Reliance on analytics is not the problem

[Fire Dan Gladden]

Poor, poor Bill James.  Based on many of the comments here, it sounds like if the majority of TD readers had their way, they would go back in time and make Bill James "disappear"...

I have been wondering for a long time why everybody seems to blame all of the bad things that happened this year on "Rocco's overreliance on analytics".  Does he use analytics?  Sure... so does everybody else in baseball (not to mention football, basketball, hockey, curling, and any other activity that analytics can give you an advantage).  But why the attack on the analytics usage?  More directly, why the attack on specific parts of his analytic usage?

Defensive shifts
General defensive placement
Pinch hitting
Lineups
Defensive replacement

All of these rely (to some degree) on advanced analytics.  Rarely are any of these discussed here.  The vitriol is always around pitching usage, but really seems to stem from Sonny Gray's complaining about not being allowed to go through the third time of the order and some relievers pitching poorly in the middle of the season.  

The problem with analytics is the same problem with everything else: it is not 100% accurate.  But they do increase your chances.  I remember the comments when Gardenhire was the manager:  He never uses analytics!!  The Twins are light years behind on metric usage!!

How quickly the pendulum swings...

Call me a Rocco apologist.  That's fine.  I have said many times before I have no clue if Rocco is a good manager.  Too many variables the general public knows nothing about.  What I would actually like to hear are logical arguments that Rocco's reliance on analytics actually cost the Twins games, not the emotional responses that come with the Twins collapsing.

[jmlease1]

100%. I think some of the complaints about Rocco & analytics are more of a complaint about the current state of baseball, where certain fans hate starters being pulled at 90-100 pitches, rather than being allowed to go as deep in the game as they can, and only getting pulled when they run into trouble. Or want more base-stealing. Or hate strikeouts. So I'm not sure it's really about Rocco all the time, he's just the focus as our manager for what they feel is a bigger problem (but maybe don't/can't articulate it like that)?

Of course, I'm also the guy who thinks fans tend to overrate the impact of the manager on a team; I think there are only a handful of superior managers (and even they can't win without the talent) and increasingly few terrible ones. Most of them fall in the mushy middle and in aggregate only go plus/minus 2 wins based on strategic decisions throughout the year.

I don't mind analytics in baseball at all. It's always been the most stat-driven sport and the least interactive by the players compared to basketball, hockey, football, etc when players have to be in synch at all times or things can go off the rails quickly. Digging in on the stats has always been part of what's fun for me in baseball.

[Riverbrian]

I am always amazed by how easily blame can be confidently matter-of-factly assigned to a process or person that you have no access to.  

[wsnydes]

The game has changed.  And has arguably changed faster and more dramatically then it's ever changed before in its history.  Not all fans are going to keep up with that.  Not all of them will even want to, and that's fine.  That's the risk of change.  Analytics aren't going anywhere, so those that choose to stay do need to adapt to what the game has become.  

Analytics are everywhere.  Not just baseball.  Not just sports.  Everywhere.  Not using them is simply putting yourself at a disadvantage.

[Trov]

On 11/12/2022 at 7:05 PM, Fire Dan Gladden said:

Poor, poor Bill James.  Based on many of the comments here, it sounds like if the majority of TD readers had their way, they would go back in time and make Bill James "disappear"...

I have been wondering for a long time why everybody seems to blame all of the bad things that happened this year on "Rocco's overreliance on analytics".  Does he use analytics?  Sure... so does everybody else in baseball (not to mention football, basketball, hockey, curling, and any other activity that analytics can give you an advantage).  But why the attack on the analytics usage?  More directly, why the attack on specific parts of his analytic usage?

Defensive shifts
General defensive placement
Pinch hitting
Lineups
Defensive replacement

All of these rely (to some degree) on advanced analytics.  Rarely are any of these discussed here.  The vitriol is always around pitching usage, but really seems to stem from Sonny Gray's complaining about not being allowed to go through the third time of the order and some relievers pitching poorly in the middle of the season.  

The problem with analytics is the same problem with everything else: it is not 100% accurate.  But they do increase your chances.  I remember the comments when Gardenhire was the manager:  He never uses analytics!!  The Twins are light years behind on metric usage!!

How quickly the pendulum swings...

Call me a Rocco apologist.  That's fine.  I have said many times before I have no clue if Rocco is a good manager.  Too many variables the general public knows nothing about.  What I would actually like to hear are logical arguments that Rocco's reliance on analytics actually cost the Twins games, not the emotional responses that come with the Twins collapsing.

I am not anti analytics, but just as I was upset that Gardy never used it, I feel Rocco, and company, fall to it as all decisions seem to stem from it.  Here is the thing, every time a coach makes a move, it will be questioned if it does not work out.  If it does, it will normally be praised, and if it is not under the normal type of moves it will get praised even more.  

The one thing I always try to remind people of analytics looks at hundreds of thousands to millions of situations over history, and never look at the situation at the time.  For example, the historical of scoring 1 run with a run with a runner on first with 0 out is 18.7% with runner on 2nd and 1 out is 25.1%.  So bunting seems to be supported here to try and get more runs.  However, the analytic people then point to the chance of scoring more than 1 is lowered.  So by bunting a guy to 2nd may increase the chance of 1 run, but actually decreases the chance of more than 1 run.  

So the analytics people say bunting is a negative thing and rarely do it.  Because you would be more willing to do it when 1 run will win the game.  However, if you never practice it, now you are doing things you do not do.  

Also, there are times where making sure you get a run will make a difference.  Looking at individual game situations are still important.  In terms of shifts I think people give them too much credit.  You still have 7 defenders, catcher and pitcher need to start where they do, but the other 7 get to move, or did, and need to defend the same overall area.  Teams took data to see where a guy was more likely to hit a ball and try to defend that area.  When it "stole" a hit we all were like ugh the shift, but when the hitter hit a hole where a non shift would have made the out we would point out how the shift did not work out that time. 

It was all relative, and moving defenders around the field was never new, it just got highlighted when 2nd basemen were playing in short right and SS on other side of base.  As the saying always went, just it it where they are not defending.  

[Major League Ready]

2 hours ago, Riverbrian said:

I am always amazed by how easily blame can be confidently matter-of-factly assigned to a process or person that you have no access to.  

This is something that quickly illustrates the relative level of experience assessing business issues / strategies.  People who regularly assess business issues do not offer an opinion until they have a significant portion of the information.  Even, then, you will not get the kind of matter of fact conclusions we see from sports fans with limited information on a constant basis.  

[Hosken Bombo Disco]

On 11/12/2022 at 7:05 PM, Fire Dan Gladden said:

The vitriol is always around pitching usage, but really seems to stem from Sonny Gray's complaining about not being allowed to go through the third time of the order and

I compared Sonny Gray’s game logs from 2019 and 2022, and from what I am seeing, the weight of evidence falls in favor of leaving Sonny Gray in to pitch longer. Obviously I was not watching those Reds games in 2019 however! But I would be glad to engage in a deeper batter-by-batter examination of those games compared to the 2022 games and have my mind changed. 

[Fire Dan Gladden]

22 minutes ago, Hosken Bombo Disco said:

I compared Sonny Gray’s game logs from 2019 and 2022, and from what I am seeing, the weight of evidence falls in favor of leaving Sonny Gray in to pitch longer. Obviously I was not watching those Reds games in 2019 however! But I would be glad to engage in a deeper batter-by-batter examination of those games compared to the 2022 games and have my mind changed. 

Could you please expand on your stance? 

Everything I have seen or heard regarding Sonny Gray pitching longer into games is 100% against it,  As an example, here is an article from August of this year:
https://zonecoverage.com/2022/mn-twins-news/sonny-gray-must-pitch-more-efficiently-to-avoid-the-early-hook/

I have seen other articles providing similar types of data.  This isn't just a Twins thing either, the Reds were also doing it.  I can't seem to find anybody (other than a frustrated Sonny Gray) that can show the data supporting him pitching longer and would love to hear a counter-argument.

[chpettit19]

1 hour ago, Hosken Bombo Disco said:

I compared Sonny Gray’s game logs from 2019 and 2022, and from what I am seeing, the weight of evidence falls in favor of leaving Sonny Gray in to pitch longer. Obviously I was not watching those Reds games in 2019 however! But I would be glad to engage in a deeper batter-by-batter examination of those games compared to the 2022 games and have my mind changed. 

I think the challenge at using Gray as our test subject here is that people want to point at his usage as being based on not wanting him to see the order a 3rd time. I think there was some of that, but I think the bigger variable was health. I don't think the Twins went into 2022 saying "let's trade our most recent #1 pick for a guy we think can only go through an order twice." I think his inability (for years now) to stay healthy caused his usage to change from what him and the Twins wanted it to be.

Spring training started late for everyone. It started even later for Sonny as he was traded right as it was about to get going and he had life things to deal with with that change and he didn't get the ST innings in that he needed. So his first start was a little short (4.2 innings, 2 earned). His next one he got hurt after 1.2 innings and went on the IL. So now he's had an extra short ST and an IL stint after 1+ starts (April 16th).

So when Sonny came back the Twins were understandably cautious with him. 4 innings in his first start back (May 7th) followed by 4.1 innings (May 13th). In his 3rd start back from the IL he went 6 innings. Followed that with a 7 inning start and a 6 inning start. Then went back on the IL (May 29th). So the Twins haven't made it out of the first 2 months of the season and he's hit the IL twice. If someone wants to make the case that he isn't being used long enough in games you point out that through 7 starts he only has 3 that he went into the 6th, but that's ignoring a whole lot of context.

After coming back from his 2nd IL stint they were again a little cautious. 5 innings (June 15th) followed by 4 innings (he was bad and gave up 4 runs that game). Then in his 3rd start back from his 2nd IL stint he went 7 innings.  His next 3 starts were shorter, but he also wasn't good in them. He wasn't pulled cuz of "analytics" (in quotes cuz it really is analytics that tell you those number of runs in those number of innings is bad, but those are "acceptable" analytics to the anti-analytics crowd) in games where he went 4.2 and gave up 5 earned, or 3.2 innings and 6 earned. The next start after those 3 bad ones he went 6 innings. He was then in basically a 5/6 inning per start range for the rest of the year until he got injured again in September. 

The problem was that when he had his now infamous statement of frustration about not going deeper into the game it was during his hot stretch and people went back and looked and his innings/start numbers were very low. They ignored the context at the start of the year that caused that and decided the Twins were just using every starter as 2 times through the order guys when it wasn't the case. The Twins had some starters (Bundy and Archer as the obvious examples) that weren't talented enough and the plan from the jump was to use them in shorter stints and piggyback them with Winder. Then Winder got hurt and messed that up. Ryan and Gray were given more leash until Sonny hit the IL twice in his first 7 starts and Ryan got crushed by COVID and wasn't the same guy for the next month after he returned. This fed the "Twins use analytics too much and Rocco has no feel for the game!" crowd by adding some context ignoring data points to their narrative. The Twins had a more individualized plan to start the year, but it gone blown up by injuries. My problem with them is they never adjusted and just went through the rest of the year overusing 1 inning guys and it cost them a lot of games.

[stringer bell]

29-year-old Sonny Gray also might not be the same as 32-year-old Sonny Gray. Gray's primary pitch in 2022--2-seam fastball--wasn't overpowering, but fooled hitters with it's arm side break. I don't know if Gray was more dominating in '19 or if his pitch mix was different. 

It did seem that Gray was coming off or building up from injury for much of the '22 season. There are so many factors. Maybe I'm biased by the grunting, but it seemed that Gray was putting maximum effort into every pitch and I'm wondering if he emptied the tank with 85-95 pitches.

 

 

[stringer bell]

24 minutes ago, chpettit19 said:

I think the challenge at using Gray as our test subject here is that people want to point at his usage as being based on not wanting him to see the order a 3rd time. I think there was some of that, but I think the bigger variable was health. I don't think the Twins went into 2022 saying "let's trade our most recent #1 pick for a guy we think can only go through an order twice." I think his inability (for years now) to stay healthy caused his usage to change from what him and the Twins wanted it to be.

Spring training started late for everyone. It started even later for Sonny as he was traded right as it was about to get going and he had life things to deal with with that change and he didn't get the ST innings in that he needed. So his first start was a little short (4.2 innings, 2 earned). His next one he got hurt after 1.2 innings and went on the IL. So now he's had an extra short ST and an IL stint after 1+ starts (April 16th).

So when Sonny came back the Twins were understandably cautious with him. 4 innings in his first start back (May 7th) followed by 4.1 innings (May 13th). In his 3rd start back from the IL he went 6 innings. Followed that with a 7 inning start and a 6 inning start. Then went back on the IL (May 29th). So the Twins haven't made it out of the first 2 months of the season and he's hit the IL twice. If someone wants to make the case that he isn't being used long enough in games you point out that through 7 starts he only has 3 that he went into the 6th, but that's ignoring a whole lot of context.

After coming back from his 2nd IL stint they were again a little cautious. 5 innings (June 15th) followed by 4 innings (he was bad and gave up 4 runs that game). Then in his 3rd start back from his 2nd IL stint he went 7 innings.  His next 3 starts were shorter, but he also wasn't good in them. He wasn't pulled cuz of "analytics" (in quotes cuz it really is analytics that tell you those number of runs in those number of innings is bad, but those are "acceptable" analytics to the anti-analytics crowd) in games where he went 4.2 and gave up 5 earned, or 3.2 innings and 6 earned. The next start after those 3 bad ones he went 6 innings. He was then in basically a 5/6 inning per start range for the rest of the year until he got injured again in September. 

The problem was that when he had his now infamous statement of frustration about not going deeper into the game it was during his hot stretch and people went back and looked and his innings/start numbers were very low. They ignored the context at the start of the year that caused that and decided the Twins were just using every starter as 2 times through the order guys when it wasn't the case. The Twins had some starters (Bundy and Archer as the obvious examples) that weren't talented enough and the plan from the jump was to use them in shorter stints and piggyback them with Winder. Then Winder got hurt and messed that up. Ryan and Gray were given more leash until Sonny hit the IL twice in his first 7 starts and Ryan got crushed by COVID and wasn't the same guy for the next month after he returned. This fed the "Twins use analytics too much and Rocco has no feel for the game!" crowd by adding some context ignoring data points to their narrative. The Twins had a more individualized plan to start the year, but it gone blown up by injuries. My problem with them is they never adjusted and just went through the rest of the year overusing 1 inning guys and it cost them a lot of games.

I would differentiate a bit between Archer and Bundy, but it might be splitting hairs. Archer wasn't going to go more than 5 innings due to his injury history. Even getting him through 5 seemed to be an adventure and more than once he rather suddenly imploded. There seemed to be ample reason to cap Archer at 5 innings and/or twice through the order. Bundy, OTOH, did have one of the longest starts for a Twins pitcher (8 full innings). Even when he was rolling, his margin for error was tiny. An inch of break or a mph or two of velocity took him from acceptable to batting practice. It is understandable that he too couldn't be given much rope.

Having Ryan and Gray out for large chunks of the season and then following them with Bundy and Archer was a recipe for failure. I don't think the Twins had anyone better and injuries to Winder, Paddack and Ober took away any chance to supplant the two veterans. Better health will likely make for a deeper staff with far more deep (6+ inning) starts. Going into 2023, I think the Twins can be comfortable letting Ryan go on a 180 IP per season clip, probably the same for Gray if he can avoid non-arm muscle pulls. I'd imagine they'd be cautious with Ober, although his injury problems weren't to his arm and he looked very good at the end of the season. 

 

[chpettit19]

7 minutes ago, stringer bell said:

I would differentiate a bit between Archer and Bundy, but it might be splitting hairs. Archer wasn't going to go more than 5 innings due to his injury history. Even getting him through 5 seemed to be an adventure and more than once he rather suddenly imploded. There seemed to be ample reason to cap Archer at 5 innings and/or twice through the order. Bundy, OTOH, did have one of the longest starts for a Twins pitcher (8 full innings). Even when he was rolling, his margin for error was tiny. An inch of break or a mph or two of velocity took him from acceptable to batting practice. It is understandable that he too couldn't be given much rope.

Having Ryan and Gray out for large chunks of the season and then following them with Bundy and Archer was a recipe for failure. I don't think the Twins had anyone better and injuries to Winder, Paddack and Ober took away any chance to supplant the two veterans. Better health will likely make for a deeper staff with far more deep (6+ inning) starts. Going into 2023, I think the Twins can be comfortable letting Ryan go on a 180 IP per season clip, probably the same for Gray if he can avoid non-arm muscle pulls. I'd imagine they'd be cautious with Ober, although his injury problems weren't to his arm and he looked very good at the end of the season. 

 

Yeah, I could've stated the Archer and Bundy situations better, but my post was already getting lengthy. Really I was just trying to say they went into the year not expecting those 2 to go deep into games and they appeared to have a plan to use Winder as a piggyback to help with that. But then the whole plan got blown up with injuries and they just never seemed to figure out a new plan so destroyed their already short pen with the short start followed by a whole line of 1 inning relievers nonsense.

Agree they can let Ryan and Gray go get as many innings as their performance allows moving forward. I have no expectation that Sonny will even get to 160 IP, but it shouldn't be because they don't want him seeing an order a 3rd time and I don't think that was ever the real concern with him. He's just not durable. Ober is a huge wildcard to me. No idea what to expect out of him or Maeda next year.

[Doctor Gast]

19 hours ago, chpettit19 said:

I think the challenge at using Gray as our test subject here is that people want to point at his usage as being based on not wanting him to see the order a 3rd time. I think there was some of that, but I think the bigger variable was health. I don't think the Twins went into 2022 saying "let's trade our most recent #1 pick for a guy we think can only go through an order twice." I think his inability (for years now) to stay healthy caused his usage to change from what him and the Twins wanted it to be.

Spring training started late for everyone. It started even later for Sonny as he was traded right as it was about to get going and he had life things to deal with with that change and he didn't get the ST innings in that he needed. So his first start was a little short (4.2 innings, 2 earned). His next one he got hurt after 1.2 innings and went on the IL. So now he's had an extra short ST and an IL stint after 1+ starts (April 16th).

So when Sonny came back the Twins were understandably cautious with him. 4 innings in his first start back (May 7th) followed by 4.1 innings (May 13th). In his 3rd start back from the IL he went 6 innings. Followed that with a 7 inning start and a 6 inning start. Then went back on the IL (May 29th). So the Twins haven't made it out of the first 2 months of the season and he's hit the IL twice. If someone wants to make the case that he isn't being used long enough in games you point out that through 7 starts he only has 3 that he went into the 6th, but that's ignoring a whole lot of context.

After coming back from his 2nd IL stint they were again a little cautious. 5 innings (June 15th) followed by 4 innings (he was bad and gave up 4 runs that game). Then in his 3rd start back from his 2nd IL stint he went 7 innings.  His next 3 starts were shorter, but he also wasn't good in them. He wasn't pulled cuz of "analytics" (in quotes cuz it really is analytics that tell you those number of runs in those number of innings is bad, but those are "acceptable" analytics to the anti-analytics crowd) in games where he went 4.2 and gave up 5 earned, or 3.2 innings and 6 earned. The next start after those 3 bad ones he went 6 innings. He was then in basically a 5/6 inning per start range for the rest of the year until he got injured again in September. 

The problem was that when he had his now infamous statement of frustration about not going deeper into the game it was during his hot stretch and people went back and looked and his innings/start numbers were very low. They ignored the context at the start of the year that caused that and decided the Twins were just using every starter as 2 times through the order guys when it wasn't the case. The Twins had some starters (Bundy and Archer as the obvious examples) that weren't talented enough and the plan from the jump was to use them in shorter stints and piggyback them with Winder. Then Winder got hurt and messed that up. Ryan and Gray were given more leash until Sonny hit the IL twice in his first 7 starts and Ryan got crushed by COVID and wasn't the same guy for the next month after he returned. This fed the "Twins use analytics too much and Rocco has no feel for the game!" crowd by adding some context ignoring data points to their narrative. The Twins had a more individualized plan to start the year, but it gone blown up by injuries. My problem with them is they never adjusted and just went through the rest of the year overusing 1 inning guys and it cost them a lot of games.

Thank you Chpettit for that very detailed synopisis of the starting pitching situation. The problem with starting pitching wasn't that we weren't pitching them enough it was because we were pitching them too much. Winder was used rarely used more than 2 innings. Because the rotation quickly became decimated, they removed Winder was removed from RPing & thrown into the rotation. Where he quickly was overstretched & landed on the IL.

I'd like to use Bundy as an example. Not a fan of Bundy but looking at Bundy's past poor performances & sole good year in the shorten '20, I profiled him as a good pitcher if you limit his innings. In the beginning of the season he was doing pretty solidly well & he began to cry out for more innings & that he felt strong. Then we started to extend him he started to weaken & have ineffective innings to the the point where he couldn't bounce back & had to leave the game early. He still pitched 140 innings but how many were quality innings? If he was maintained at 5-6 innings per start (depending on pitch count). He would maintain a rythm & be able to bounce back to give you another quality start each time. At the end of the year (in theory) he could have given you 30 full games of 160+ innings and I'd bet most them would be quality innings. 

I'm absolutely in favor of analytics but my main problem with Baldelli is his formulas & mindsets that he's never able to adjust them depending on the situation. My preference would have been if Sheldon was the manager & Baldelli learn how to manage from him as a bench coach.