Where is the research frontier for baseball analytics?

[ashbury]

Anyone who is expert in modern statistical methods may want to have a look at this fellow's master's thesis:

 

https://www.retrosheet.org/Research/Calzada/CALZADA-THESIS-2018.pdf

 

I'm pretty versed in certain areas of math, but am relatively weak in stats and this paper on forecasting baseball players is well beyond my pay grade.

 

He comes from a strong computer science program, so I consider this a little better vetted than the average collection of formulas and tables.

 

I am intrigued by the visualizations he provides, although not able to fully absorb what they intend to impart. If I can't understand the pretty pictures, what hope is there for the actual content.

 

My main point in posting this is to say that the cutting-edge in baseball analytics is a lot farther out there than people sometimes realize. It's not just a matter of whipping together some formulas, tweaking the coefficients to attain some new level of precision. Moreover, the guys who can do this are not inexpensive to hire - this guy's LinkedIn Page suggests to me that a baseball team can't hire him for an intern's wages, which has been my impression of past hiring practices by the Twins in the area.

 

It seems to me that a minimum standard for the Twins' analytics staff is that they have someone who can understand this paper - preferably several who can at least debate knowledgeably on the methods that this author uses. If they can't, they will continue to play catch-up with the teams that can do this. Maybe this paper is where the cutting edge is, maybe it isn't - but I want my Twins to know. Small differences in player evaluation will determine who makes the winning offers for the right free agents and trade acquisitions, and who pounce first during the amateurs draft.

 

If Twins personnel browse through this site, I urge them to pay attention to this paper. The subject matter concerns only major-leaguers, but the methods look like they could be applied to young'uns too, and that might be a bigger payoff.

[nicksaviking]

Finished, that was a nice easy read for my coffee break.

[bird]

Thanks for the link, Ash.

 

I'm going to get to this stuff about writing some lines of computer code and doing some fancy arithmetic right after I read about how teams are taking advantage of the new scientific frontiers in neuroplasticity.

[Parker Hageman]

[Parker Hageman]

 

 

In all seriousness, in conversations with people inside the department, it's clear that there is a significant value placed on the education level of the new hires, not their baseball-related pedigrees. One employee told me they actually prefer to have people who know nothing about baseball but are highly skilled in their trades. 

[nicksaviking]

 

In all seriousness, in conversations with people inside the department, it's clear that there is a significant value placed on the education level of the new hires, not their baseball-related pedigrees. One employee told me they actually prefer to have people who know nothing about baseball but are highly skilled in their trades. 

 

That's going to be pretty divisive I'm sure. I've worked for more than one company though where they preferred to hire people with no experience so they could mold the employee to their vision without the employees having preconceived ideas about how the job should work.

 

Of course that means the organization better have good ideas to teach these people.

[Parker Hageman]

 

That's going to be pretty divisive I'm sure. I've worked for more than one company though where they preferred to hire people with no experience so they could mold the employee to their vision without the employees having preconceived ideas about how the job should work.

 

Of course that means the organization better have good ideas to teach these people.

 

Right, it's more of the right mix than anything else. They do have baseball people -- including Alex Hassan -- within the department. The idea is why limit yourself as an organization to make "knowledge of or experience in baseball" for a data scientist rather than hiring the best minds period. 

[ashbury]

In all seriousness, in conversations with people inside the department, it's clear that there is a significant value placed on the education level of the new hires, not their baseball-related pedigrees. One employee told me they actually prefer to have people who know nothing about baseball but are highly skilled in their trades. 

 

That's going to be pretty divisive I'm sure. I've worked for more than one company though where they preferred to hire people with no experience so they could mold the employee to their vision without the employees having preconceived ideas about how the job should work.

 

Of course that means the organization better have good ideas to teach these people.

This is a complicated aspect. Back in the 70s or 80s, investment firms started hiring Physics majors with no background in finance - the so-called Rocket Scientists. This paid off handsomely, as those noobs brought in a fresh perspective that the old guard didn't have the mathematical chops to carry out.

 

Actual ignorance of the field they're being hired into isn't an actual positive. The idea was that it's easier to teach Finance to someone with a background in applied Calculus, than to teach Calc to the Finance guys. Nowadays pretty much any MBA program involves exposure to the "Quants", and those who go on and specialize in quantitative methods need to be pretty good at higher math.

 

Same here. To a degree, it's probably easier to teach the rudiments of baseball talent evaluation to someone with highly evolved stats skills, than to teach what's in the linked research paper to baseball experts. Eventually, there will be a meeting somewhere in the middle, where future front offices take this particular skillset for granted as necessary to their jobs.

 

My view is that there doesn't have to be friction. The master's thesis in that paper looks like it used data available from SABR, stats concerning measured performance. But, you could just as easily use data from an internal database, involving quality judgements like swing quality, or even just "The Good Face" - even if the numbers have no precision, someone with a 10 on the Good Face scale and another with a 1 conveys some meaning. The statistical tools are good at locating correlations, and it might very well be possible to capture some value out of what scouts have always instinctively felt. If that initiative panned out, scouts would continue to use their expertise to judge prospects by their Face, or (more likely) by their swing.

 

The Parker Hagemans of the world actually should fit in very well with the framework a good analytics team would construct. All the statistical methods in the world won't buy you anything if you lack quality input.

[Jim Hahn]

I didn't read the article, I suspect it to be beyond my former high school math teacher level. I do find this whole analytics thing interesting. Ryan's focus was scouting, and he seemed to keep the analytics dept., really the whole FO, at a bare minimum. Clearly the analytics as well as the rest of FO as been beefed up considerably under Falvey. What is less clear is what is going on with scouting. It has been reorganized and some long time scouts have been fired.

 

I have read little more than that. Has that department been beefed up? What are are the qualifications and backgrounds of the replacements? As was said, analytics is only as good as its input. There are number of qood prospects and young major leaguers who are products of the old regime. Not everything they did was a complete failure as some posters seem to suggest. If someone has a little insight into what is going on with the scouting part of Twins organization, i would like to read about it.

[Parker Hageman]

 

I have read little more than that. Has that department been beefed up? What are are the qualifications and backgrounds of the replacements? As was said, analytics is only as good as its input. There are number of qood prospects and young major leaguers who are products of the old regime. Not everything they did was a complete failure as some posters seem to suggest. If someone has a little insight into what is going on with the scouting part of Twins organization, i would like to read about it.

 

The Twins have eliminated a lot of their advance scouting staff, beginning in the last few years of the Ryan regime, preferring to go with video and data (this is not uncommon). 

 

They still have a lot of good baseball people in place and they've added people like John Manuel who had a wellspring of contacts through baseball. Thad Levine very much emphasizes the scouting aspect of the game -- they just have pivoted how they are attacking it. 

 

 

 

 

[Jim Hahn]

Thank you. I appreciate your response. I had read about Manuel, thought he was more of a FO type than an actual scout, I appreciate the clarification. I did know about the video advance scouting as well. If you have more information on how their amateur scouting has pivoted and you can share it, I would be interested in that.