There’s a fellow up around Ithaca, New York who is spending a lot of time with statistics and how they relate to the New York Yankees. His name is Matthew Frank, and he’s doing some interesting work in the field of sabermetrics. I’ve been intrigued by his data since I began to read his work, and I wanted to find out about the man and discuss his statistical methodology. Dr. Frank was kind enough to share some time with me and answer my questions.
What is your educational background, and what got you interested in sabermetrics?
I received a PhD in Social Psychology and have expertise in research methods and statistics. I currently teach statistics at a central New York college.
I’ve been interested in statistics and baseball for most of my life. I’ve focused on the use of research methods and statistics in baseball for several years now. Like some other people with academic backgrounds, I find statistics and baseball not only an interesting challenge, but a great deal of fun.
Who are your sabermetric influences? Bill James? Eric Walker? Earnshaw Cook?
First, my background is in research and statistical methods, and I have actively worked to develop my own statistical framework to analyze baseball. However, it is absolutely a fact that if James and others hadn't done a lot of the leg work to develop such things as runs created stats, win-shares, ERA+ etc, the field of sabermetrics wouldn't be what it is today. And almost certainly I wouldn't have the advantage, indeed luxury, of following in the footsteps of some major contributors to understanding baseball with a richer statistical perspective than the one that existed for much of the Twentieth Century, without James' and other people's contributions.
Essentially, I have found James' work a great starting point to fine-tune and even further elaborate his contributions. Stats I've introduced like CR+ (created runs plus), DERA (defensive independent ERA+) and MDERA (minor league equivalent defensive independent ERA+) are all variations on work that James began long ago. I hope any small contribution that I have added to his and other people's work, is useful for consumers of baseball statistics.
A guy like Billy Beane put baseball on notice that rational, empirical analyses needed to mature within baseball. That is, if a team on a shoestring budget like the A's could consistently make the playoffs by making solid judgments based on good empirical analysis, then other teams needed to make an adjustment. We've seen that happen over time, where now most teams have people who do number crunching. The key here is to understand what numbers need to be crunched. That is where the real added value comes in, when we look at analysis of any kind.
So by using statistical analysis successfully to compete with bigger market teams, Beane demonstrated the value of good value decisions, and corroborated what James and other sabermetricians were showing by their work. I should add that not all teams will benefit evenly from the contributions of strong empirical analyses. Obviously, once wealthier teams begin to employ such methods they will be able to gain advantages that smaller markets couldn't due to financial limitations. We have begun to see such a development over the last several years with the Red Sox and Yankees.
That team that makes the consistently sound decisions and can deploy the resources to execute a strong plan has the best chance to reap the rewards of the marriage between rational, empirical analyses and financial fire-power. The key as always will be sound baseball decisions.
How much can you tell us about how you’ve developed your systems?
I have not published any of the work I have introduced at nyyfans.com. However, the statistics I have introduced have been tested for their reliability and validity. Since many of my ideas are built on earlier work, the reliability and validity of the measures have components which have already been around and tested.
I’m excited about how each of the major elements of what I have introduced can be easily used to estimate how much a given player actually contributes to winning baseball games. That is so for both pitchers and position players.
I’m excited about the futures of Joba Chamberlain and Phil Hughes. Who isn’t? But please tell us – in layman’s terms – why you think they will post a 150+ ERA?
First it is useful to picture any statistical projection as one looking through an aperture (the aperture width and direction is set by the current conditions of interest - more positive indicators tilt the aperture ever so slightly above the horizontal). Close by the aperture (in the near future) the range of possible outcomes seems fairly narrow, but the farther away one looks the extreme ranges that represent all possibilities get much wider apart. Imagine that the wider range indicates that in the future, farther into the future, it becomes more likely that given the current vantage point (aperture setting), it is still reasonable to envision very positive outcomes (seen at the top-most edges of the range), and it is possible to see an ever wider range of possibilities that will leave the outcome grounded (poor outcomes). Still, with the current information available, the aperture is pointed in the direction where it appears the most likely path appears to be.
So when one does projections based on current information about athletes, you try to aim the aperture so that is dead-center on what appear all possible outcomes.You can see using this perspective that projections, based on time-lines have a very high margin for error the farther one looks into the future. Many things are possible, and can be envisioned as possible outcomes, the farther one looks. However, if one continues to take a snapshot through that aperture, frequently, over time, and the path seems steady, one can gain confidence that the setting of how that aperture is pointed is reasonable (At each point along the way).
Phil Hughes through 2006 produced MLEs that placed him at a solid ERA+ of around 130 for three years running (this is very impressive indeed, and nearly unprecedented for a mere teenager - what Hughes was for that period).Hughes' 2007 season, truncated by injury appears to be merely a blip on the way as it turned out that over his last six or seven appearances (all of Sept and into the playoffs), Hughes ERA+ was over 160. Now some might argue that I am cherry picking that data, but I would like to remind everyone that Hughes was coming back off two injuries, possibly slipped into a few bad mechanical habits, and greatly benefited by the return to his side of Dave Eiland (at the beginning of September).
I've written over the last year, about how important Eiland has been to Hughes over the last several years of Hughes' development. I see no reason to think that somehow the aperture should be lowered on what one should expect from Hughes going forward. As for Chamberlain, I think it is now fairly well accepted among most baseball analytical types, that Chamberlain has special talent. The closest comparable for Chamberlain is Justin Verlander. I compared Chamberlain to Verlander last offseason, and pointed out that Chamberlain was likely to follow along the path that Verlander paved in the minors.
Not only did Chamberlain get to the majors quickly, as did Verlander, so too does Chamberlain have a skill set comparable to Verlander with his great fastball and secondary pitches that are out pitches. Both come to the majors as power pitchers with a fair amount of control.
One can see that if that Verlander comparison continues to hold up, Chamberlain ought to be posting ERAs in the mid-threes on a consistent basis, or even better. I expect that Chamberlain ought to post such results consistently and during his peak performance years (barring injury or physical issues that affect the trajectory of his career path) he ought to have peak ERAs consistently over 140 (I expect an average projection for Chamberlain' career ERA+ given the above caveats to be about 140 or so.
For Hughes I expect his career ERA+ to exceed 150. I am well aware that the number of pitchers who have produced careers like that can be counted on a mutilated hand. That doesn’t change the projection. What could change the projection is if Hughes doesn’t produce ERA+ as good or better than he did those minor league equivalents.
Again, with such predictions they are based on what the aperture's direction and focus appears to be telling me now about both guys. Further snapshots might focus the detail along that path, or indicate a path either more or less impressive than that.
Tell us about Dave Eiland and how/why he will help the young pitchers.
Eiland has worked with Hughes for two years now, and understands how to maximize his talent mechanically. Eiland worked with Chamberlain and Kennedy in 2007 and all three will benefit by having a pitching coach at the ready, willing to jump in and correct mechanical flaws and to remind each of them what they need to be doing to be the best pitchers they can be. The Yankees work hard to analyze pitcher's style, find ways to improve it, work with the pitchers to implement it, and reinforce it through repetition.
The kids in the Yankees farm system have a lot of confidence in what Eiland does with them, and we shouldn't be surprised by the success shown in the minors and the majors by recent Yankee pitching prospects. We can expect more of the same from other pitchers who have benefited from the work of Eiland and Nardi Contreras the organizational pitching guru.
You calculated projections for the Yankee pitchers in 2007 that included an ERA+ of 119 for Mussina and 137 for Clemens. Obviously that didn’t happen. What can you tell us about these projections and how you arrived at them?
Statistical estimates are rarely right on target. They are estimates, not prophecies. Having said that, Mussina’s career ERA+ was 125 going into 2007. Mussina’s ERA+ in 2006 was 125. Expecting Mussina to have a slight drop off from career and recent results was reasonable given his age. However, Mussina had an injury to his hamstring which really affected his conditioning, after the season started. Mussina looked like an old, washed up pitcher in 2007, he might not be.
The Red Sox are making a similar choice right now, when they are considering giving Schilling an incentive laden contract. I think when you have players who could drop precipitously due to expectations about decreasing performance due to age, it might make good economic sense to encourage the use of incentive laden contracts in signing such players.
Clemens came off three straight years where he put up tremendous ERA+ numbers. His career ERA+ going into 2007 was about 140. The short term trend was favorable for Clemens to have a strong 2007 season. Apparently his body was more vulnerable to injury that it had been in the recent past.
Yes, both projections were way off, but I think extenuating circumstances may have had a lot to do with how far off those projections ended up being.
Do you have any thoughts on Shelley Duncan’s projections with the Yankees?
It is encouraging to see that the 27-year-old Shelley Duncan’s major league performance so far is similar to what one would expect from his recent minor league equivalents. If he can hold onto that, he might end up being an important contributor to the 2008 Yankees.
Can you remain objective while measuring the Yankees?
Researchers are trained to be on the look out for bias, and to correct for it. Bias can come in a multitude of forms, yet one tries to eliminate any tendency that one has to base estimates, assumptions, or judgments on anything that can’t be empirically supported. No one is completely objective, and when I have my fan hat on, believe me, I’m no more objective than the bloke at the local bar.
Do you aspire to work within baseball?
If such an opportunity were to present itself I would seriously consider it. I have not yet pursued such employment.
With people like Billy Beane, and J.P. Ricciardi in GM roles, and Paul DePodesta working behind the scenes with the Padres, how do you see the role of sabermetricians in front offices going forward?
I think it will definitely grow. There are so many opportunities to do empirical analyses within baseball, not only on the field of play but throughout an organization, that it is bound to grow. That will be so because there are tremendous benefits for those organizations which identify what systems to study, and who are able to extract meaningful information about those systems. Teams that do not have strong analytical support in their organizations will tend to filter to the bottom of teams within the baseball pecking order. Those teams that have strong analytical departments will be able to compete effectively with organizations with more money that aren’t as strong analytically.
How much has Bill James meant to the Red Sox?
James’ contributions have meant a great deal to the Red Sox. Much more than is generally known. I say this with confidence because James is a very knowledgeable and creative analyst. That doesn’t mean that the Red Sox have an inherent, overwhelming advantage going forward, because even James has limitations about what he knows or where he can apply his analytical resources.
Did you ever have a man-crush on Scott Hatteberg or Jeremy Brown?
Who hasn’t?
Give the readers a piece of info that will really surprise them.
Many scientific discoveries over the next twenty years in both biology and in astronomy will be made by machines with forms of artificial intelligence.
WCBS Newsradio 880 is the flagship radio station of the CBS Radio Network, CBS News and the broadcast home of the New York Yankees. Four decades of broadcasting New York tri-state area news, New York tri-state area traffic, and New York tri-state area weather and backed by the global resources of CBS News, WCBS always has the edge on breaking news and the day's essentials. WCBS consistently reaches over 2 million listeners each week and you can listen on air, online and on demand to WCBS' personalities including Charles Osgood, Andy Rooney, Kim Komando, Anthony Dias Blue, Bob Lape, Charles Grodin, Dave Ross, Wayne Cabot, Michael Wallace, Mary Alice Williams, Jim Tailor, John Sterling, Suzyn Waldman, Pat Carroll, Pat Farnack, Ed Crane, Jared Max, KC Connors, Tom Kaminski and Craig Allen. www.wcbs880.com
E-Mail Story
Print Story
