model to calculate the value of players (regardless of their understanding of the parameters).  Wouldn’t it stand to reason that managers would be similarly valued (and in turn, compensated) based on their win values?  And if not, how long will it be until managers are compensated for their win values? 

It’s a difficult thing to quantify, of course, which presents many tangible problems with my theory.

In theory, yes, it does stand to reason that managers would be valued in a way similar to players. But I just don’t think that’s what is being done (of course, like you, I have no actual evidence).

Besides the fact that the link I provided before doesn’t work, it was the wrong data as it was for 2007. Here are the 2009 managerial salaries.

http://www.sportsbusinessdaily.com/article/116188

I just don’t know how it’s possible for Torre ($7.5MM) to provide more than double the value of Piniella ($3.5MM). Or be 2.5 times more valuable than Bobby Cox ($3MM). Or, from the bottom of the pay scale, 15 times better than Bob Geren ($500K). Are Joe Maddon and Bud Black equivalent since they were paid the same amount?

I wouldn’t rule out me failing to grasp some system guiding these valuations, but it looks like managerial salaries are largely based on other factors, some subjective (e.g., reputation) and measurable (e.g., marketability).

As you said, though, even if more rigorous evaluations are not currently in place, when will they be? I don’t know, but could teams perform similar analysis for other key off-field personnel, like GMs. scouts, medical staff, etc.?

In theory, yes, it does stand to reason that managers would be valued in a way similar to players. But I just don’t think that’s what is being done

I think you’re probably right.

it looks like managerial salaries are largely based on other factors, some subjective (e.g., reputation) and measurable (e.g., marketability)

You do bring up a good point about marketability.  The sabermetric community largely overlooks this, but it’s certainly worth something.  But, again, I think you’re probably right.  And based on my understanding of the game and these basic assumptions, to quote you:

they’re doinng it wrong

Continuing

As you said, though, even if more rigorous evaluations are not currently in place, when will they be? I don’t know, but could teams perform similar analysis for other key off-field personnel, like GMs. scouts, medical staff, etc.?

It’s extremely difficult to quantify a manager’s impact (at least disregarding his on-field decisions, and at least I think), and it would seem to be equally difficult to quantify the value of FO execs or baseball employees in general.  But I think (and I sincerely hope) that as our baseball knowledge expands, we will learn to quantify these things.  I hope it’s sooner rather than later.

In response to Greg:  never ever have I said that the true talent is static.

Always and forever have I said that the true talent changes on a second-by-second basis.  The reason is that players are people not automotons.

What I maintain is that if you focus ONLY on the performance numbers (6 IP, 6 runs), you will NEVER be able to make it change your opinion as to the player’s true talent level (it’ll go from say a “82”, on a 0-100 scale to an “81.5").  Basically, it’s useless.  Now, for extreme situations, as we’ve shown in The Book, it moves SLIGHTLY more, but not that much more.

What does count is if the player himself tells you he feels different.  Or you, as a trained observer, see something.  But, the player and the observer has to come to that conclusion WITHOUT looking at the numbers.

If Rick Ankiel is throwing pitches 10 feet in the air, and more than once, that is about as an extreme condition as you will get.  And if Ankiel himself tells you that he’s on the brink, or a trained observer sees him on the brink, then you can estimate his true talent level as changing drastically.

This is all consistent with what’s in The Book…

In response to Greg R:  Roy Halladay, CC Sabathia and other top pitchers don’t get taken out at less than 100 pitches. 

Halladay was taken out at 99 on opening day, 95 on may 22@ atl (they pinch hit for him in a scoreless game) once for injury reasons and once on his first game back after the injury.

CC was taken out at 96 on opening day, once for injury reasons (and at 99 on his first day back), once at 94 while up 7-2, and the last two games of the year.

Lincecum and Santana were taken out a little more often than these two, but the point holds - top pitchers get removed due to fatigue concerns, not because of ineffectiveness.  Excluding Lincecum’s 78 pitch opening day start, none of the four pitchers you listed was taken out before 80 pitches, and rarely before 100.

I guess that means A.J. Burnett is not a great pitcher, as he got taken out after only 53 pitches in Game 5 tonight.

Just sayin’

I haven’t looked at the historical numbers but it seems to me that ALL pitchers get taken out early if they are getting hammered (like tonight).  Of course great pitchers are much less likely to get hammered early.  Of course it is actually sensible for a manager to tend to leave in great pitchers that are getting hammered and take out those who aren’t good in the first place.