The Book Blog Comments

How much should minor leaguers make? (01:15:34-05:00)

2012-02-23T01:15:34Z

Dirk Hayhurst talking with a teammate:

“Maybe I’m wrong for thinking this, but it makes me wonder why there is such a huge gap between the guys up here and the guys in the minors. I mean if you just spread out the smallest portion of all this to the guys below, it would make their lives so much easier, don’t you think?”

“That’s a terrible idea,” said Bentley.

Moments later, Bentley explains:

“This is the only level you can make an impact at. It’s the only one that matters—the only one people care about. All the rest of that stuff is just practice to get here.”

“But—“

“No buts.” He stopped me. “This is the only league that matters. Your career in baseball starts here.”

I don’t know how much money the minor leagues takes in. If you figure about 10,000 home dates at 4000 people per game at 10$ per ticket, that’s 400MM$ for all the minor leaguers, some 8000 of them. That’s 50,000$ of revenue generated per player. Half of that to the players in salary, and that’s 25,000$.

That’s why players get their signing bonuses when they are drafted, because that’s how they can make money (on their potential) before actually being called to MLB. For the in-betweeners, guys who are not believed to have high potential and so are drafted low with a low signing bonus, and at the same time, also have a small chance of making MLB, those guys are going to get very little money for several years.

Presumably, this acts as an incentive to make them develop “better, stronger, faster”, like Lee Majors.

Let’s say you make the floor minor league salary 100,000$. What would be the effect there? Chances are, you have lots more competition, so alot of the bubble guys who would play for little money to begin with, get squeezed out by guys who are just as good, but would have otherwise chosen a profession of engineering or accounting or electrician. Is this a good thing? I don’t know.

I don’t know what the right answer is. But whatever compensation system you derive will have some consequence, many expected (be it intended or unintended), and others that are unexpected.

What Dirk seems to be really saying is that the minor league players deserve some union, so someone can speak on their behalf.

Not everything you learn in college is true (duh)… (22:31:14-05:00)

2012-02-22T22:31:14Z

By mgl8@cox.net

Almost anyone who has ever taken a stats class in college and especially one in medical school (or graduate social psychology) has been taught this:

If you do an experiment and you get a P value of .05 (your result is 2 SD from the mean of the null hypothesis) that means that you have a 2.5% chance of making a Type I error (concluding that the null hypothesis is false when in fact it is true) if you reject the null hypothesis. (and of course if you can live with that, go ahead and reject it!)

The fact that that (rejecting the NH because you can “live with” a 2.5% chance of being in error) may not be very satisfying to some people not withstanding, is that even correct - that when P=.05, you have a 2.5% chance of making a Type I error if you reject the null hypothesis?

Where does the 2.5% chance come from? It is, of course, the probability that your result will be at least as far away from the mean of the null hypothesis as it was, assuming that your experiment yields a “normal” (bell curve) distribution of expected results. In other words, if you performed your experiment an infinite number of times, 2.5% of the time your results would be as far away from the mean as you got, or further, if the NH were true.

Notice that I bolded “at least as far away.” Not exactly as far away as you got, but at least as far away as you got. In fact, you got the absolute minimum you can get and still qualify for a spot in that 2.5% of the area under the bell curve. It is like if I got together with everyone who scored in the top 10% of a certain test and we all celebrated how smart we were, but it turned out that I came in 100 out of 1000. In other words, I just barely made the cut - everyone else at the party was smarter than I (assuming the test measured smartness)!

That is what happens when you declare a P value. You declare it on a value, like my score on that test, that “just barely makes the cut.” Yet the 2.5% includes every other value further from the mean than the one I got in my experimental results, and none closer.

What right do I have to proclaim that I have a 2.5% chance of making a Type I error in rejecting the NH? Surely my value needs to be somewhere solidly within that area before I can declare that my chances of making a Type I error is 2.5%?

Would you cut to a regularly scheduled show, if the main event ran long? (17:27:16-05:00)

2012-02-22T17:27:16Z

By Tangotiger

Non-sports post.

After the Heidi debacle, it became clear that you can’t f-ck with an audience that is invested in the outcome of a show, be it an NFL game or otherwise. You wait until Miss Universe gets the crown on her head, before closing the show.

As if cutting off Frank Sinatra wasn’t bad enough twenty years ago, now the latest sensation, their own Adele, got the same treatment from the Brits. And good for her for expressing herself.

***

And in completely unrelated news: government-mandated transvaginal ultrasounds as a precondition? Really? Really?! A footnote in American history that hopefully died today.

I wonder if Jon Stewart and Amy Poehler had any influence on this?

This week in chart failure (17:02:06-05:00)

2012-02-22T17:02:06Z

By Tangotiger

Real data:

Faux News version of real data (focus on the last data point):

Also, the scale on the left is out of proportion. If you compare the 8.8% point to the 9.1% point, we see there’s two lines of gap, meaning 0.3% = 2 lines, or 0.1% = 0.67 lines. 1% point should be 6.7 lines. Now, compare the gap between 8% and 9% on the y-axis. It’s 3.5 to 4.0 lines!

If we treat the 9% on the y-axis as valid, then where it shows 10%, it should actually be showing about 9.5% or 9.6%.

Glove-slap: Flowing Data.

Who’s evaluating the 2011 forecasts this year? (16:26:27-05:00)

2012-02-22T16:26:27Z

By Tangotiger

Anyone going to step up? Anyone? The hard part is collecting all the data, and matching all the players. If someone ELSE does all that hard work, I can step in and do the rest.

The test is pretty simple.
1. Calculate wOBA for every forecast, and for the actual. I’ll do something simple like
numerator = 0.7*BB + 0.9*1B + 1.3*(2B+3B) + 2*HR
denominator = BB+AB

It really doesn’t matter much what you do here. You just need something that focuses on the important stats, and make sure everyone forecasted those stats.

2. Calculate each population mean, by weighting by actual PA (AB+BB). For missing players, either give them the population mean (you HAVE to do this for Marcel, since by definition, Marcel has no missing players), or set the wOBA at 20 points below the rest of the population mean.

3. Recalculate new population mean (where applicable).

4. Baseline each player to a common mean (set to .330, but it doesn’t really matter what you set it to). So, if the pop mean in #3 is .327, and you have a player forecasted for .377, his adjusted forecast is .380.

5. Calculate the difference for every player.

6. Present the average absolute difference, and the RMSE, and in both cases, weighted by the actual PA.

That’s it, that’s the basis.

Then you can do fun stuff, like splitting by career performances. Guys with 1500 or more PA in the last 3 years, guys with fewer than 250 PA, guys who had a .380+ wOBA in the last three years, shortstops, etc, etc. Look for whatever attribute of a player you want. And compare the systems, and look for bias.

Bueller?

MLB 2012 Odds: BetOnline (12:21:21-05:00)

2012-02-22T12:21:21Z

By Tangotiger

Glove-slap Vegas Watch.

One SD is 8.0 wins, which is right where it should be.

95.5 Phila.
94 Detroit
94 Texas
93 New York
89.5 Los Angeles
87.5 Tampa Bay
87.5 Boston
87.5 San Fran.
87 Cincinnati
87 St. Louis
85.5 Atlanta
84.5 Arizona
82.5 Miami
81.5 Toronto
81.5 Milwaukee
81.5 Colorado
81.5 Los Angeles
81 Wash.
78.5 Kansas City
77.5 Chicago
75.5 Cleveland
74.5 New York
74 Minnesota
73.5 Chicago
73 Pittsburgh
73 Oakland
72.5 Seattle
71 Baltimore
70.5 San Diego
62.5 Houston

K minus BB differential or ratio? (07:11:37-05:00)

2012-02-22T07:11:37Z

By Tangotiger

I’m having a discussion with Bill James on the virtue of the K minus BB differential, compared to the ratio. Bill took the challenge (sub only). There are 3 things he did there that would cause bias, because he looked at all data from 1947-2010:
1. Because K rates have been on a steady rise, and there’s been a shift in batting average on balls in play between 1992 and 1994 (.280 pre-1992, .300 post-1994), that the groupings that Bill has done may have an era bias. This is why I find it helpful to look at the data from 1993-present, because there’s been a pretty strong line drawn there, in terms of looking at unadjusted data.

2. I had said it should be K minus BB per PA, not per IP. It’s not that big a deal, but it still has some bias to it.

3. And of course, HR exploded at the same time BABIP jumped (between 1992-1994). And so, if you have more guys in the high K minus BB per PA group, you’d also have higher HR rates as well.

***

I did my own study. I looked only at 1993-2010. I grouped all data by player, so we have 2513 pitching lines (no PA limit). I calculated a “walk” as BB-IBB+HB. I calculated each pitcher’s strikeout minus walk and divided by plate appearances.

I then grouped them into 5 categories:
Group 4: differential of at least .14 per PA
Group 3: differential of .10 to .14 per PA
Group 2: differential of .06 to .10 per PA
Group 1: differential of .03 to .06 per PA
Group 0: differential of less than .03 per PA

I also calculated their runs allowed per 27 outs (i.e., per 9 IP):

Grp    IPouts    diff/PA    R/27    6-14*diff/PA
4    206352     0.169      3.66      3.63 
3    436797     0.116      4.33      4.37 
2    866050     0.079      4.80      4.90 
1    530103     0.047      5.26      5.34 
0    208194     0.009      5.94      5.88

Grp is the groups I noted above.
IPouts is IP*3.
diff/PA is the strikeout minus walk differential per PA
R/27 is runs allowed per 27 outs (or per 9IP)
6 - 14*diff/PA is simply six minus 14 times the diff/PA in the previous column

As you can see, a very strong relationship.

Bill however upped the ante, and looked at within each differential group at the K/BB ratio.

So, what I did was flag every pitcher with at least a 3 K/BB ratio as “high ratio”, anyone with under a 1.25 K/BB ratio as “low ratio”, and the rest as “average ratio”.

First up is for Group 4:

Ratio    IPouts    diff/PA    R/27    6-14*diff/PA
Aver.    70668     0.157      3.76      3.80 
High    135684     0.176      3.61      3.54

As we can see, the high ratio pitchers do indeed get a better value than their differential would suggest, but the difference is pretty small.

Group 3:

Ratio    IPouts    diff/PA    R/27    6-14*diff/PA
Aver.    372113     0.115      4.36      4.40 
High    64684     0.126      4.16      4.24

In this case, both of them are pretty much off by the same amount relative to expectations of using differentials only.

Group 2:

Ratio    IPouts    diff/PA    R/27    6-14*diff/PA
Aver.    862868     0.079      4.80      4.90 
High    3182     0.078      4.90      4.91

In this case, the lower ratio pitchers are actually the better pitchers.

Group 1:

Ratio    IPouts    diff/PA    R/27    6-14*diff/PA
Low    8204     0.033      5.46      5.54 
Aver.    521899     0.047      5.25      5.34

Here, both groups, the Low and the Average ratios, are off by a similar amount from expectations using differentials only.

Group 0:

Ratio    IPouts    diff/PA    R/27    6-14*diff/PA
Low    166582     0.004      6.09      5.94 
Aver.    41612     0.027      5.31      5.62

The biggest gap is here, with the differentials suggesting that the Low Ratio and Average Ratio pitchers should be off by only 0.32 runs, but are instead off by 0.78 runs. The worse performing group is the low K/BB ratio pitchers.

Overall, we see that while the ratio may have some additional information for us, a simple and straight strikeout minus walk differential per PA is a great indicator of performance.

Two players have the same stats: one is much younger. Which one will be better next year? (01:18:55-05:00)

2012-02-22T01:18:55Z

By Tangotiger

Someone basically asks this question.

If we believe in humans learning and facing new environments, then it should be clear that we need to include the player’s age, relative to the age of his opponents, when forecasting.

I believe Brian/Oliver has specific translation numbers to this effect.

Knuckleball pitchers: all of them (14:49:58-05:00)

2012-02-21T14:49:58Z

By Tangotiger

Here you go.

Glove-slap: Joe.

Proper compensation for Epstein? (13:57:13-05:00)

2012-02-21T13:57:13Z

By Tangotiger

This one should be easy to figure out, no?

John Buck, Placido Polanco, Joaquin Benoit, Juan Uribe all recently signed a 3-yr 18MM$ contract (more or less), as free agents. Epstein is signing an 18MM$ deal, but the Redsox still own one more year of rights to him.

So, here’s the question: if any of those players were being paid at their fair value with their original team, but had one year to go, how much would a team give up for them? That is, Uribe is with the Whitesox, he’s due to earn 7MM$ in his last year, he’s expected to perform as if he’s worth 7MM$, and then the Giants want to trade for him with that one year to go (and then offer him a contract extension). What do the Whitesox get for him (setting aside the free agent compensation rules)?

In the case of Esptein, he’s a 3.6MM$/yr asset. We have to therefore think of a lower-priced asset, like, Matt Guerrier (who signed a 3yr/12MM$ deal). What would the Twins have asked for him if he was already getting 4MM$?

It seems to me the market for that kind of player, one already paid his full value (*), with one year of exclusive control, must be pretty weak. I’m thinking some mop-up reliever should be fair compensation, and fit in with the precedents of other baseball assets similarly traded.

(*) I’m presuming the Redsox are paying him his full value, and the Cubs are paying more because they are giving him more responsibilities.

Paper on predicting pitches (to be presented at Sloan) (13:32:38-05:00)

2012-02-21T13:32:38Z

By mgl8@cox.net

http://www.sloansportsconference.com/?p=6137

Click on the link on that page to download or view the entire paper.

I read it. It is short. I don’t understand the model. I am not really sure what their “naive” model is. It sounds like it is simply a pitcher’s overall (across all situations, counts, etc.) fastball percentage, but they don’t really say.

If it is, of course anyone can come up with a model that incorporates inning, count, score, batter, etc. that is a lot better, so I have no idea how to judge their results.

Also, one of their parameters is “score differential.” They found very little weight to be attached to that parameter. That doesn’t make sense since surely if you are ahead by lots of runs in the late innings you are supposed to throw lots of fastballs. The reason their results don’t make sense, I think, is that they defined score differential as the absolute value of the difference between the scores. That is ridiculous of course. Pitchers are going to have markedly different approaches depending upon whether they are UP or DOWN in the game, especially when the margin is large. If I am up a lot of runs, I can throw mostly fastballs. If I am down lots of runs, I need to mix up my pitches as much as a I can to prevent any more runs from being scored.

So the reason they got around the same fastball rate regardless of the “score differential” was probably because the change in pitching approach when down and when up cancelled one another out.

If they didn’t consider the “sign” of the score (whether the pitching team was ahead or behind) in their model, I can’t imagine that it can be a very good model.

Anyway, if anyone wants to read this (as I said, it is a short paper), I would like to hear your comments. I am going to the conference and would like to ask the authors some questions…

When is it a partnership and when is it a competition? (13:25:59-05:00)

2012-02-21T13:25:59Z

By Tangotiger

Marc asks if having more sports teams would help or hurt the Mariners.

We can even think more locally. Are you better off having a restaurant as part of the strip of a whole bunch of other restaurants, or are you better off being the only place in town? Out of sight, out of mind? Or exclusively located?

Why did it take so long to invent these things? (12:02:12-05:00)

2012-02-21T12:02:12Z

By mgl8@cox.net

Wheels and long handles on luggage (that actually work).

Self adhesive stamps and envelopes.

Any others along these lines (not like the steam engine or the fax machine)?

Going through the Sauer/Hakes Moneyball data (17:11:09-05:00)

2012-02-20T17:11:09Z

By Tangotiger

Sauer and Hakes were kind enough to give me their dataset. As you know, I took exception to their findings, because I don’t believe their model captures the reality of baseball. That while they did a great job in identifying the parameters to study, the way they combined those parameters left alot to be desired.

As I noted earlier, in the 2004 equation, it implies that two players with equivalent hitting stats will be paid more in the outfield than the infield. This makes no sense of course. Positional scarcity (which comes about because the fielding talent is greater in the infield) would demand that the infielders are paid more for equivalent hitting. A baseball fan will accept this as true. A GM will pay for this as true. A model that doesn’t reflect the reality of what we know by our guts and what actually transpires on the field has shortcomings.

So, what happens in the 2004 data? Using their data (setting filters on year=2003, freeyear=2004, free=1, inf=1) we have 16 infielders that became free agents for the 2004 season. The most prominent of these players is Miguel Tejada, who we know signed a huge 6/72MM deal with the Orioles. This deal, however, was backloaded. His actual payout for the 2004 season, using the Hakes/Sauer data, is 4.8MM$. Therein lies the problem. It’s irrelevant what he was actually paid for that season, since he did sign a 6-yr deal. The quickest way to note his salary is to just give him each year at 12MM$. This makes an enormous difference.

(Joe Sakic, in the prime of his career, once signed a front-loaded offer sheet from the NY Rangers for 3/21, with 17MM being payable in the first year. We’re not going to sit here and say that we should count his salaries as 17, 2, 2, are we?)

There were 3 other infielders who had similar hitting stats to Tejada: Luis Castillo, Joe Randa, and Mark Grudzielanek. Let’s look at each of these guys.

Grudz, according to their data, was a free agent entering the 2004 season. In 2003, for the Cubs, he earned 5.5MM$. He was still with the Cubs in 2004, but for 2.5MM$. In 2003, he had one of his best hitting seasons, and even got a couple of MVP votes (at the low end). In 2004, he played 81 games. I can only presume that Grudz was injured, and so, the Cubs got the injury discount.

Joe Randa was 34 years old.

Luis Castillo signed a 3/16 deal starting in 2004. The Sauer/Hakes data shows him with a 4.7MM$ deal (which is not much different from 5.3MM$, but still works against him, and infielders). Castillo, as we know, is an onbase machine. Plus a good fielder. And yet, all he got was a 3/16 deal? His OPS+ in the three preceding seasons (2001, 2002, 2003) were: 81, 95, 106. But, the Sauer/Hakes model only looked at his 2003 performance (106). So, it looks like he’s as good a hitter as Tejada, when in fact he was not. Tejada’s OPS+ from 2001-2003: 109, 128, 111.

The fifth best hitting infielder was probably Todd Walker, and we know he’s a disaster as a fielder. So, if he signs a low contract, we know that teams are valuing based on that fact.

Then there’s Greg Norton, who signed a 600K deal, even though his OBP/SLG were a bit above league average and he was a 3B. How does that happen? Well, Norton played in Coors in 2003, so his performance was actually pretty bad as a hitter. And, he was a backup. Now, they do include a PA parameter, but that’s not enough to undo this damage. Their model has him at 1.2MM$, and he was paid half that (barely above the minimum level). Indeed, by their regression equation, it is virtually impossible to fit the data to get a 600K free agent salary.

I took the 6 infielders with the best hitting stats in 2003, and I was able to explain an anomoly for each one. The end-result is what the regression equation told us: infielders should get paid less than outfielders for the same offensive performance. While that, inandof itself, does not necessarily damn the study, it points to bias or data anomoly. The Tejada salary misuse is fairly damning. We have pointed to many other issues in past threads as to scratch our heads at what the regression equations are saying. Overall, the results simply point to claims that do not seem reasonable, and the supporting evidence has enough questions as to invalidate the study.

***

I was listening to NPR on my drive home, and I missed most of the inteview, but from what I gathered, they were talking about the recently deceased James Lilley, U.S. Envoy to China during Tiananmen crackdown. And they recounted a story as to how someone told him “your data is wrong” and his response was “then I have to retract my opinion”.

Hopefully, we’ve given Sauer/Hakes enough to think about here as to question their data and methodology.

***

UPDATE: I decided to look at the 2001 data, which is the one where the OBP parameter is negative.

I set the filter to year=2000, free=1. I expected to see all the free agents. That’s not what I found. This “free” parameter is not valued corrected. For example, Barry Bonds’ contract expired following the 2001 season (free agent entering 2002). But the “free” parameter is set to 1 for every year in his data set (1999, 2000, 2001, 2002, 2003). So, it can’t be used as a stand-alone like that. I don’t know if this impacts their data. I have to also set the freeyear=2001 in order to get my list.

This list is a total of only 13 players, which seems awfully small to me. Maybe something happened in 2001, I don’t know. But, most importantly, this is the year that ARod signed his contract! Talk about an outlier to end all outliers, and then need a regression equation to try to fit that data point, along with free agent Jeff Reboulet who signed at close to the league minimum. Reboulet had an OBP of .325 and ARod of .420. Of course you can’t give too much weight to the OBP. How could you? Plugging ARod into their 2001 equation, and you get an estimated salary of 8.4MM$. Ouch. Reboulet, he of the lower SLG than OBP (hard to do) comes in with a forecasted 762K$ salary to compare to his 450K$.

Regardless of whether I’m misreading the data regarding the counts, I do know that ARod played for the Rangers in 2001, that means that he’s part of this dataset. And since their regression equation seems out-of-whack, and “coincidentally” it also happens to be the largest monster contract of all-time, well, that pretty much explains that, doesn’t it?

***

Here’s their dataset, which they authorized me yesterday to release, but I missed that email somehow. Now, you guys have a crack at it: moneyball-data.xls, and in stata format: lewis9903d.dta.txt (Note that after you download, rename the file to remove the .txt.... it was the only way the software let me upload it.

At the request of the authors, they want to point out that the data is sourced from Lahman DB and Pappas salary data.

Malkovich, Malkovich, Malkovich (15:40:20-05:00)

2012-02-20T15:40:20Z

By Tangotiger

Not since John Malkovich entered his own brain have we seen such overuse of the word Lin.