Good read.

In the sports context, especially baseball where they are still assigning new numbers for things, it's very easy to lose standard solid decision making.

I've often said, analytics isn't having the data-its figuring out which data is important at any given time or situation.  Analytics has been part of the game from the beginning. Just more and different data now. Many SABR discussions about did we discover something new or just learn how to quantify something Earl Weaver just did.

Much angst among fans comes from not understanding that your coach/FO thinks something is more important than what you think.  They usually have more information too.

Thanks for the conversation starter.

I come at things from a business analytics perspective, and this article pretty well encapsulates the attitude I saw across 35 years or so.  An airline for example might use a form of analytics to minimize crew costs when scheduling their fleet for a month, and might have to use the same tools to recover when a snowstorm snarls their schedule.  In neither case would they simply turn off their brains and rely on the output from the computer without incorporating their significant expertise in their own industry; before the advent of software powerful enough to address the problem, they had plenty of heuristics they used to get quite-good solutions.  But in the first case they have the luxury of looking at the minimum-cost solution and seeing whether they can improve robustness or union satisfaction or a multitude of other factors, while in the other case they may be willing to take "any" quick and dirty solution the software offers quickly rather than wait for a truly minimum-cost solution to fix the immediate problem (the algorithms being quite time-consuming if allowed to run to completion), tweaking minor logistics issues by hand if necessary.

A big thing baseball has in common with airlines is fuzzy data and huge uncertainties.  Airlines deal with uncertain demand, can be hit by weather (as mentioned), and small random delays can have a cascading effect on a schedule.

I don't have a similar inside look at what baseball teams do, but from the outside looking in, I don't detect nearly the arrogant analytics point of view that gets attributed to them.  The Twins analytics representative who spoke at a panel discussion at the SABR national convention in August described the team's approach as analytics-assisted.  That's how industry looks at it too.

One fun tidbit from that panel was the insight that after every game, Rocco and the analytics team go over the ways that he deviated from the plan going into the game.  People on the outside call that plan a "spreadsheet" to be disparaging, but the point is that Rocco routinely deviates from it (multiple times per game, it was implied), and the analytics team wants to understand.  Not the other way around.

Of course, I could be projecting my own approach onto what they said. 😀

This is more of a gut feel, but I do feel like front offices, not just the Twins, do lean closer towards “blindly trusting” the numbers. Or that we have just passed through a phase of it. 

For example, exit velocity. The faster a ball travels to a point on the field, the less likely a fielder will get there to field it. Makes sense.

As a result, we see players with hard swings and high exit velocities being scouted and promoted. But maybe the cost is faster swings and more strikeouts. On the Twins, you see it in newer players like Larnach, Julien, Wallner. The cost is faster swings and more strikeouts. Are they coached to just swing hard, let er rip and see what happens? I don’t see the ‘two strike approach’ much anymore.

When these types of guys lose their hitting eye, it gets ugly fast. 

Shifitng gears slightly, you also see decisions taken out of players’ hands. Look at the ‘contact play’ when a runner is on third base. More and more, players seem instructed to just run. Dont even think about it. Just 15 minutes ago, bottom of the 3rd inning tonight, Austin Martin was on third base, Trevor Larnach hit a comeback grounder to the pitcher, and Austin Martin was stranded about 30 feet off of third base. Tagged out. We saw Louie Varland throw out a guy in a similar situation recently. Not even a competitive play.  Player just ran on contact. Players are trusted less and less to use their own judgment. 

It’s not just the Twins. I’d say it’s a majority of teams. 

Amyway.

Rant over—sorry this thread didn’t take off. Thanks for your contributions here and hope for many more!!

On 9/24/2024 at 8:51 PM, ashbury said:

An airline for example might use a form of analytics to minimize crew costs when scheduling their fleet for a month, and might have to use the same tools to recover when a snowstorm snarls their schedule.  In neither case would they simply turn off their brains and rely on the output from the computer without incorporating their significant expertise in their own industry; before the advent of software powerful enough to address the problem, they had plenty of heuristics they used to get quite-good solutions. 

I'm an career aviation guy so this hits the mark for me.  Many, many cases of safety equipment, procedures and design in aviation come from near misses.  A very high percentage of those misses were due to a pilot knowing their aircraft and realizing something was lying to them.  They knew (and know), inside and out, how things worked and from that had the confidence to override a machine.  I fully support all the new avionics systems, for they pay me quite well and provide obvious benefits, but I have significant concern that a sort of dependency on them is the primary flight safety concern we have going forward.

I have similar concerns with education in a general sense.  I don't have kids in the systems but I can't shake the feeling that "you have a calculator in your pocket crowd" are winning out.  Automate all you want, someone still has to know how things work under the hood.  Basics still need to be taught, drilled even.  Nobody can discover anything new if they run around re-discovering things known in the 15th century.

Very interesting to listen to the SABR interviews, I missed those.  The interview with the Twins guys was great, I'm more positive on this front office than ever.  They are certainly saying the right things when it comes to using the data.  I'm not surprised, as they have the correct motive of winning games.  I suspect most teams are that way.

Where its going wrong is that most of the people we hear talking about numbers have a different motivation. 

I've been giving the Rates and Barrels podcast a couple month try lately.  I just stopped listening as I was thinking about this post and the nagging issue I've had with them.  It is the absolute nerdiest podcast you could imagine.  There is no other place I can find to slop around in super deep analysis and I love about 90% of it. 

The other 10% makes it unlistenable.  Everything ultimately comes down to how it effects their fantasy baseball teams.  I could not care less about fantasy baseball.  I want to know how the numbers effect the game of baseball and they have different motivations.  Their judgement with numbers is clouded looking for something different than the game demands.  Then, fans and media get stirred up over different priorities.  The fantasy implications on pro football has basically made me stop watching.  Play the game, not fantasy numbers.  It's a perverse flip.

Speaking of stirred up media, the new (8-25) Gleeman and the Geek podcast features a fully plumed peacocking Gleeman, who obviously told us so.  I generally like the listen, but Gleeman has a real blind spot on anything payroll related.  Talking payroll, he'd fit right in here.  There was a moment in this show where he was listing all the ways he was right about the collapse coming where he paused, almost realizing what he was saying.  He mentioned the whole team not hitting multiple times and he caught himself realizing its been the hitting for two months, which was not anything like what he was predicting.  Nevertheless, he persisted.  It's all because of payroll, still, whatever. 

To his credit, I'm pretty confident that when he calms down a bit he'll put all the blame for the collapse where it belongs, mainly the hitting group.  Nobody predicted that, at least not to this degree, but the narrative he's married to is currently preventing him from getting to using the correct numbers.  He honestly should be on a different team beat but the thought of him moving, and telling the story about thinking about moving is hilarious.  He holds some sort of younger fandom pain that he can't get past.

I may be thought of as some sort of contrarian in these parts, and while that may be true by result, its certainly not intentional.  Everything I comment on is something I'm trying to understand from Rocco or the FO or ownerships chair.  Having sat in similar chairs to some of them, I'm thinking about different things than most.  The conversation sometimes gets confused where I have to clarify what I would do vs what I think they are doing but I generally start from their angle. 

I do that for the simple reason that I can accept they have more information than I do and confidence in them to use it correctly.  This allows me to not get too bent out of shape about anything.  Current result notwithstanding, I still see the trajectory of this franchise as very much up and sustainable. 

I'm well aware the way I think isn't compatible with the modern sports world, at least the way the media and sites like these drive it.  It was good to hear the Twins guys say that most of their data stuff is proprietary but the inability and sometimes unwillingness of the media to discuss the numbers in a manner similar to the way the teams are trying to use them is a major disconnect.

21 hours ago, Hosken Bombo Disco said:

This is more of a gut feel, but I do feel like front offices, not just the Twins, do lean closer towards “blindly trusting” the numbers. Or that we have just passed through a phase of it. 

 

Rant over—sorry this thread didn’t take off. Thanks for your contributions here and hope for many more!!

After listening to the team talks at the SABR convention,  I feel pretty confident they are not anywhere near blindly trusting any numbers.  They clearly stated they are using mostly proprietary systems and having debriefs with Rocco about why things didn't work.  In simple terms, they are making it up as they go.  Unfortunately, if they get something wrong we all see it work out very publicly.  Looking back over the last few years, their course corrections are generally pretty good. 

And I can't speak for Ash, but some threads don't need to take off.  Sometimes we need to have good thoughtful discussions among ourselves.  😀

On 9/27/2024 at 3:40 PM, Jocko87 said:

And I can't speak for Ash, but some threads don't need to take off.  Sometimes we need to have good thoughtful discussions among ourselves.  😀

Quantity plus quality is always desirable.  But of course I'll take the latter over the former.

Also of course, "thoughtful discussion" is in the eye of the beholder.  If in disagreement with my position - wait, nah.  😀

In my opinion...

1. All teams are using analytics. How each individual team weights individual data is most likely the differences from organizations in terms of application of the data. 

2. The data is important but it is yesterday's news. I'm sure that there are moments when it can be predictive but a pitcher can render previous data useless by learning and perfecting a new pitch or a batter adjusting his swing for better timing. All teams use data but all teams also employ a boat load of coaches that are employed for the purpose of making players better and confounding that previous data rendering it useless at least temporarily. Data and coaching must be married together.... it's goofy to ignore one or the other. 

3. When people say Rocco only cares about the spreadsheet in an attempt to vilify him. The first thing that pops into my head... how do they know what is on the spreadsheet. How can they be sure that Rocco didn't ignore the spreadsheet instead of faithfully follow it? 

On 10/3/2024 at 12:59 PM, Riverbrian said:

ignore the spreadsheet

ignore the spreadsheet?

On 10/3/2024 at 12:59 PM, Riverbrian said:

In my opinion...

Another great post rb. 

I like your point about what a pitcher can do to confound the best laid plans of analytics. Think about how many runners Pablo Lopez (or say Jose Berrios) can leave on base. “They shouldn’t be as good as they are! How are they so lucky?” It is just an instinct for pitching there, for how to get a batter out when needed, for what pitch to throw in any given situation. Or how & where to throw a given pitch that the catcher calls, or is called by the dugout. Especially getting batters out like Wallner, Larnach or Julien (sorry if any of those three are anyone’s favorite.)

My gut tells me this is part of why offense is down; enough pitchers who know how to pitch on a gut level. Greg Maddux was the best. Bailey Ober has this skill too. If Paul Skenes has this instinct, combined with his physical talent and stuff, he will be this generation’s Clayton Kershaw or Pedro Martinez.

“gut” can’t be measured of course; but neither can “hotness” (talking about the 2024 Detroit Tigers, not Ryan Gosling 🙂)

Pure baseball talent will be cool again someday.

Lots still to talk about. We are not done with this subject. 

12 hours ago, Hosken Bombo Disco said:

ignore the spreadsheet?

This was a perfect post. 

18 hours ago, Hosken Bombo Disco said:

ignore the spreadsheet?

Is this shorthand for "you can't handle the truth", or some other part of the movie?  I don't have encyclopedic knowledge of that lengthy scene.

3 hours ago, ashbury said:

Is this shorthand for "you can't handle the truth", or some other part of the movie?  I don't have encyclopedic knowledge of that lengthy scene.

I asked "When people say Rocco only cares about the spreadsheet in an attempt to vilify him. The first thing that pops into my head... how do they know what is on the spreadsheet. How can they be sure that Rocco didn't ignore the spreadsheet instead of faithfully follow it? 

Hosken Bombo Disco brought up a perfect scene from a Few Good Men. 

Jack Nicklaus playing the character of Colonel Ketchup is being grilled by Ted Cruz playing the role of Lt. Coffee. 

With Nicklaus on the witness stand.  

Lt. Coffee: Any Chance Howie Kendrick ignored the spreadsheets? 

Colonel Ketchup: Ignored the spreadsheets? 

Coffee: Any Chance he forgot about it? 

Ketchup: No

Coffee: Any Chance that Howie Kendrick left your office and said the Analysts are wrong? 

Ketchup: No

Coffee: When Howie Kendrick spoke to the team and ordered the players to pinch hit for Larnach... any chance they ignored him?  

Ketchup: You ever played a major league baseball game, son? 

Coffee: No Sir

Ketchup: Ever faced a 98 mile per hour fastball? 

Coffee: No sir

Ketchup: Have you ever placed your bat in another man's hands and asked him to put his in yours. 

Coffee: No sir

Ketchup: We follow analytics, Son. We follow analytics or people strikeout. It's that simple... Are we clear? 

Eventually, this leads to "You can't handle the truth. 

Alan Arkin is a great screenwriter. Perhaps the best. 

I guess analytics is stupid then. Case dismissed.  Didn't realize it would be so easy.

3 hours ago, ashbury said:

I guess analytics is stupid then. Case dismissed.  Didn't realize it would be so easy.

It was meant as a joke, nothing more. Riverbrian saw what I was going for. Not even a put down or snide remark, just an opportunity to be silly for a moment. Honest. 

10 hours ago, ashbury said:

I guess analytics is stupid then. Case dismissed.  Didn't realize it would be so easy.

My apologies... I didn't intend for this reaction. Hosken made me laugh... after that I was just being me. 

It's an interesting discussion... To get it back on the rails. What each team prioritizes interests me.

Every team has data... to not utilize analytics would be a big disadvantage... every one has the same data... which limits the advantage though.  

There is so much data that any attempt at a composite is just going to leave a huge pile in the middle. They will have to do some picking and choosing to pull players out of that middle.

So I imagine that what teams emphasize is going to cause variance from team to team.

That could be why the Twins would pinch hit in the 3rd inning when other teams don't. Even though everyone has the same data. 

On 10/3/2024 at 10:59 AM, Riverbrian said:

In my opinion...

1. All teams are using analytics. How each individual team weights individual data is most likely the differences from organizations in terms of application of the data. 

2. The data is important but it is yesterday's news. I'm sure that there are moments when it can be predictive but a pitcher can render previous data useless by learning and perfecting a new pitch or a batter adjusting his swing for better timing. All teams use data but all teams also employ a boat load of coaches that are employed for the purpose of making players better and confounding that previous data rendering it useless at least temporarily. Data and coaching must be married together.... it's goofy to ignore one or the other. 

3. When people say Rocco only cares about the spreadsheet in an attempt to vilify him. The first thing that pops into my head... how do they know what is on the spreadsheet. How can they be sure that Rocco didn't ignore the spreadsheet instead of faithfully follow it? 

On 10/4/2024 at 6:48 PM, Hosken Bombo Disco said:

ignore the spreadsheet?

Apologies accepted of course, and I'm sorry that I blew up at an attempt at a joke (which I confess I still don't quite see the purpose or relevance, but moving on...)

One problem with blog posts is the TD messaging system doesn't let me know there are replies to the main entry, only the replies to my own replies.  Now there's a bunch of content to wade into and address, since I last looked.  So I'll reply to one reply from each of you - that way TD will inform you both - but I'll cover more territory than just these.

What I've tried to do is to stake out a middle ground where it comes to analytics.  The article that I referenced in the OP was a little unusual, in that it was in regard to nothing like baseball analytics, nor business analytics, but instead a rather more esoteric topic ("effective altruism" if I have the gist of it correctly), which I don't pretend to understand in its own right - the author uses terminology that is probably familiar to philosophy grad students or economic theorists, both of which I ain't.  But that's what made it interesting to me: the author still advocated for a middle ground where it comes to statistical methods in support of arguments in that field (whatever the heck that field actually is).

So... to address Brian's post above, item 1 draws no reaction from me since we're in agreement.  But bullet 2 seems to simply ignore what has been said.  It's a strawman to think that analytics people develop some kind of result, package it with a bow on top, present it to whoever is supposed to receive it, and then go on vacation for 11 months and come back in a year to solve some other problem for all time, and collect a fat paycheck for their effort.  No, analytics is a constant, and iterative, process.  Do you really think that a team of number-crunchers will overlook the fact that performance is constantly changing?  That the underlying facts on which the study was based could have changed overnight?  Airline analytics teams may develop a forecast of demand for individual routes for the coming year, so that they can devise a strategic plan for aircraft and crews - and then as the year goes by, they revise the tactical side of the plan as demand rises or falls in a given city or a specific route - and at an even smaller level they devise recovery plans due to weather.  Though I have less insight provided by insiders where it comes to baseball, it still seems impossible that the Twins analytics team develops a static picture of, say, Pablo Lopez, and then does not take into account if he develops an effective new pitch like a sweeper, or simply that the results keep changing when he pitches.

But baseball being a dynamic game, of pitchers adjusting to batters and batters adjusting to pitchers and the "book" on rookies and veterans alike constantly evolving, don't you think that the forecasts based on past data kind of take this into account anyway?  The romance of sports is strong, and we all love a story where we learn that Lopez's recent success is due to his fancy new sweeper. But we've all seen cases where someone's success spawns an article about why it happened, only to see that very same player lay an egg the very next game after the article.  Yes, the forecasts may not have a very long shelf life if used at an extremely granular level to predict what Pablo Lopez or Alex Kirilloff or Manuel Margot will do in their next game or next at bat.  You can't predict the next roll of the dice either, but that doesn't mean you discard past events, and maybe your analysis even identifies that the dice being used today are loaded somehow.

The third bullet item about ignoring the spreadsheet is spot on (other than my detestation of the word which is used to imply something different than what goes on at decision-making time).  The analytics team has stated publicly that Rocco deviates from the plan going into a given game all the time, and they specifically discuss those changes with him afterward, which in turn shapes their planning for the next game or whatever

I don't know how deviating in certain ways turns into ignoring the spreadsheet.  And I especially don't understand how that in turn becomes "You. Can't. Handle. The. Truth."  Nor why it's a perfect post.  I guess I should let it go, but I like a good laugh, sometimes even at my own expense*, and I feel left out on this one.

As for HBD's followup, regarding runners left on base, I can't find where on b-r.com there are stats for LOB for individual pitchers.  There are team LOB on the main page for each major league season (the "MLB Summary" link).  LOB is such a mixed bag anyway - I don't see a strong correlation of it to high OR low scoring totals - the White Sox had a putrid offense this year and rather low LOB, but not as low as the Royals or Guardians (among a few others) who scored more runs per game. Their pitchers on the other hand did lead the majors in LOB, but they still gave up a high number of runs because they were putting runners on base in the first place. And HR are a complicating factor, because they clear away the LOB.  But I don't know how to find that kind of information for individual pitchers - which seems strange - so I can't evaluate whether Lopez and Berrios really do "bear down" when runners are on base to any degree that other pitchers don't.  When in doubt, I don't assume any mystical abilities that can't be measured - certainly that marks me as being of the analytical mindset - and what we think we see in one game can be negated in the next.  Sonny Gray had a great, clutch performance in the post-season last year, except he couldn't follow it up the next time out, whereas Lopez was nails in both his outings, and the realist in me says to treat them both as small sample size outcomes from two very smart and brave pitchers.  Anyway, long way around, I don't know whether or not Lopez and Berrios are wizards with LOB, with better gut and instinct than usual, and I'm not especially willing to grant that they are, in order to follow a discussion on that tangent.  What I do lean toward is that LOB correlates strongly with WHIP at the team level (where b-r.com offers easy access) - if you put men on base at a higher rate than average, then you probably will leave more on base when the third out is recorded, unless (there's that complexifying factor again) you give up a lot more HR than average.  The Mariners and Tigers had lower WHIP and also lower LOB than any other teams. (Both did at least okay with HR, so I just dunno.)

I think I wanna take a step back from the tangents, though, and go back to the article I linked to. To me, it covered several points, which IMO serve to knock down a few straw men arguments.

• "The ‘All or Nothing’ Assumption".  The author says that people get annoyed by language that hedges (“very rough guess,” etc).  Folks act as if an analytics team hands down absolutes and treats data as facts.  That's a straw man.  Baseball analysts are keenly aware that the data itself contains all kinds of traps, not only because the data itself could be inaccurate but because the data involves human beings striving in an environment purposely designed to challenge them.  It's difficult hitting a round ball with a round bat, as they say, and because both opponents are trying their best I don't believe in "luck" but I do believe that results may be less, or more, "reproducible."  In the words of Ebby LaLoosh, sometimes you win, sometimes, you lose, sometimes it rains.  Even minor league data has some predictive value because at even the lowest level, you have young men matched up against other young men of approximately equal innate talent and experience, and who manages to achieve more might be likelier than the next player to achieve more at the next level too, given that the abilities can continue to develop and the amount of experience increases.  But such results contain unknowable factors such as one player or another working under a handicap such as a nagging injury or a crisis at home for some reason.  Forecasting methods look for trends within piles of uncertain data, and those trends are thus themselves uncertain so any competent analyst is pretty humble about touting such things as "truth."  It pleases folks to portray such other folks as "arrogant," when in my experience most of them are anything but.
• "Practical rationality requires good priors".  This one's probably not so controversial or subject to strawmen.  If a result looks revolutionary, or even just ground breaking, check it again.  And again.  It's probably wrong for some reason you overlooked - a hidden bias in the data (or your way of analysis) is often the culprit.  Failing to try to poke holes in your own result can come across as arrogant, when it's finally exposed, but probably is more in line with simple human frailty and wanting to do something important.  That's where having a team, not just some unique guru, can be put to good use, even if you accomplish half as much because you're required to spend half your time sanity-checking a fellow worker's results.
• "It’s also quite possible to start with accurate or reasonable numbers, and be led to utter insanity due to doing the wrong things with them."  Probably my biggest critique of the Twins' analytics team, strictly outside-looking-in, is what seems a reliance on hiring a lot of college grads with some learning in basic stats, versus sprinkling in also a few expensive guys with training in really advanced statistics.  I have a bias toward MBA types, not the general business students from a one-year program with a smattering of accounting, marketing, and so forth, but the so-called Quants or Rocket Scientists who really dug into the relevant fields of analytics during two years of study that go deeply into statistics but also related fields.  Wharton, Sloan - those kinds of credentials aren't fake, but they're expensive if you're not bringing in someone as an exec.  These people not only know a lot of stuff that can be used to guide a team of generalists, relatively speaking, but they also have developed a network of people they can reach out to when trying to solve a particularly knotty problem.  It's still possible to do "the wrong things with" data, but knowing a wider range of techniques than the average person, especially the newest methods (which means hiring someone like that every few years) can go a long way.
• "Don’t Blindly Dismiss All Numbers".  This is the other side of the cautionary messages for the analytics folks themselves, and represents the middle ground that other folks seem to find offensive.  I don't know what else to say.  Extremism at either end of the spectrum usually strikes me as wrong.
• "Anyone more quantitatively-inclined than I am is a blind number-cruncher; anyone less inclined is stupid." This is part of a longer section on Fallacies, all of which are good to remind oneself now and then, but this one seems particularly relevant at TD.  And perhaps it's the one that a middle-of-the-roader like me is especially prone to, LOL.  Just because you staked out some mythical middle-ground doesn't make you right either.  But in any case we could do with less name-calling - and "spreadsheet" is basically "blind number-cruncher" in different clothing.

Anyway, I held off until now on delivering my own synopsis of the article, to see what discussion might evolve organically, but there's my starter set.

* A year or two ago my daughter told me, "nobody cares about your inner monologue," in response to some rambling story I told in a family gathering, and it was a good antidote to whatever I happened to be spewing at the time. 😀

I am mostly neutral on the quantitative & hard data side of analytics. I trust that teams are finding data that helps them. My guess is that front offices have instructed analytics people to seek exit velocity (EV) and see that as their holy grail, with everything that goes into achieving a high EV and the benefits that come from putting a ball in play with high EV. I am also observing costs and downsides of chasing EV. I just see a lot of teams that all kind of look like each other (to me) in MLB today. Kind of like all those cookie cutter stadiums built in the 50s and 60s. The same, but different—but basically the same.

Another guess with my EV theory is that teams mostly know what the other teams are doing. “That team is going for EV, and so are we.” Different from Oakland 20 years ago and Oakland’s search for on base percentage (OBP), when other teams didn’t really pay heed to it. 

The irony Oakland and the Moneyball story is that Michael Lewis didn’t really write much about the Oakland pitching staff in those years, which was very good. The story was mostly about the OBP. 

With my theory about teams prioritizing EV, I also concede that I am likely to be wrong or only seeing a small part of the picture. 

With all that said, I am more interested in the qualitative softer side of analytics, questions like “how did the Detroit Tigers sustain that hot streak in the end?” or “can pitch framing be coached” or “why is Jhoan Duran so poor in non-save situations” or “did Royce Lewis’s slump coincide with the Twins asking him to play second base?” or “why did all the Twins infielders except for Carlos Correa all go into a slump at the same time?” And so on, 

Are these valid questions for an analytics department today? Is today’s analytics asking them?

4 hours ago, ashbury said:

So... to address Brian's post above, item 1 draws no reaction from me since we're in agreement.  But bullet 2 seems to simply ignore what has been said.  It's a strawman to think that analytics people develop some kind of result, package it with a bow on top, present it to whoever is supposed to receive it, and then go on vacation for 11 months and come back in a year to solve some other problem for all time, and collect a fat paycheck for their effort.

I'm glad we agree on Bullet Point 1 and on bullet point 3... I also don't like the use of spreadsheet because it is used disdainfully around these parts.  

I'm a little surprised that you have a dispute with bullet point 2. It's really not a strawman at least not intended to be. I was very careful to use "predictive" "previous data" in my post.

Of course your team of analysts are still working. I would never suggest such things. But... let me list two thoughts for your consideration so we can try to remove some straw. 

1. Small Sample Sizes and the Long Game. 

2. Roster Adjustment difficulty. 

That predictive data is used to compile your roster. Rosters are set in the spring. Once set... adjustments are hard to make because the decent replacements are on other rosters and those not on rosters are most likely not on a roster for a reason and you are not going to be in a hurry to make those adjustments even if you could... because you shouldn't adjust on small samples.

Of course, the analytic staff is still working. The information that they compile is important because it can be passed on to hitting and pitching coaches in the moment . The slider is eating you up... Let's try this. Your changeup isn't moving and has lost it's effectiveness. Let's try this. If the player is capable of adjustment... he just might confound previous data. 

However, you can't make that predictive data (that built your roster) work for you in a small sample. You have to play the long game. Margot isn't going to be released after producing a .508 OPS in April. It was the data that brought him here. It's the increasing of that sample... the long game... that keeps him here still pinch hitting in September despite 0-30. Meanwhile... someone somewhere... don't ask me who because I don't know... but someone... somewhere... there is a player in AAA who is cranking out 4 digit OPS because he fixed a hole in his swing... because he is listening to his coaches who are still coaching him... yet he remains frozen for lack of opportunity while the organization waits for the Margot numbers to arrive... and often times... they just don't arrive. 

All teams have this quandary to reckon with... although... some are more stubborn than others. This is where I usually come in conflict with whatever is guiding the decision making. I admit that I get a little impatient here but... I come by it honestly because I never want to see another Logan Morrison again. In the lineup every damn day. 

6 hours ago, ashbury said:

"The ‘All or Nothing’ Assumption".  The author says that people get annoyed by language that hedges (“very rough guess,” etc).  Folks act as if an analytics team hands down absolutes and treats data as facts.  That's a straw man.  Baseball analysts are keenly aware that the data itself contains all kinds of traps, not only because the data itself could be inaccurate but because the data involves human beings striving in an environment purposely designed to challenge them.  It's difficult hitting a round ball with a round bat, as they say, and because both opponents are trying their best I don't believe in "luck" but I do believe that results may be less, or more, "reproducible."  In the words of Ebby LaLoosh, sometimes you win, sometimes, you lose, sometimes it rains.  Even minor league data has some predictive value because at even the lowest level, you have young men matched up against other young men of approximately equal innate talent and experience, and who manages to achieve more might be likelier than the next player to achieve more at the next level too, given that the abilities can continue to develop and the amount of experience increases.  But such results contain unknowable factors such as one player or another working under a handicap such as a nagging injury or a crisis at home for some reason.  Forecasting methods look for trends within piles of uncertain data, and those trends are thus themselves uncertain so any competent analyst is pretty humble about touting such things as "truth."  It pleases folks to portray such other folks as "arrogant," when in my experience most of them are anything but.

As Mike Sixel has been saying recently… It’s a dial… not an on/off switch. I don’t ask for absolutes because they can’t be given… especially in baseball. You have a 36% chance of this guy reaching base compared to a 34% chance of this guy reaching base. If the 66% happens to rear its ugly head. People tend to act like that 36% guy was 100%. I find it odd that so many can expect the level of perfection that they do. I’m sure that auto-mechanic who is pissed off because Jeffers struck out with one out and the bases loaded has never made a mistake in his own life… but in his quest for perfection from others… I’d ask him to at least consider if he would be so mistake free in the garage if other auto-mechanics were trying to knock the wrench out of his hands while he is trying to get his job done. If I was looking for a General Manager replacement. One of the first things that I would look for is the humility to know without a doubt… that it is an inexact science. I want the candidate to know that they are going to be wrong often… I want them to know that it’s not what they do but what they do after. If a GM candidate is giving me absolutes. I’d tell him to run for political office… because the voters eat that absolute crap in big spoonful’s even though it’s nowhere near reality.

6 hours ago, ashbury said:

"Practical rationality requires good priors".  This one's probably not so controversial or subject to strawmen.  If a result looks revolutionary, or even just ground breaking, check it again.  And again.  It's probably wrong for some reason you overlooked - a hidden bias in the data (or your way of analysis) is often the culprit.  Failing to try to poke holes in your own result can come across as arrogant, when it's finally exposed, but probably is more in line with simple human frailty and wanting to do something important.  That's where having a team, not just some unique guru, can be put to good use, even if you accomplish half as much because you're required to spend half your time sanity-checking a fellow worker's results.

Playing Devils Advocate. Very rarely do I see a new plan work right out of the box. Typically… It makes its debut… followed by the originally unplanned necessary corrections to the plan. Granted… I’m not talking about space travel with highly combustible stuff around human beings in space suits. They should probably test that kind of thing over and over and over again.

7 hours ago, ashbury said:

"It’s also quite possible to start with accurate or reasonable numbers, and be led to utter insanity due to doing the wrong things with them."  Probably my biggest critique of the Twins' analytics team, strictly outside-looking-in, is what seems a reliance on hiring a lot of college grads with some learning in basic stats, versus sprinkling in also a few expensive guys with training in really advanced statistics.  I have a bias toward MBA types, not the general business students from a one-year program with a smattering of accounting, marketing, and so forth, but the so-called Quants or Rocket Scientists who really dug into the relevant fields of analytics during two years of study that go deeply into statistics but also related fields.  Wharton, Sloan - those kinds of credentials aren't fake, but they're expensive if you're not bringing in someone as an exec.  These people not only know a lot of stuff that can be used to guide a team of generalists, relatively speaking, but they also have developed a network of people they can reach out to when trying to solve a particularly knotty problem.  It's still possible to do "the wrong things with" data, but knowing a wider range of techniques than the average person, especially the newest methods (which means hiring someone like that every few years) can go a long way.

I can’t speak on the resumes of the Twins staff… but that entry level compensation for these positions is going to bring in entry level. In consideration of the important work being done in finding and developing a better baseball player.  I have often wondered if baseball franchises have it backwards by pursuing expensive free agents and relying on interns in the front office. Maybe they would get more bang for their buck to change the pay scale for front office personnel to bring in the best and the brightest analysts, scouts, coaches. By doubling their salary to entice the best and the brightest… they might produce an extra one, two or three better baseball player and that just might pay for itself. Just a wonder in my head.

To take a left turn from that… When I think of “starting with reasonable numbers and the wrong things being done with them”. I think about that focus on the left handed batter vs the left handed pitcher. Too much focus on the 25% of left handers who throw pitches. They should be focusing on the 75%. To me... that's taking accurate measurable data and getting the application completely backwards. 

6 hours ago, ashbury said:

"Don’t Blindly Dismiss All Numbers".  This is the other side of the cautionary messages for the analytics folks themselves, and represents the middle ground that other folks seem to find offensive.  I don't know what else to say.  Extremism at either end of the spectrum usually strikes me as wrong.

• I’m a centrist in most everything. I’m such a centrist that my first thought is… Who is Richard Y. Chappell? He writes logically but I don’t know who he is? It’s hard to pull me a direction. I’m a centrist.  Politically… I have no yard signs. I prefer to stay out of manipulation because manipulation is all it is because manipulation is what moves the needle. I’ve had many conversations (not willingly) with passionate participants on both sides and both sides scare the pants off me. It’s black and white to both sides. It’s almost life and death… they speak in absolutes. These people are going to vote… they are going to decide. The people who act like a grand canyon of real estate exists between 83 Wins and 87 wins scare the pants off of me. They will also vote and I have no pants. Like Mike Sixel has been saying lately… It’s a dial… not an on/off switch. Like I’ve been saying for a long time. The Margins are THIN… Paper thin.

6 hours ago, ashbury said:

"Anyone more quantitatively-inclined than I am is a blind number-cruncher; anyone less inclined is stupid." This is part of a longer section on Fallacies, all of which are good to remind oneself now and then, but this one seems particularly relevant at TD.  And perhaps it's the one that a middle-of-the-roader like me is especially prone to, LOL.  Just because you staked out some mythical middle-ground doesn't make you right either.  But in any case we could do with less name-calling - and "spreadsheet" is basically "blind number-cruncher" in different clothing.

I have been wrong before. I will be wrong again. I am under the impression that those on the extreme edges are wrong more often but will not admit or recognize being wrong. But who knows… I could be wrong about that.  

16 hours ago, Hosken Bombo Disco said:

 

12 hours ago, Riverbrian said:

 

I appreciate the flurry of activity, and want to say that I'm on the road starting today (Arizona Fall League, baby!!!!) so I might not get around to giving your thoughts the time they deserve for a little while.

Also, I might not have made clear in my lengthy previous response that, while I did respond to some or all of your comments, the second half (with the bullet items) was about the article I linked and was not directed toward anyone in particular at all.  I don't know who Richard Y. Chappell is either, except he seems to be a philosopher who somehow dabbled in this. 😀

On 10/6/2024 at 8:55 PM, Riverbrian said:

Playing Devils Advocate. Very rarely do I see a new plan work right out of the box. Typically… It makes its debut… followed by the originally unplanned necessary corrections to the plan. Granted… I’m not talking about space travel with highly combustible stuff around human beings in space suits. They should probably test that kind of thing over and over and over again.

Even though I’m late to the party, I thoroughly enjoyed the article and I completely agree that the answer is both/and not either/or.  I also like your above statement about plan development and change.  I offer the following on it.

I am a musician (retired band conductor actually) and as such, have studied (or was forced to study) a great deal of music from different composers and different eras.  I found much of it quite confusing as describing one composer’s music compared to another’s can be pretty esoteric.  However, in graduate school, I had a music theory professor who described composers in terms of what their process was.  His thesis was essentially this:  There are three parts in the process of composing and developing music.  Composers are either innovators, codifiers, or exploiters.  A good example of an innovator would be Beethoven or Wagner - what they did changed the world of music, even if they didn’t have it all figured out yet.  It’s why they are still famous.  Innovators always get lots of credit.  Exploiters are people that take other peoples’ mistakes and learn from them, then figure out how to do something better than anyone else can, even after others have moved on.   In the music world that’s J.S. Bach, whose own sons thought he was old fashioned and stuck in the past.  However, because his music is the best possible example of its type, it remains popular hundreds of years later and he is on the “Mt. Rushmore” of composers.  Guys like him are kind of the GOAT.  It’s the codifiers who don’t get a ton of credit.  They are taking that innovation - using and applying it in new and different ways.  Sometimes they are successful, but often they aren’t or their success is middling at best.  We don’t name a lot of these guys because they get lost in the process of “progress”.  Some of these guys were excellent composers in their own right but didn’t take the ideas to their peak.   I think that all of this applies to a lot of different things, including the use of analytics in baseball.  

We are currently deep in the codifying world.  There were some great innovators that had some success and now everyone is trying to figure out how they can do it better than anyone else.  The problem is that within the certainty that statistics provide there is indeed a lot of uncertainty.  The Twins are no different in this than anyone else, even though there is WAY too much kvetching about “Rocco’s Spreadsheet” (I did like the line - “how do we know what is on it?”  That was funny and so so true.). At this point (maybe thankfully) we are still quite a distance from figuring it all out so I don’t think that we really have any exploiters yet, even though some would like to claim that they are.  

So. . . What does all of this mean?  Not much except that we keep on trying, mixing in a both/and kind of way in hopes of perfecting at least one part of it that we can exploit against the rest of the league. . . . until the next week when the rest of the league figures out what we’re up to and adapts to it.  Argh!

This might be the answer, but what do I know?  I’m just the (retired) band guy!