Archive for September, 2010

Identifying Prospects

Request:

1) What traits or skills do MLB teams scout for, and what do they expect players to develop over time?
2) What traits/skills do teams avoid? How do they estimate injury risk, and do they do this well? Can they?
3) What pitches generate more injuries? It seems that pitchers who throw a curveball more often get injured more (think Ben Sheets, Chris Carpenter, Stephen Strasburg). Is this really the case, and if so, is it worth the risk?

I don’t know exactly what teams look for in players beyond the five tools, tall pitchers, and possibly the “good face.” Kevin Kerrane wrote a marvelous book on scouting in the 1980s Dollar Sign on the Muscle, which follows the lives of several scouts and discusses the characteristics they look for. While there is some agreement over what makes a baseball player good, different scouts and organizations have their own philosophies as to what characteristics mark future success.

I can’t comment on how to predict success based on personal observations (a technique one of my professors referred to as “ocular least squares”), but I have looked for makers for success in minor-league performance statistics. I report my results and explain my methods for predicting success in the Chapter 8 of Hot Stove Economics. I even put a dollar value on prospects using these characteristics.

The difficulty with picking out major-leaguers before they’re ripe is that while most future big-league players excel in the minors, many bad players do as well. Looking beyond the slash stats reveals some common characteristics of big-league players, and some of the stats I found useful for predicting major-league success aren’t necessarily stats that I find to be the most useful stats for evaluating players once they make it. For example, I rarely look at the batting averages and strikeout rates of major-league hitters, but I find that high batting averages and low strikeout rates are important predictors of major-league success. You can succeed in the big leagues with a low average and striking out a lot, but even players who struggle in these areas typically handled the bat much better in the minors. Also important are a player’s walk rate and isolated power. If you have patience and can hit the ball hard, you’re more likely to succeed in the majors that players who lack these skills. And you can’t look at minor-league performances without also accounting for age. A twenty-year-old who’s treading water in Triple-A may have more promise than some of the older guys having success at the level.

Another interesting finding was that the stats below High-A ball have no predictive power. At this level, predicting success requires personal observations of trained scouts.

As for how players skills develop, I’ve done some work looking at how major-league players improve and decline over their careers. For hitters, batting average and power peak in the mid-to-late-20s, but these skills see minimal improvement and decline. The ability to walk improves into the early-30s, but the age-range of peak performance is less than it is for batting average. Pitcher strikeout ability is at its greatest almost as soon as pitchers enter the league, but this ability doesn’t diminish as fast as other skills. Like hitters, pitchers improve in walking into their early-30s. This is likely the result of acquired knowledge that allows older players to succeed, even as their physical athletic skills are deteriorating.

As for identifying injuries, that’s something that is not well-understood outside of baseball. I would hope that teams are conducting their own internal analyses of injuries, but most of that knowledge is kept private. Baseball injury data is just starting to become available where we can look at factors that influence injuries. The research being done in sports medicine journals is good and is still developing. What I have found interesting is that the medical community seems to have a better grasp on youth injuries than it does on adult injuries. For example, I’ve got a study on my desk that looks at factors that impact elbow injuries among youth pitchers—arm fatigue and mechanics seem to matter, but curve balls don’t. Play tracking systems like Pitchf/x, and motion analysis technology like Dartfish should help us better predict and prevent injuries for all players.

Why Are Modern Pitchers So Fragile?

Another request.

Why are modern pitchers so fragile? They pitch fewer innings per game, start fewer games, and have more days rest between starts. In addition they have much better training and medicine to cope with the stresses of pitching.
I have heard it repeated that there will no longer be 300 game winning pitchers. What happened to the Nolan Ryans and the Bob Gibsons?

Unfortunately, I don’t have very good data to examine exactly how total pitches thrown have changed going too far back in time. But, given the pattern going back to the late-1980s, I think it’s safe to assume that the extreme loads of pitchers are declining, even though the average pitching load has remained constant at about 99 pitches per game.
Maximum pitches thrown per game

Using some simpler measure of workloads, innings pitched, the pattern is interesting. The figure below shows the change in total innings pitched in a season over time for the maximum number of innings pitched, and by 95th and 75th percentiles, and the median (minimum 10 games started).

IP

Though there has been a general decline in pitcher workloads over time, there was a bump in the late-1960s and early-1970s, when pitching loads increased over what they where in the 1960s. Since 1962, when the leagues both started playing 162 games a season, there have been 65 pitcher-seasons with 300 or more innings pitched. The last one occurred in 1980 when Steve Carlton threw 304 innings. Below are a table of the number of 300-inning performances by seasons and a list of those performances by pitchers.

Year	Count
1962	1
1963	3
1964	1
1965	2
1966	4
1967	1
1968	4
1969	9
1970	4
1971	4
1972	4
1973	7
1974	8
1975	4
1976	2
1977	4
1978	1
1979	1
1980	1
Total	65
Pitcher		Year	IP
Vida Blue	1971	312
Bert Blyleven	1973	325
Jim Bunning	1967	302.33
Jim Bunning	1966	314
Steve Carlton	1980	304
Steve Carlton	1972	346.33
Jim Colborn	1973	314.33
Larry Dierker	1969	305.33
Don Drysdale	1965	308.33
Don Drysdale	1962	314.33
Don Drysdale	1963	315.33
Don Drysdale	1964	321.33
Bob Gibson	1968	304.67
Bob Gibson	1969	314
Dave Goltz	1977	303
Bill Hands	1969	300
Catfish Hunter	1974	318.33
Catfish Hunter	1975	328
Fergie Jenkins	1968	308
Fergie Jenkins	1969	311.33
Fergie Jenkins	1970	313
Fergie Jenkins	1971	325
Fergie Jenkins	1974	328.33
Randy Jones	1976	315.33
Jim Kaat	1975	303.67
Jim Kaat	1966	304.67
Sandy Koufax	1963	311
Sandy Koufax	1966	323
Sandy Koufax	1965	335.67
Mickey Lolich	1974	308
Mickey Lolich	1973	308.67
Mickey Lolich	1972	327.33
Mickey Lolich	1971	376
Juan Marichal	1966	307.33
Juan Marichal	1963	321.33
Juan Marichal	1968	326
Sam McDowell	1970	305
Denny McLain	1969	325
Denny McLain	1968	336
Andy Messersmith1975	321.67
Phil Niekro	1974	302.33
Phil Niekro	1977	330.33
Phil Niekro	1978	334.33
Phil Niekro	1979	342
Claude Osteen	1969	321
Jim Palmer	1970	305
Jim Palmer	1976	315
Jim Palmer	1977	319
Jim Palmer	1975	323
Gaylord Perry	1974	322.33
Gaylord Perry	1969	325.33
Gaylord Perry	1970	328.67
Gaylord Perry	1972	342.67
Gaylord Perry	1973	344
Steve Rogers	1977	301.67
Nolan Ryan	1973	326
Nolan Ryan	1974	332.67
Bill Singer	1969	315.67
Bill Singer	1973	315.67
Mel Stottlemyre	303	1969
Luis Tiant	311.33	1974
Wilbur Wood	320.33	1974
Wilbur Wood	334	1971
Wilbur Wood	359.33	1973
Wilbur Wood	376.67	1972

Thus, it seems that teams tried to ramp up pitcher workloads just prior to the modern decline. What happened in the 1960s and 1970s that caused an increase in pitcher workloads? Did teams realize the ramp-up was a mistake, which caused the trend to reverse? This was an era of low offense, and the mound was lowered and the designated hitter added as a response. Was there a shift in pitching philosophy or did something structural cause this shift? I’m open to suggestions. The bump may offer a clue.

Aside from the bump, what has caused the declining trend in workloads? Most obviously, the rise of the five-man rotation gave pitchers fewer games to cover. On top of this, teams began to rely more on relievers within games pitched than in the past, going with fresh pitchers in late innings rather than asking starters to pace themselves. The number of complete games has declined continuously since the late-1970s.
CG

I think the decline in pitching loads is less a response to a toughness of pitchers than it is a change in pitching philosophy. Every year, someone is supposedly going to go to have four-man rotation, but then we never hear any more about it. Whether that is because no manager has the guts to stick to a plan that is easy to criticize when injuries that were bound to happen anyway happen, or because the five-man rotation actually leads to better pitching, I don’t know. I have found that days of rest appear to have little impact on performance, so I don’t believe there is any performance benefit from giving pitchers more rest days in a five-man rotation. I’d love to see a team go with a four-man rotation from start to finish.

As for the increased use of relievers, I believe managers have discovered that 100% mediocre arms can be more effective than paced good arms. Teams can increase their chances of winning by going to the bullpen, exploiting match-ups, and pinch-hitting for pitchers. Furthermore, I’ve found some evidence that fewer pitches per game can improve future performance among starters.

In summary, my best guess is that the decline in pitching loads is part fad (four-man rotations) and part innovation (relievers can be better than paced starters).

UPDATE: In my initial cut and paste of the 300-inning pitchers I accidentally left off six seasons at the bottom of the table. I have added the missing seasons.

Determining Awards: Results Versus Performance

I’m starting on blogging requests. This one actually came in before I opened the queue.

How much do you think the Cy Young voters should take into account things like BABiP? On the one hand, it’s not [Tim] Hudson’s fault he’s getting lucky and getting outs, and that is his job. On the other hand, it’s not all Hudson’s performance that has led to all those outs. My tendency is to want the BBWAA to hand out individual awards based on performance and not based on results. How much should results matter over performance, if at all, in your opinion?

Sports awards are kind of silly when you think about it. We use awards for the arts when there are no winners (e.g., Oscars, Emmys, etc.), but sports competitions have winners. Why bother? And the ESPYs? Who watches that? But, for whatever reason, awards have been around for a long time, and it’s fun to argue over who is the best. So, it’s a worthy topic of discussion.

The results versus performance debate is an interesting one. If I was trying to predict future outcomes, then distinguishing true performance from luck in outcomes is of utmost importance. Awards are backwards looking, and in sports competitions all we care about is outcomes. At the end of the regular season, we don’t pick post-season participants based on Pythagorean records or some other luck-sanitizing measure. For individual awards, should the standard be different? On the hitting side, I developed PrOPS to identify when players may be over- and under-performing based on the way they hit the ball. Fortuitous bounces and wind gusts may push a good hitter into the elite category, but I wouldn’t pick a player with a higher PrOPS over a player with a higher OPS to win an offensive award. I think what actually happens on the field matters more.

Here is another example. I don’t think Jose Bautista will ever hit 50 home runs in a season ever again, but should the flukyness of his season take away from the luster? Maybe he’s not in the MVP discussions, but if he was, I don’t think any favorable match-ups, wind-conditions, or other factors beyond his control that may have helped him outperform other players should put him out of contention for the award. The reason I’m reluctant not to single the performance out as an aberration is that an alternate explanation for Bautista’s improvement is that the took active steps to play better. Even if I can specifically identify good luck he benefited from, there is also the chance that there is unidentified bad luck that I am missing.

When it come to pitchers, that analysis gets a little more complicated. Batters do most everything they do by themselves. Pitchers need the help of fielders to get outs. Batting average on balls in play is something that pitchers have little control over, and BABIP is heavily influenced by randomness. I’m a strong-DIPS proponent. I think the evidence is clear that pitchers have very little control over hits on balls in play, and even on extra-base hits on balls in play. When I see a pitcher like Tim Hudson near the top of the league in ERA with a strikeout-to-walk ratio of less than two, I have a hard time treating Hudson as pitching as one of the league’s elite pitchers. Tim Hudson is a good pitcher, and at times this season he has been one of the best, but the award should go to the best pitcher for the entire season. Even if Hudson led the league in ERA, I couldn’t support him for the Cy Young.

It may seem that I am judging Bautista and Hudson by different standards by ignoring luck for the former but not for the latter, but I’m really focusing on different types of randomness. The role that randomness plays for pitchers is different than it is for hitters, because the main metrics that we use to judge pitchers are heavily polluted by factors beyond their control. If a pitcher gets lucky in striking out batters or preventing walks and homers, I’d have no problem supporting a Cy Young campaign, even if I thought there was little chance that he could continue to perform with the same level of outcome success.

So, when its outcomes versus performance, I prefer to focus on outcomes, but I there has to be some accounting for luck. And due to the nature of the luck they experience, I think we have to treat pitchers differently than we treat hitters when accounting for luck.

Taking Requests

The hectic summer took me away from the blog more than I had anticipated, but things are beginning to lighten up to where I can settle back into a routine in October. If you have any topics that you would like me to write on, let me know via the comments, e-mail, Twitter, or Facebook.

On the JQAS Study of Attendance and Winning

The other day, the Wall Street Journal posted a note about an interesting research finding regarding baseball attendance and winning. According to research published in the Journal of Quantitative Analysis in Sports increasing attendance increases the home team’s chance of winning. The finding makes some intuitive sense. If players are motivated and umpires are influenced by a large boisterous crowd, then teams might want to make more of an effort to get fans to the ballpark. I’ve had a few people ask for my opinion of this study, and since I am quite familiar with the paper I will offer a very blunt negative assessment.

When I first saw a draft of this paper as a referee for another journal, I was intrigued by the finding. I reviewed the paper positively, but I still had a few questions that I thought the authors needed to address before the results could be accepted and published according to the quality standards of the journal. A few months later a revised paper was submitted to me, but I was not pleased by the revisions.

In my initial referee report, I suggested using a method that I had used for investigating how umpires are influenced by outside pressure using QuesTec. The authors bizarrely interpreted my suggestion and did something that made little sense. But more importantly, they credited a source that they did not use, and that source happened to be me. Using the citation that I had provided, they stated that in my work I had found that QuesTec monitoring limited racial bias by umpires. I was shocked to read this, because I have done no such research. The authors simply fabricated this. I can only guess their motives: I suspect laziness in an attempt to placate an annoying referee.

If the authors had lied about something so simple and easy to verify, what had they done behind the scene, where numbers can easily be manipulated? I looked at the results more closely, things looked fishy, and the explanations in the text didn’t make much sense. I wrote up my report, in which I stressed the severity of the academic integrity violation and expressed my other concerns about the research. I recommended that the editor reject the paper, and he agreed with my recommendation. In his letter to the authors, he also noted the false citation.

Jump ahead to earlier this year. I stumbled across the paper at JQAS. To my surprise, the paper had been published with the offending text that I had identified in my report. I contacted JQAS editor Ben Alamar to tell him the saga of the paper. I was most upset by the academic dishonesty, but I was also concerned that my work was being cited as finding results that I didn’t find. Dr. Alamar responded promptly and stated that he would discuss the matter with the editorial board.

Soon after I initiated my complaint, I received an e-mail from one of the study’s authors Erin Smith. She apologized to me and stated that the incorrect text should not have referred to my book but to some commentary that I provided on another study of racial bias among umpires in The New York Times. I replied to Ms. Smith that I appreciated her apology; however, this did not explain why the error made it to the JQAS. Ms. Smith was unaware that I was an anonymous referee on her paper, that I was someone who had previously pointed out the error to her, and that I was aware that another journal editor had also pointed out this error to her. Yet, the offending text remained in the paper. Ms. Smith was lying again, and I never received another response from her.

A few weeks later, I received an e-mail from JQAS editor Ben Alamar in which he stated, “I just wanted to let you know that we have finished our review and have rejected the paper based on the incorrect citation of your work.” So, you can imagine my surprise when I read an article on the study in the mainstream media. When I went to the JQAS’s website, I found the paper still published with the following appended.

Erratum
Please note that the following statement has been retracted from Page 4:

“Bradbury (2007) shows racial discrimination is less likely to occur when the umpires are monitored by an electronic pitch tracking system called QuesTec.”

There is no argument in the paper by Bradbury that such discrimination is likely to occur.

The offending passage was removed with explanation; however, this is not what I had been told would be done to rectify the situation. I contacted Dr. Alamar to request an explanation, but he has not yet replied to me.

As you might imagine, I am not particularly happy about this affair. I don’t like academic dishonesty, I don’t like being lied to, and I’d rather spend my time doing other things. The thing that annoys me the most is that none of this should have happened. Ms. Smith could have removed the offending passage, or Dr. Alamar could have just told me that the journal would publish the paper with an erratum. I still wouldn’t think much of the paper or the decision to publish it, but that would be the end of it.

I would also like to note that this is the second time that I have identified serious errors published in JQAS articles (I identified coding irregularities in a paper that claimed to find performance spikes among players included in the Mitchell Report) and nothing was done about it.

Interesting Facts about My Dad

Please excuse my personal post. My father died yesterday after a long battle with Progressive Supranuclear Palsy (PSP).

Tom Bradbury, journalist and advocate, dies

Paul Thomas “Tom” Bradbury, 67: Journalist and ham-radio whiz

»» Interesting Facts about My Dad