Article: Cereal and Statistics

[PeanutsFromHeaven]

I tend to spend my mornings with four things: a bowl of cereal, a cup of tea, my MLB At Bat app and a folded over copy of The Economist magazine.

 

As I gorge on both carbs and information, I'm happy to know that the Twins have room to grow and that seemingly every night someone outdoes themselves (even if someone else does worse).But I'm also happy to balance the Twins in one corner of my mind and world events (like Greek debt defaults and Indian corruption trials) are squarely in another.

 

So it's a little startling when those two worlds actually do collide.

 

At the start of this month The Economist published an article entitled "Every step they take" about the ease with which advanced statistics have permeated the sports world, in particular the plethora of statistics that the MLB Network can show on an average broadcast. From pitch speed and curvature to fielder reactions, speed and routes.

 

None of what they covered will be terribly groundbreaking to the most hardcore of Twins fans. The folks who analyze the team and read amateur blogs know about such statistics already, there's little that I (or a bunch of International foreign affairs nerds) can add to it.

 

 

 

Not the cover of this article, but appropriate.

But what I can say is that when the world of statistical analysis collides with the world of my breakfast table, we've crossed a threshold. This isn't in reference to Moneyball or "advanced statistics" that make oblique allusions to the wider world of baseball knowledge. This is detail. This is substance. This is specific and detailed analysis smack dab in the middle of mainstream media.

 

The Twins may be using more advanced metrics under Paul Molitor, they may not. Terry Ryan may have turned a corner in his evaluation systems, he may not. But when statistical analysis is offered openly to students in Sri Lanka and professors in Peru and bureaucrats in Burundi and shopkeepers in Slovakia, then you are past the "wait and see" stage. You're past the "consider all aspects" phase. Statistical analysis isn't advanced knowledge, it's mainstream, and if evaluations and judgement aren't made on those metrics...well...

 

Put it this way: I'd like the front office of my favorite baseball team to be as well- informed about recent advances in baseball statistics as people half-way around the world who've never seen a game, but who love to read.

 

I hope they are. I hope Ryan and Molitor talk about such topics. I hope I'm underestimating them. And if not...they can always come over to my house for breakfast.

 

Click here to view the article

[BrianTrottier]

"Statistical analysis isn't advanced knowledge, it's mainstream"

 

Comments like this make me shudder. You don't just "learn" statistical analysis by reading the Economist (or TD, for that matter). It takes years of studying and hard work, and even then the professionals are still constantly reading, learning, and updating their tool kits.

[COtwin]

He isn't advocating that TR pull out his calculator and make up a new metric, that is what a degree in statistics is for. He is advocating that the trends that statistical analysis can provide should no longer be poo-pooed by our favorite baseball team. If you are looking for an edge than advanced analytics can provide a bit of that. Data analysis isn't just for Bill James. Anyone who can see a trend in numbers can do it, someone who has years of experience can do it better.

I think he is pointing out that data analytics is moving into many facets of life and that if people managing billions of dollars will risk the GNP of a small nation using it, than maybe we could risk shifting the infield more or make our next free agent signing using a bit more number crunching.

[Linus]

Here is what I don't like about all the discussions of Twins and advanced metrics: the assumption that the Twins aren't using them or are somehow too stupid to comprehend.

 

I like all the information available about our favorite game.  But its not all created equal.  Some of it should be rightfully heeded and some rightfully ignored.  That is the true key in this discussion.

[Seth Stohs]

 

Here is what I don't like about all the discussions of Twins and advanced metrics: the assumption that the Twins aren't using them or are somehow too stupid to comprehend.

 

I like all the information available about our favorite game.  But its not all created equal.  Some of it should be rightfully heeded and some rightfully ignored.  That is the true key in this discussion.

 

I would hope no one would really think like that first paragraph. I know some are quick to write about it or talk about it on the radio, but I would hope even they wouldn't really think and believe it. Of course they use it, and of course they are smart enough to comprehend. 

 

 

[PeanutsFromHeaven]

As always, comments on my articles offer fresh views on my comfortable positions and easy pieces of writing. It's not always easy to hear this feedback, but it's invaluable.

 

I didn't intend to over emphasize the supposed "ease" of using statistical analyses (as BrianTrottier explained), or the "stupidity" of the front office (as Linus suggested), but I can definitely see ways that it reads that way. The biggest issue that I tried to bring out is the public perception of the Twins and Statistics (i.e. TR gives them short shrift--or "poo poos" them, as COTwin notes, but Molitor has started using them), and since data analytics are publicly everywhere, it would be reassuring to have them publicly acknowledged in the Twins. It doesn't *have to be* public knowledge for the Twins to use them, I don't *have to be* informed about the Twins strategies, it's just a topic to consider.

 

As ever, thanks to all for feedback and responses, this is why I'm always happy to post writing here, even if I miss the mark.

[ShouldaCouldaWoulda]

Let's be honest, statistics are usually overwhelming...even for the most of us on here. Sure, most of us understand the value of the newer simple stats, like OPS+ or DRS, but there are just so many micro-stats that are hard to understand the meaning of in the correct context, and how to add all of those together to find future success in certain areas and for certain players. 

 

I do think that, overall, many of us tend to subconsciously see the Twins as "Old School." Part of that is the "Twins Way" type of generalizations over the years. It almost feels like the Twins were ahead of the curve in the past by the "Twins Way" approach, but that has now passed them by. Small things like the Twins seeming to favor scouting and "toolsy" guys in drafts seems to be another area that they are less reliant on statistics or analysis, and sometimes that works out even. Again, the Twins just "feel" like an organization less reliant on statistical analysis. That of course does not mean that it is true by any means. It is the bias that most of us has grown to believe, and we therefore look for confirmations of our biases. I am certainly VERY guilty of this. We might be right even, but we don't know.

 

Comments like TR saying that 5 good SP's give us a better chance to make the playoffs, than do 1 ace and company. He is probably right about that, but that also depends on what he defines what "good" SP is. I still believe that a few stud SP's or at least 1, can be supplemented with fairly good SP's with the same amount of money than it would be to get 5 good SP's. Obviously, combining Nolasco + Santana+ Pelfy money could get us a pretty good SP, and still leave us with options in the minors or cheaper dumpster diving options, which is supposed to be TR's strong suite anyway. But, at least the guy is weighing these things out with some sort of game theory approach that is purposeful, which I can take comfort in. I don't need to agree with it. 

 

I do wonder on how much we are using stats, in what areas, and to what degree. Are we digging deep? Is it all done in general by the same people, or is it split up into micro areas by different people? How is this information being translated and explained to the FO and coaching staff and players? How closely and well do the analysts and staff communicate, or analysis and FO? We don't know any of that, and likely won't know it. 

 

I sometimes wonder if a baseball team would be ran well if it was put together more like a football coaching staff. With specific specialized coaches for defense, hitting, catching, and pitching. I know we have those coaches, but it usually seems like just some random title that guys get placed into. Instead of defensive gurus, or the like. Then, those coaches should their own specific staff of analysts breaking down their area of specialties. The team manager just needs to be explained the simplified explanation of what they are doing, but just let them do there thing, unless you want to overrule certain things. Then, a whole other set of analysis's working with the FO for what players to aim to get or trade for in the draft, international market, FA's, and trades. Not just some analysts that cover it all, and coaches coaching everything in general. Just my thoughts, but that probably made no sense anyway.

[drivlikejehu]

The Twins clearly use statistics, no one has ever questioned that. The issue has been how they use them and to what end(s). For instance, RBI is a statistic, but it would be inexcusable to include it as part of a player evaluation. 

 

The thing is - a lot of the Twins' strange moves would confuse the oldest of the old school just as much as a statistician. If you went back in time to meet with team executives from the 1950s or even 1920s, they wouldn't understand calling up Stauffer either. I'm not sure that things like that, or Jason Bartlett, etc., are about statistics at all. Do you really need wOBA to know that Danny Santana shouldn't be a DH?

[TheLeviathan]

 

For instance, RBI is a statistic, but it would be inexcusable to include it as part of a player evaluation. 

 

RBI shouldn't be a go-to, but inexcusable to even consider?  Going a bit far and a pet peeve of mine with talking baseball stats.  No metric or stat is inexcusable or worthless.  In fact, the more of them we consider and overlap only enrich the context of the evaluation....not diminish it.

[drivlikejehu]

 

RBI shouldn't be a go-to, but inexcusable to even consider?  Going a bit far and a pet peeve of mine with talking baseball stats.  No metric or stat is inexcusable or worthless.  In fact, the more of them we consider and overlap only enrich the context of the evaluation....not diminish it.

 

This is not a correct understanding of statistical analysis. Adding in lots of overlap and noise is actually a bad thing.

 

RBI does not tell you anything new - it's just a noisy, variable combination of other data points. If my only other alternative to evaluate a player was, say, HBP, then I would definitely use RBI, but there is zero reason to use it in the real world for player evaluation.

[old nurse]

RBI does measure how well someone hits with men on base, Sort of like timing someones speed by counting 10001,1002, 1003. Dang, that guy is fast getting to first base.

[jimmer]

 

RBI does measure how well someone hits with men on base, 

The RBI stat itself, by itself, doesn't tell you how a guy hit with men on base. 

 

In fact, if there's a guy at 1B, and the batter gets a hit and and that guy on 1B doesn't score, there is no RBI. The guy on 1B falls under the 'men on base' category, but the batter didn't get an RBI.  So how does RBI measure how well that guy hit with men on when he got no RBI?

 

BA/OBP/SLG% with RISP measures how well someone with with RISP.  You can get an RBI while grounding out to 2B, he didn't hit well with RISP, but he got one.  

Edited May 22, 2015 by jimmer

[TheLeviathan]

 

This is not a correct understanding of statistical analysis. Adding in lots of overlap and noise is actually a bad thing.

 

RBI does not tell you anything new - it's just a noisy, variable combination of other data points. If my only other alternative to evaluate a player was, say, HBP, then I would definitely use RBI, but there is zero reason to use it in the real world for player evaluation.

 

RBI as a statistic certainly has a great many weaknesses, but you could use the statistic as a window into other ones to examine why a player has had such success (or lack thereof) in driving runners.  Perhaps it's your team's lineup construction or other issues, but I don't feel the need to dismiss the statistic entirely.  Even if better alternatives exist.

 

Sometimes the surface level questions "why is this guy struggling to post RBI?" can lead to deeper questions that require better measures to answer.  Ultimately, it comes down to what questions you're asking and the best stat to find the answer.

[Craig Arko]

If you want to find the relationship between multiple variables, it's called multi-variate (or multiple) regression. Probably a second year college course for a Statistics major, maybe third year for someone in experimental sciences. At least back in the last century.

 

http://www.statsoft.com/Textbook/Multiple-Regression

[Richie the Rally Goat]

"Statistical analysis isn't advanced knowledge, it's mainstream"

 

Comments like this make me shudder. You don't just "learn" statistical analysis by reading the Economist (or TD, for that matter). It takes years of studying and hard work, and even then the professionals are still constantly reading, learning, and updating their tool kits.

its baseball, and there are dozens of websites, blogs, YouTube channels, and TV networks that dissect every facet of the game. It's not curing cancer or decrypting the human genome, it's a game.

 

Sure Jack Goin has a far deeper analytical baseball insight than anyone at the economist, or me as a supply chain manager with a degree in manufacturing management, but don't underestimate the statistical understanding of a premier economics magazine.

[drivlikejehu]

 

RBI as a statistic certainly has a great many weaknesses, but you could use the statistic as a window into other ones to examine why a player has had such success (or lack thereof) in driving runners.  Perhaps it's your team's lineup construction or other issues, but I don't feel the need to dismiss the statistic entirely.  Even if better alternatives exist.

 

Sometimes the surface level questions "why is this guy struggling to post RBI?" can lead to deeper questions that require better measures to answer.  Ultimately, it comes down to what questions you're asking and the best stat to find the answer.

 

RBI doesn't really have weaknesses; it is what it is. It counts when a particular game event occurs (a run is batted in). The only thing that can go wrong is for the counting of those events to be misapplied, e.g., by mistakenly believing it should be part of an evaluation of a player's ability.

 

Evaluating the value of a player's offensive output is pretty easy - early versions of linear weights existed a century ago. Computers allowed for much more accuracy, of course, but that was a data issue rather than a statistics challenge.

 

The more complicated statistics aim to infer measurements that can't be made directly - defense, for example. How to approach player contracts can be quite complicated, for another. 

 

 

[TheLeviathan]

 

The only thing that can go wrong is for the counting of those events to be misapplied, e.g., by mistakenly believing it should be part of an evaluation of a player's ability.

 

 

Well, I'll just let it go since you prefer to be condescending, even after you mistakenly accused me of misunderstanding statistical analysis.  I hope the Twins are willing to use RBI if it is relevant to the analysis of a player and it most certainly can be, but need not be for all.

[drivlikejehu]

 

Well, I'll just let it go since you prefer to be condescending, even after you mistakenly accused me of misunderstanding statistical analysis.  I hope the Twins are willing to use RBI if it is relevant to the analysis of a player and it most certainly can be, but need not be for all.

 

I was not condescending. Plenty of fans and commentators put more value on RBI than you yourself do. My point was that RBI itself is not problematic in any way - not something to be 'dismissed,' as you had suggested I was wont to do. It just doesn't work for player evaluation because practically all the relevant aspects of RBI are represented more accurately in linear weights, and the latter doesn't have the same problems with timing and context.

[TheLeviathan]

 

 not something to be 'dismissed,' as you had suggested I was wont to do.

 

"Inexcusable to include" seems a lot like "dismissed" to me. It also seems strange to me to suggest that there is no conceivable type of player analysis in which RBIs might factor in as a relevant variable to consider for pattern or relationship analysis.

 

Perhaps all you meant was that a player's RBI total isn't a very fair or conclusive way to judge a player.  I agree.  It also seems to me that your initial statement was stronger than that.

[DocBauer]

What's the old expression? "There are lies, damned lies, and there are statistics." For the sake of this umpteenth discussion, we could easily change that to; "there are statistics, damned statistics and lies."

 

I don't have a problem with measurement, statistics or any sort of comparison. But I find it odd just how many different stats, measurables and comps have been created for baseball. My goodness, if you just put in the research, you could come up with a hitting statistic for Mauer, on weekdays, not raining, facing RH's with runners on base, less than 2 outs, against pitchers with at least 2 vowels in their last name. Hell, there probably is already!

 

"You know, Mauer was clutch on Tuesdays and Thursday's against all the Smith's and Johnson's of the world, but don't EVER let him face a pitcher on Wednesday or Friday who has 3 vowels in his name. He's .25 below league average with runners on and no outs and the sun shining with the temperature above 55 with those guys on the hill."

 

I mean, it gets ridiculous. If you think a stat is relevant, great. RBI is a great example. Overblown? Irrelevant? Tell that to the manager and the team trying to score more than the other guys.

 

Someone hits lead off and #2 in a lineup. Why? To get on base and help set up RBI opportunities for the next guys to bat. Why does Plouffe hit #4 for the Twins? Because he's good at collecting RBI there. And he's better at it than other guys who DON'T hit there. If they were better, then they'd hit there instead of Plouffe. Are there multiple factors that can influence RBI opportunity and thus also RBI production? Absolutely. But to arbitrarily dismiss a statistic because it isn't perfect or old school is kind of ridiculous. We're never all going to agree on this stuff, but minds need to be open both ways.

[drivlikejehu]

 

"Inexcusable to include" seems a lot like "dismissed" to me. It also seems strange to me to suggest that there is no conceivable type of player analysis in which RBIs might factor in as a relevant variable to consider for pattern or relationship analysis.

 

Perhaps all you meant was that a player's RBI total isn't a very fair or conclusive way to judge a player.  I agree.  It also seems to me that your initial statement was stronger than that.

 

By player evaluation, I meant determining the talent level of a player. RBI does not contribute to that analysis... I believe that to be an uncontroversial view, that you at least seem to mostly agree with.

 

Beyond that, it's probably more of a semantic distinction than anything else. Is an analysis of batting with RISP an analysis of RBI? I would say no, the former being a more granular level of examination, but I suppose you could say that it is.

 

Either way, an MLB organization should be operating at a level of sophistication such that basic counting stats don't add anything to the evaluation.

[jimmer]

http://www.fangraphs.com/library/stats-to-avoid-runs-batted-in-rbi/

 

http://www.fangraphs.com/not/rbi-what-is-it-good-for/

 

 

[TheLeviathan]

 

By player evaluation, I meant determining the talent level of a player. RBI does not contribute to that analysis... I believe that to be an uncontroversial view, that you at least seem to mostly agree with.

 

See I took player analysis as also including analysis of their tendencies.  I doubt there are many multi-variable regressions in which RBIs are valuable, but I wouldn't exclude the stat completely.  It may be a quick and dirty window into a deeper issue in some situations.  

 

But neither of us know what the Twins are or are not doing in order to analyze their players, I just hope they aren't artificially restricting their analysis because something is seen as antiquated.  (Or, worse yet, not antiquated enough)

[drivlikejehu]

See I took player analysis as also including analysis of their tendencies. I doubt there are many multi-variable regressions in which RBIs are valuable, but I wouldn't exclude the stat completely. It may be a quick and dirty window into a deeper issue in some situations.

 

But neither of us know what the Twins are or are not doing in order to analyze their players, I just hope they aren't artificially restricting their analysis because something is seen as antiquated. (Or, worse yet, not antiquated enough)

Well, coming full circle, I meant looking at RBI just as part of a general evaluation, not within some kind of real statistical analysis.

 

I doubt the Twins would do that, but they do make strange public comments at times to explain their moves, so its hard to be sure.

 

We know that the Twins still are behind most clubs in terms of analytical personnel, and a lot of what Goin has talked about relates to scouting enhancements more than statistical analysis.