Tangotiger Blog

Wednesday, October 28, 2020

Maybe taking out Snell was a bit too early?

I’ve posted on Twitter some data. I wrote a chapter in The Book on the topic. Plenty of others have chimed in, all concluding the same thing. Times Thru Order effect is real, and it does not take a backseat to a pitcher that is mowing down his opponents.

I’m here to offer a possibility that under a specific situation, maybe there is an overriding concern. We’ll get there in a sec. Here’s a data chart, and I’ll explain what it is. (Click to embiggen)

All the data is from 2010-2019, courtesy of Retrosheet. Chart on the left is regular season and chart on the right is the post-season. The columns from -1 to 9 refers to the strikeout - walk differential for the first 18 batters faced. -1 means more walks than strikeouts. 9 means 9 or more strikeouts than walks. All the others are exact counts. Walk is really walks + hit batters, with IBB removed. Snell was a “9”, meaning the best.

The columns “1” and “2” means 1st time and 2nd time through the order. The data is wOBA. Now, don’t pay too much (or any) attention to those two columns. Those are selection biases and we’ll learn nothing from them.

The column we care about is “3” meaning the performance 3rd time through the order. Since we selected our strikeout-walk differential based on the first 18 batters faced (in-sample), we are interested in what happens after that (out-of-sample). And that is the third time thru the order.

Now, let’s start with the regular season. We notice that the wOBA gets progressively lower, the more the strikeout-walk differential. That’s really an issue of bias, because the better pitchers will be part of the higher differential groups disproportionately. So, we expect that value to behave as it does based strictly on the quality of the pitchers in each of those groups.

Now, the right thing for me to do is to look at the actual quality of pitchers in there. I will leave that to the Aspiring Saberist. What we can do instead is use this as a baseline of sorts as we now turn our attention to the post-season chart.

First, we’ll notice that in the post-season, third time through, with the strikeout-walk differential at 5 or less, the performance is roughly the same. Since the post-season is made up of good pitchers to begin with, we don’t expect the progressive drop we saw from the regular season.

That said, then the fun happens. We see a big drop at a strikeout-walk differential of 6. Then a huge drop at 7. And all by its lonesome is a .214 wOBA with a strikeout-walk differential of 9+. Now, we are only talking about 9 pitchers (all huge names, as you’d expect: Cole and Kershaw twice each, Kluber, CC, Max, Stras, Verlander). And they totalled just 49 plate appearances. One standard deviation is 71 wOBA points. If we assume a .300 level talent, that’s only 1.2 standard deviations.

But, 49 sounds like a lot, and if the non-data folks want to hang their hats on something, it’s that one. That in the post-season, when a superstar pitcher is mowing down batters, they went out to pitch at a better-than-Mariano level.

And to give them more ammunition, we can combine the strikeout-walk differential at 7+ (so those last three lines). That’s 341 batters faced at a wOBA of .259 (meaning Mariano level). And one standard deviation is now down to 27 points. Which is actually 1.5 standard deviations.

Therefore, I will give them that as a reasonable possibility. Just like we DID find evidence that a pitcher’s talent level does improve in the 9th inning when he’s going for a perfect game, maybe there is a brief change in talent level for a pitcher, who is at the highest level, pitching at an even higher level, and on the greatest stage, the post-season.

That said, these star pitchers did not pitch at the level that their first 18 batters suggested, that they were “unhittable”. The best you can see is that they pitched like an average Mariano Rivera. Which is of course great. And naturally, you don’t pull Mariano Rivera from a game.

But everything I’ve said comes with a degree of uncertainty. You can make the argument that Blake Snell third time through drop in performance gets cancelled out by what we saw, and so we had an average Blake Snell. And you don’t pull an average Blake Snell.

Or, all of this is really grasping at straws, that the regular season provides so much data that we can conclude that Snell was losing effectiveness all the while our eyes were lying to us.

Anyway, the floor is open to the Aspiring Saberists.

Tuesday, October 27, 2020

How to handle the 2020 season for forecasting the 2021 season

Background

​I’ve avoided thinking about this topic when it came to 1981 and 1994. It was long ago, it’s an inconvenience, and sometimes it’s just easier to assume that everyone was injured for one-third of a 162 game season. Not to mention you can create a model with 1996-2019 data, pretty easily.

But now we have 2020 to contend with. I was asked to create the WARcels for 2021 and I knew that that thing I’ve avoided for twenty years needed to be met head-on. And true to form, the method I will be describing is going to be as simple as I can make it. If someone wants to take the idea here and expand it and make it better, go for it. I like to keep things simple. That’s the whole idea behind the original Marcels, and now the WARcels.

Remember, unlike other 60-game performances by players, where those players likely played 60 games because they were injured or unproductive, in this case, it’s because the entire league was limited to 60 games. And so, what we really want to do is “fill in” the rest of the 102 games, but do so WITHOUT pro-rating. That’s because pro-rating will pro-rate both the talent and the luck (good or bad). And we definitely don’t want to do that. We also do NOT want to use performances from 2019 and 2018 and 2017. We just want to deal with 2020 data and see how can we possibly fill-in 102 team games for each player.

Understanding the data

Ok, now we have our objective, let’s create our model. The first thing I did was rely on the generosity of Baseball Reference and their daily WAR feeds. I grabbed the feeds of June 5 2019, June 6 2018, June 10 2017 to roughly correspond to ~60 games per team. I will call the plate appearances (PA) and WAR in these files as PA1 and WAR1. I then subtracted those from the end-of-season WAR of 2017, 2018, 2019 to get the rest-of-season PA and WAR, which I will call PA2 and WAR2. So, the “1” designation is for the first 60 games, and the “2” designation is for the next 102 games. More or less.

I settled on every batter with over 100 plate appearances, which as the luck of luck would have it gave me exactly 900 players (average of 300 per season, or 10 per team).

The highest WAR1 belonged to Bellinger in 2019 at 5.56, followed by Mike Trout in 2018 at 5.4. Then Jose Ramirez in 2018 with 4.27, Trout again 4.12 in 2019. The next highest WAR1 was Mookie Betts, 4.11 in 2018. Then Lindor also in 2018 at 3.62. And on and on. What I then did was take the top 30 in WAR1 and put them in the first group. Their average WAR1 was 3.35. Then I took the next 30 highest WAR1 and put them in the second group, with an average WAR1 of 2.48. And on and on. So I ended up with exactly 30 groups of 30 players in each group.

I then asked this question: how much WAR did those players generate in the rest-of-season, what was their WAR2? The top group had a 3.31 WAR2. In other words, the 30 top players, in ~60 team games had 3.35 WAR and basically matched that in ~102 team games with 3.31 WAR. This repeated itself with the next group, those with the 2.48 WAR1 had a 2.51 WAR2.

Indeed, roughly speaking, the top 450 players over these three seasons (150 players per season) had a WAR2 that basically matched their WAR1. In other words, if you want a rule of thumb: whatever it is that an average or better player did in 60 games, expect him to match in the next 102 games. So, if you want to end reading right now, just double every average or better player’s WAR of 2020 and that’s their full-season 2020 WAR.

For those still interested in more keep reading.

Into the weeds

We’re going into the weeds now. The core of WAR is made up of two components: PA and wOBA. Indeed, that’s the whole idea behind Quick WAR. So, what we’ll do is get a wOBA-equivalent value by reverse-engineering WAR and PA. To do that, we take WAR1 divide it by PA1, which gives us wins per PA. We then multiply by 10, to give us runs per PA. And multiply by 1.2 to give us wOBA units per PA. Those familiar with Fangraphs Guts know what I’m talking about here. Finally, we add in .280, which is roughly the replacement level wOBA value. And there you go, we have a wOBA-equivalent value of WAR. We do this for all the PA1 and WAR1 values to give us wOBA1.

Now, to convert wOBA1 into a future_wOBA, we need Regression Toward the Mean. And that is going to be the easy part. What we need to do is apply a Ballast, adding in 300 PA of league-average performance. So, we take wOBA1, convert it into a wOBA over league average, multiply that by PA1/(PA1+300) and that’ll give us the regressed-wOBA over league average. Add the league average back in and we have regressed-wOBA.

Finally, we need a future_PA. This one is even easier. While 102/60 is 1.7, we won’t presume everyone will get 70% more PA in rest of season. Rather, we can look at our 2017-19 data and see what happened. And what happened is that they got roughly 45% more PA in rest of season. So, future_PA is 1.45 times PA1.

With future_PA and future_wOBA in hand, we use quick WAR to convert to future_WAR. And so, we can compare the actual WAR2 to the xWAR2 (future_WAR) and see how it holds up.

This is what that looks like for 2017-2019. (Click to embiggen)

That’s a reasonably good fit (for binned data anyway). If I showed you the individual data points, it would look like there’s very little pattern. When you report binned data, don’t bother reporting the correlation, as that’s highly deceptive.

And this chart shows part of the data of above, charting the xWAR2 on the x-axis against the same WAR2 on the y-axis.

So in 2020, Freddie Freeman and Jose Ramirez got 3.4 WAR in the first 60 games. That translates to 6.4 WAR in a full-season. Betts goes from 3.0 WAR to 5.7 WAR. Tatis comes in at 5.6 WAR.

Next up will be pitchers. I haven’t looked at the data yet, but I hope the translation process is similar.

Monday, October 26, 2020

Statcast Lab: Batter-Runner v Outfielder, Play at 2B, part 2

I recently introduced the Outfield v Runner confrontation. The core concept is to compare the distance of the fielder to the base (second in this case), to the distance of the runner to the same base. Because each runner has his own running speed, we convert his distance from feet to seconds. If for example Byron Buxton is 90 feet away and he runs at 30 feet per second, then we know he’s about 3 seconds from reaching his target base.

We have a chart of all batted balls with bases empty, where the batter got a basehit and they have to decide whether they’ll freeze themselves by holding at first base, or whether they will go for two. And if they go for two, they’ll either be sniped down by the outfielder, or they’ll have successfully swiped the extra base. (Click the image to embiggen all of the charts below.)

When we look at it league-wide, patterns emerge. The yellow circles are all those plays where the batter held up at first base. Virtually every circle above the red line were singles. The red line is the no-go line: plays that are so obvious to be singles that even the weakest arm in the outfield would nab the runner, so well over 99% of the runners don’t try.

The green circles are all those plays where the batter ended up with a double. Virtually every circle below the green line were doubles. The green line is the go line: plays that are so obvious to be doubles that even the strongest arm in the outfield would not nail the runner, so well over 99% of the runners take that extra base.

Those green dots are officially singles, but because of a fielding error, the batter ended up at second base.

The red dots are all the outs on base. In between the go/nogo lines is where you have the decision making zone: there is a smattering of singles and doubles, meaning that the runners are in a bind trying to figure out whether to go for it or not. And that because of all those red dots, the outs. The plays are close enough that basically anything can happen. This is where the action is, this is where the runners make the call, and this is where we evaluate the arm of the outfielder (both in strength and accuracy).

So let’s do that. Let’s evaluate the outfielders. We’ll start with everyone’s favorite outfielder, Jackie Bradley Jr. JBJ not only led Catch Probability for 2020 for his range, but he is highly regarded for his arm. What we can do is take the above chart and filter it down to an individual outfielder, namely JBJ in this case. This is how it looks.

Every runner above the no-go line didn’t go, and every runner below the go line went. All of that is noise, and tells us nothing about the respect runners have for his arm. Where we learn that is in the decision-making zone, that region between the go/nogo lines. JBJ had 13 runners that had to make a choice, and only two of those runners went for it (and one of those was right near the go-line). Two out of thirteen is 15%, which is well below the league average of 35%. So we can conclude that runners have a good amount of respect for JBJ. JBJ is able to freeze the runners.

The most respected arm in the outfield (at least insofar as we are only looking at plays at second base) is Whit Merrifield. This is his chart. He had 11 runners in the decision-making zone and none of them went for it.

You will have noticed something so far: the outfielders with the most respected arms are not throwing any runners out. That’s the double-edged sword: the outfielders are so respected that the runners are holding up. While they won’t get official credit for a non-assist, Statcast can now recognize frozen runners. We’ll get them that respect metric soon enough.

Here is Bryce Harper, who is also respected as you can see by the decision-making zone (two of eleven runners going for it, and both of those were right close to the go-line). But, what is that red dot in the no-go region? It seems ridiculous that a runner would go for it. Well, this is that play. And as you can see, it’s not that the runner went for it, but that he was so complacent rounding the bag that Harper managed to double him off first base. That’s another thing we can do here, is find all those extreme cases.

On the flip side is the least respected arm, Corey Dickerson. Corey Dickerson is actually number 2 in the league for best in Jump (outfielder with the best combination of reaction, route, burst). But, as you can see, when it comes to his arm, runners run on him each time they have a choice.

They even run on him when there’s no reason at all to run on him. This one was the ultimate in no-respect, and the runner paid the price for it.

Finally we have Christin Stewart, who runners also don’t respect, and maybe they should start to. Five times they had a choice to make, and all five times they went for it. But twice he sniped them down.

Now that we have all this data, we can start to create leaderboards, and ultimately come up with a rating for outfield arm. This particular snippet is only for runner plays at second base. Eventually, we’ll include doubles/triples, first-to-third, as well as plays at home. We’ll get there soon…

Friday, October 23, 2020

When you shift the infield, how should you shift the outfield? Part 0

​This is going to be a high-level research. Almost no controls, just to get the landscape. Hence the part 0. If this wasn’t Friday night, I’d spend a bit more time on this, so this just lays the groundwork for future research. I will get back to it.

As we learned, where you put the CF is highly impactful of batter performance. Now, how about the combination of the infield alignment AND the outfield alignment? Looking at all LHH v LHP, and they wOBA .313 whether with the infield shift or not. Now, remember I said no controls. It’s not the same population. If I controlled for the population, then we’d see a 20 point difference. But, I’m curious to see what happens if we break that down between the CF being placed on the left side or the right side.

• .303 wOBA when on left side, with or without the shift
• .324 wOBA on the right side without shift, .322 with shift

In other words, moving CF to the right side, shift or no, will increase wOBA by about 20 points. Therefore, we want the CF on the left side, and its impact will be equally felt shift or no.

We see a somewhat similar story with LHH against RHP:

• .327 wOBA when on left side, shift or no
• .361 wOBA when on right side without shift and .352 with shift

In other words, you really want to keep the CF on the left side. And moving the centerfielder to the right side will have some 30 points higher impact on wOBA.

Now, what about RHH? Against LHP we see this:

• .361 wOBA when on left side no shift, .375 wOBA on left side with shift
• .324 wOBA when on right side no shift, .366 wOBA on right side with shift

Now, that’s interesting. With a RHH, having the CF on the left side of the field doesn’t have that much of an impact on wOBA, shift or no. It has some, but not a great deal. But, when we put the CF on the right side of the field, the wOBA explodes by 41 points with the shift. The best combination for the defense is no shift, CF on right side. The worst combination is CF on left side with the shift (in other words, just leaving the RF and 1B all alone out there).

Finally, RHH v RHP:

• .346 wOBA when on left side no shift, .365 wOBA on left side with shift
• .311 wOBA when on right side no shift, .358 wOBA on right side with shift

Functionally the same thing as RHH v LHP. Best combo for defense: CF on right side with no shift. Worst combo: CF on left side with shift.

Again, I should point out that there’s no controls. So, the next step is to control for the batter-pitcher (or batter-pitcher-fielding team) combo to see what impact the combination of infield and outfield shifting has on the batter.

Thursday, October 22, 2020

Unit Sphere: Spin Axis

​For those not familiar with Alan’s trajectory calculator, it is easily one of the most indispensable tools in the saber toolshed. He also describes the spin axis based on the movement of a ball. That is, not the spin axis out of the pitcher’s hand, but rather the average spin axis over the flight of the ball. We’ve been treating the two as being equivalent, but Jared Hughes would suggest otherwise. In any case, that’s a discussion for a future thread.

For this thread, I just want to show what Alan’s spin axis looks like (wb, wg, ws from his calculator), when it is unitized (meaning the radius of the three dimensions of the spin axis adds up to one) and plotted at the league level. And since it is three dimensions, then we’ll plot it in three dimensions. Here’s how that looks like for a RHP (for LHP, the wb values remain the same, with the wg and ws signs flipped). To navigate, use the left, right and wheel on your mouse.

Black dots are the medians, and gray dots are reference points.

http://tangotiger.net/spin/spin3D_M.html

How close is Mookie Betts to being great enough to be in the Hall of Fame

Setting aside whatever rules are in place, I asked readers how they would want to vote for the Hall of Fame (if they had the vote) for Ken Griffey Junior at various stages of his career, as well as Ted Williams. And the consensus was that they’d vote for Junior after his 1997 season and Ted Williams after his 1947 season. This is how they stacked up:

• After 1997, Junior had 41 wins above average (in equivalent of 7.5 162-game seasons)
• After 1947, Williams had 42 wins above average (in equivalent 5.5 seasons)

So pretty clearly, they are looking for players to cross that 40 WAA level. That’s one of the things I do with my polls. I don’t ask: How many WAA are you looking for. Rather I ask an indirect question and reverse engineer how they are really thinking. So, 40 WAA is our threshold. That’s not to say you can’t make it in the HOF at 30-39 WAA, but that once you cross that 40, you are in.

Here’s Mookie Betts so far:

• After 2020, Betts had 33 wins above average (in equivalent 5.2 seasons). He is one year, maybe two, from getting to the 40 WAA level, and be considered a Hall of Famer by those who follow me.

Since someone will bring up Mike Trout:

• After 2016, Trout had 36 WAA in 5.0 seasons (aka slightly better than Mookie Betts)
• After 2017, Trout had 41 WAA in 5.7 seasons (aka slightly worse than Ted Williams)
• After 2018, Trout had 49 WAA in 6.6 seasons (aka noticeably better than Junior)

I also asked my followers about Bobby Orr. The consensus was after the 1971-72 season, the equivalent of 5 80-game seasons. So, that’s the Ted Williams level, of five years at the highest level of play. That’s what everyone is after. Bobby Orr notably at that point would have been only 24 years old! So he reached the Hall of Fame level at age 24, with five Norris (best defender), three Hart (MVP), and two Smythe (Stanley Cup MVP to go with the two Cups). He also somehow won the scoring title… as a defender.

Tuesday, October 20, 2020

Statcast Lab: Should the centerfielder play to pull or go the other way? Part 2 of 2

We made the case that teams should favor the CF to the opposite side, and by and large the do place them there. But, do they do it often enough?

***

Let’s look at everyone’s favorite LH pull hitter, Joey Gallo. 74% of the time, the CF is playing him to pull rather than going the other way. When he plays him to pull, Gallo’s wOBA is .355. That is, his overall wOBA when the CF is standing on the right side is .355, and when the CF is standing on the left side, his wOBA is .362. In other words, there’s a small 7 point advantage to playing Gallo to pull, and so, teams should predominantly place their CF there. And they do.

Since the data goes back to 2015, we can look at everyone’s other favorite LH pull hitter, Big Papi. 71% of the time, the CF plays him to pull, his wOBA when the CF is on the right side is .393, and it plummets all the way down to .327 when the CF plays him the other way. In other words: don’t play Big Papi to pull! There’s a difference between playing the infield to pull and the outfield to pull. You can make the argument to play Ortiz to pull on the infield side, but when it comes to the outfield, play him to go the other way.

How are teams doing? (Click to embiggen.) That redbox represents: Batters who hit at least 50 wOBA points worse when CF is standing on the right side, and that teams place a fielder there at least 70% of the time. As you can see, no one is there. Note however that there are very few batters who actually do better on the pull side. Chisenhall is the closest that teams are getting right: he hits a whopping .353 when the CF is on the opposite side, .292 when the CF is on the right/pull side. And teams are placing their CF on the pull side 51% of the time.

The orange box is: what are you doing? Only Ortiz is there, and we’ve discussed him.

The blue represents batters who hit 50 wOBA points better when the CF is standing on the right side, and that teams place a fielder there less than 30% of the time. Lots of dots means teams know what they are doing. Ben Revere is a good example: .486 wOBA when CF is standing on the right/pull side and .294 on the left/opposite side. Teams know he’s a spray hitter and that’s why they placed the CF on the right side only 3% of the time, and instead had him on his weak left side 97% of the time.

All in all, it’s hard to see why teams would place a CF on the right side for more than a handful of batters. And right now, it seems there’s very little relationship that they are making the right call.

I’ll do the RHH later today.

UPDATED:

Carlos Santana (RHH) has a .422 wOBA when the CF is standing on the right side (going the other way) and a .340 when CF is standing on the left/pull side. Undoubtedly, the CF should stand on the left side, and he does so 79% of the time (21% of the time he’s on the right side). There’s a handful of other batters like Santana, but that’s it. But whether it’s Bautista or EE or Hoskins or Dozier, teams are putting their CF on the left/pull side. That’s probably not the best choice.

On the flip side are batters like Lemahieu: .448 when CF is standing on the pull/left side and .362 on the right. And 97% of the time, the CF is on the proper right side. You can see plenty of dots there.

Rule of thumb

Place your centerfielder on the opposite side.

Statcast Lab: Should the centerfielder play to pull or go the other way? Part 1 of 2

​The top chart is LHH and the bottom is RHH. (To make it easier to remember, the Red lines is for Righties, and the bLue lines for Lefties.)

The x-axis represents the angle of the centerfielder relative to home plate. 0 is home. Negative means the CF is playing toward LF and Positive means toward RF.

The LHH chart shows that the more the CF plays toward LF, the lower the wOBA, and that the more the CF plays toward RF, the higher the wOBA. That is, if the CF plays to pull, the LHH will have a higher wOBA. The RHH chart shows a similar pattern: the more the CF plays to pull, by playing toward LF, the higher the wOBA.

And MLB teams know this (to some extent anyway).

This is how often the CF is placed on the field from left to right. As you can see, against Lefties (blue), they are placed toward LF (and so playing to go the other way). And similarly, against Righties (red), they are placed toward RF, meaning going the other way.

So, if more often than not the CF are being placed in the right spot, does this mean there’s a good reason for them to be placed on the pull side? Is it maybe the big power pull hitters that are driving the high wOBA on the pull side and the weak spray hitters that are keeping that opposite side wOBA low? In other words, is it a biased sample that is driving the wOBA we see?

No.

From 2015-2020, with the bases empty, Charlie Blackmon (LHH) faced Greinke and the DBacks 25 times with the CF on the left side and 17 times with the CF on the right side. Based on the above, we therefore expect his performance to be better on the pull side (right side) and drop going the other way (left side). As it turns out, that’s what we go: .350 wOBA on the right/pull side and .280 on the left/otherway side.

I repeated this for every combination of batter-pitcher-fielding team over 2015-2020. Remember, we are controlling for the batter, pitcher, fielding team. There’s no bias in representation in the two pools.

All the lefty batters had a .327 wOBA on the left side, and they had a .349 on the pull/right side. That’s a 22 point advantage to the defense if they put the CF to play the other way. And the story is the same for righties: .326 wOBA on the right side and .361 on the pull/left side, for a 35 point advantage to the defense to place the CF to play the other way.

Teams are placing the CF to go the other way, but not nearly enough.

And the placement of the CF is changing the approach of the batter/pitcher confrontation. When teams place their CF to go the other way (which is the correct strategy and one they do more often than not), the hit-into-play rate goes down by about 5% points. The spray and launch don’t change much at all. In other words, the batters know that the CF is well-aligned and so is now less likely to try to make contact.

In my next blog post, I’ll check to see if some batters should be played to pull and are teams playing those batters to pull.

Monday, October 19, 2020

Statcast Lab: How much space should you place between infielders, part 2 of N

This blog post is focused on RHH, with three infielders on the left side. Part 1 is here.

The wOBA is .354, which is 25 points higher than the .329 when you have two infielders on the left side. This is extremely high, and calls into question the entire concept of putting three infielders to the left side against a RHH. But is there some combination where we can put three infielders to the left side successfully?

First some spatial breakdowns.

Thirdbase:

• Small gap: less than 6 degrees from the 3B line
• Normal gap: 6 to 8 degrees from the 3B line
• Large gap: more than 8 degrees from the 3B line

Shortstop:

• Small gap: less than 14 degrees from the thirdbase (player)
• Normal gap: 14 to 16 degrees from the thirdbase
• Large gap: more than 16 degrees from the thirdbase

Secondbase:

• Small gap: less than 16 degrees from the shortstop
• Normal gap: 16 to 18 degrees from the shortstop
• Large gap: more than 18 degrees from the shortstop

Now, is there a combination where we get even league-average results (.329 wOBA)? Of the 27 combinations, there are 4 combinations that we have better than league average results. The sample size however is small to the point that it’s not even one standard deviation from the mean. But, we’re looking for any little win here.

Best Combos

The thirdbase is playing close to the line, the shortstop is playing close to the 3B, and the 2B is 8 degrees from the bag. With only 54 attempts, that wOBA was .272.

The next best combo (61 attempts, .289 wOBA): third base close to the line, the shortstop normally spaced from thirdbase, and the secondbase closer to the shortstop (10 degrees from the bag). In other words, if you shift the 3B over, make sure to get that 2B shifted over as well.

The third best combo (193 attempts, .308 wOBA): third base in a normal spot, the shortstop with extra spacing from thirdbase, and the secondbase closer to the shortstop. Now, I should remind you when I say “normal spot” I mean in terms of having three infielders to one side.

The last good combo (44 attempts, .290 wOBA): third base close to the line, the shorstop close to the thirdbase, and secondbase in a normal spacing from shortstop.

That’s it, nothing else good. As for the bad combos, there’s a bunch of them.

Worst Combos

The sixth worst is the “normal” spacing for each fielder. It’s obviously the most common, and the wOBA was .343.

The worst has the 3B and SS in the normal spot, and the secondbase closer to the bag. In other words: they went to the risk of putting three infielders to one side, but (they thought they) hedged their bets by keeping the secondbase close to the bag. That is actually the worst thing they can do. The wOBA was .372 on 968 attempts.

The next worst had the 3B away from the line, the SS staying close to the 3B, and 2B normally spaced from SS. That’s a wOBA of .394 (worse than above) but only 300 attempts. In terms of z-score, terrible but not as terrible as above.

The third worst: .370 wOBA with the 3B and SS in their normal spot and the 2B too close to the SS.

The fourth worst: .397 wOBA (highest of all, but only 236 attempts): 3B close to the line, SS wider spaced from 3B, 2B normally spaced from SS.

Overall

If you are going to put three infielders to one side, you have to move that 3B close to the line. Three infielders means you are playing to pull, so you need to close the gap to the line. Don’t put too much space between the 3B and SS. The 2B is a bit harder to pin down, but he has to be off the 2B bag to a good, but not great degree. Specifically, we get a .310 wOBA on 371 opportunities when you have this fielding alignment: 3B at -40 degrees, SS at -26 degrees, 2B at -8 degrees. More broadly: 3B has to be within 5 degrees of the line, the SS within 15 degrees of 3B and the 2B 15-20 degrees from the SS. That’s your best bet to getting a winning spatial-based alignment against RHH with three infielders.

But if you really have to do something: don’t put three infielders to the left side against a RHH so often. Or even maybe ever.

Statcast Lab: How much space should you place between infielders, part 1 of N

This blog post is focused on RHH, with two infielders to either side of the bag.

I looked at the fielding position of all thirdbase relative to the 3B line, and broke them up into three groups:

• Small gap: less than 7 degrees from the 3B line
• Normal gap: 7 to 10 degrees from the 3B line
• Large gap: more than 10 degrees from the 3B line

I did similarly for the shortstop:

• Small gap: less than 16 degrees from the thirdbase (player)
• Normal gap: 16 to 21 degrees from the thirdbase
• Large gap: more than 21 degrees from the thirdbase

And for the secondbase:

• Small gap: less than 23 degrees from the shortstop
• Normal gap: 23 to 28 degrees from the shortstop
• Large gap: more than 28 degrees from the shortstop

That gives us 27 possible combinations of spatially-based alignments. There actually was one combination that was never tried: 3B close to the line, the SS close to the 3B and the 2B close to the SS. But that’s a technicality, as to make that happen, it’s just about impossible with only two fielders on the left-side. So, we have really 26 possible combinations. Three combinations were tried less than 20 times across the league, so we are now down to 23 possible combinations.

The league average wOBA was .329 (I include errors as singles). The most popular combination was the Normal/Normal/Normal gap. The league wOBA was .335. It was also the third highest wOBA of the 23 possible combinations. A simple rule is: do something that’s not normal. Of course, you can’t be too abnormal.

Worst combos

The worst combination yielded a wOBA of .383 (54 points higher than league average). That one was based on the thirdbase being close to the line, the SS at a normal gap from the 3B, and the 2B at a large gap from the SS. In other words, both the 3B and SS shifted over by about three degrees, while the 2B didn’t move. That opened up the gap between SS and 2B too much. Result: highest wOBA and worst spatially-based fielding alignment.

The second worst combination: .362 wOBA (33 points higher than league average). In this case, the 3B and 2B played in their normal spot, but the SS was playing closer to the 3B. In other words: the gap between SS and 2B was too wide.

The third worst combination was everyone playing their normal spot.

The fourth worst combination, albeit at a wOBA of .337, was the 3B and SS in their normal spot, and the 2B shifted over to the right side. In other words: the gap between SS and 2B too wide. So at this point, we learned our lesson: don’t open up that spot between SS and 2B.

Best combos

Let’s now look at the best combinations. A very low .302 wOBA (27 points lower than league average). How did that happen? Open up the gap between 3B and SS. In other words, the “normal” SS spot needs to have a new normal, and they need to be moved away from 3B and toward 2B. This is consistent with what we just learned about the Worst Combos.

The second best combination: .309 wOBA (20 points better than league average). This one had the 3B farther from the line, while the SS and 2B shifted over as well. In other words, don’t play to pull so much.

The third best combination: .313 wOBA (16 points better than league average). This one had the 3B and SS in their normal spots, but the 2B was closer to the bag. In other words, close some of that gap between SS and 2B.

Other combos

There were other combos that showed positive results for the defense, but the sample size was a bit low. We can however try to combine some of those combinations to come up with additional rules.

• When the 3B plays close to the line, make sure you don’t shift the SS closer too much to compensate, and definitely don’t allow the gap between SS and 2B to get larger. If you follow that rule, then you end up with 20 points better than league average.

Next up, we’ll look at what happens when you put three infielders to the left side against RHH. Those who have followed my blog over the years know that this is a risky thing to do. But maybe there’s some combination that lets us get something positive out of it. Stay tuned…

Tuesday, October 13, 2020

Batting Average bias in MVP voting

Bill has a tremendous article showing that batting averages bias MVP voting to a pretty large extent. Now, if you wanted to determine the EXTENT of the bias, there’s a path there. While Bill used Win Shares as his central point, he did also do a quick overview with WAR, which is what I’ll focus on here.

First, figure out how many hits above (or below) the league average the hitter has (using AB as your opportunity number). For example, if you have a .360 batting average in a league of .260 with 600 at bats, that’s +.100 x 600 = +60 hits. You can now run a regression, but you can do a trial and error process as well, which is probably going to be more instructive. Give each extra-hit 0.01 WAR. So in the above example, you are giving +0.6 WAR. Go back to Bill’s study, and now look to see if the bias still persists. You will probably not notice much difference. Try again with 0.02 WAR for each extra-hit, then try 0.03, then 0.05, and then 0.10. You may iterate downwards as you may have overcompensated.

So what we are doing here is building-in the bias into the model, so that there is no bias in the output. Once you have something close to that, then congratulations, you have now figured out the extent that batting average biases the MVP voting.

Monday, October 12, 2020

Statcast Lab: Batter-Runner v Outfielder, Play at 2B

Our next foray into the Statcast world will be the outfielder v runner confrontation.

Way back in the early Statcast days, we did tackle the issue of Sacrifice Flies. In that case, it was more straightforward: every runner started from the same location (third base), at the same time (outfielder catching the ball), with an outfielder doing the same thing (setting themselves up for a throw home).

Now, let’s turn our attention to batter-runners deciding whether to go for two. In the illustration below, we start the clock when the outfielder first touches the ball. At that point, the outfielder is a bit over 200 feet from the target base while the runner is a bit over 90 feet to the same target of second base. This is a race: who gets there first.

Since this is a race, what we care about is the time. We need to convert distances of feet into time of seconds. Feet per seconds is speed. Therefore, if we know the speed of the runner and the speed of the outfielder throw, we can convert feet into seconds. We’re going to take this from the perspective of the outfielder, meaning we’ll use an average outfielder as the baseline, against the actual known runner.

The Play

Hampson is one of the fastest runners in baseball, with a Sprint Speed of around 30 feet per second. Since he is 94 feet from second base at the start of this segment of the play, then running at around 30 feet per second means he will get there in about 3.14 seconds (give or take).

The outfielder is 204 feet away. How much time will that throw take for an average outfielder? Suppose an outfielder throws at 85 mph, what does that tell us about how much time the throw will take? The first thing you want to do is convert miles per hour into feet per second. You do that by multiplying by 1.4667 (or 5280/3600; be mindful of integer division, as some programs will actually tell you that 5280/3600 is 1; I know, that’s totally insane). So, 85 mph is about 125 feet per second. Now, that’s a MUCH better number to know. When you look at 85 mph, it is the end of the road: there’s nothing you can do with that number on its own. This is why units are important: you want to be able to use the relevant number in a way you can relate to it. If you were driving from NYC to Boston, miles per hour makes sense. If you were having a race with a motorcycle, miles per hour makes sense. Throwing a ball on a baseball field to catch a runner? Feet per second.

Ok, so our outfielder throws about 125 feet per second. That’s at release. However, as soon as the ball is released, the ball goes into a deceleration phase: it will constantly slow down. How fast does it slow down, what is its decay rate? We could try to determine this based on physics. Or, we can just let empirical data drive it. So, what I did was looked at all 85 mph batted balls at each launch angle of one degree. I took the 10% longest hit ball at each angle. What this would approximate is finding those batted balls that had the most topspin, which is the exact rotational axis that someone throwing a baseball will throw. And the average speed for each foot was about 99.9% of the previous foot. In other words, if a ball averaged 100 feet per second over the first 50 feet, it would average 99 feet per second over the first 60 feet. I verified this against all batted balls hit 70 to 100 mph, at a distance of 150 to 300 feet (which is our universe of outfielder throws), and the decay rate of 0.9989 held up pretty well. So, mathematically, the average speed of a throw is ReleaseSpeed * .9989^Distance.

In our case of a bit over 200 feet, this sets the average outfielder to throw that ball at an average speed at just under 100 feet per second. So, with just over 200 feet at just under 100 feet per second, the flight time will be about 2.05 seconds. We also need an exchange time (from outfielder ball pickup to release), as well as tag time (from infielder catch to tag). Again relying on empirical data, we will set that value at 1.2 seconds. (As I work on this more during the off-season, that value will become more precise, but to move forward, 1.2 seconds is reasonable.) And so, with 2.05 seconds of flight time, and 1.2 seconds of transfer time, that gives us 3.25 seconds (give or take).

If you remember, Hampson will get there in 3.14 seconds, give or take, and the outfielder will get the ball in there in 3.25 seconds, give or take. This is pretty close to a 50/50 play. Which is what happened (video).

All Plays

We can now look at every batter-runner and see how well the model holds up. There are three kinds of plays:

• Did the batter hold up at first base for a single
• Did he go for two and ended up with a double
• Did he go for two and was thrown out at second base

The data points we will plot is the distance of the outfielder to our target base, as well as the time it would take the runner, from the start of the segment, to get to our target base. The data has been color coded as yellow (batter held up), green (batter went for it and was safe), and red (batter went for it and was out). Click image to embiggen.

We can draw baselines that splits the data into three regions:

• Typical singles, batter will hold at first base
• Typical doubles, batter will stand up at second
• Decision-making region, the grey area, where the batter might stay or go for it, and if he goes for it, he may get thrown out

And now that we have our regions split up this way, we’ll be able to start creating leaderboards: how often does each specific batter have to decide on whether the go-for-two is an opportunity. How aggressive is the batter-runner in going for two, and how successful is that batter-runner?

In a later iteration, we’ll also look at it from the outfielder perspective, and we’ll finally be able to quantify the “respect” of the outfielder arm.

All of that and more is what we’ll explore this off-season.

Monday, September 28, 2020

Cy Young Predictor 2020

​As you know, the simple Cy Young tracker has done very well since its start several years back. The system is simple enough:

IP/2 - ER + SO/10 + W

Easy enough to commit to memory, and clear in what it’s doing. Innings, earned runs, strikeouts, and wins. Each voter may use other metrics like WAR from Fangraphs or Reference, they may use shutouts, or FIP, or complete games, or quality of opponent. But those are tertiary level considerations, and they all end up canceling out, as all the voters are focusing, at a minimum, on the big 4.

This year will be a challenge, since a third of a season is not enough for the system to separate players. Normally, we’d get 2-4 points of separation. We can’t get that this year for the most part. Anyway, time for the predictions.

AL

1. Bieber. He’s so far ahead of everyone, it’d be a shock if he’s not unanimous. He leads in ERA, Wins, Strikeouts, and second in Innings. Regardless of whatever point system I’d create, he’d end up number 1.
2. Gerrit Cole
3. Keuchel
4. Maeda
5. Bassitt

After that we have Lynn and Ryu. Keuchel over Maeda is a bet on ERA over strikeouts. Bassitt over Lynn is a bet on ERA over strikeouts and innings. Lance Lynn will be the biggest test to the system. Any time you have an extreme case like Lynn, leading the league in innings, but with a fairly high ERA for a Cy Young candidate, it’s a test as to whether the system overweights or underweights a category.

It would seem that both Keuchel and Bassitt have to both appear together. It’s just hard to choose one over the other. With Bieber and Cole, that leaves one spot. And so, more likely Maeda over Lynn. As usual, no reliever will make an appearance in the top 5.

NL

1. Darvish
2. Bauer. They are neck and neck, both at 38 points. The difference is a rounding error, but an error in favor of Yu. Tertiary level stats like FIP favors Yu, so that’s my guess. But, this is a 52/48 kind of guess.
3. deGrom
4. Lamet. Another neck and neck, and they are even closer than Darvish/Bauer. In this case, even though the rounding error is in favor of Lamet, I’m betting that voters will use deGOAT as the tie-breaker.
5. Burnes.

Friday, September 25, 2020

Value scale of players

Ben does a good job going through the history of valuation scales. Below was the email response I gave him for this article.

Read More

Run Values By Pitch Count

The typical way we think of run values is at the plate appearance level. This is something most of us learned through my saber-hero Pete Palmer in The Hidden Game of Baseball in the 1980s. The idea is that every base-out situation has a run potential. And after the event, the new base-out provides a new run potential. The CHANGE in those run potential is what we attribute to the event. A strikeout with bases empty and 0 outs for example turns the run expectancy from .481 runs to .254 runs. And so, the change in run expectancy, or the run value, of the strikeout is -0.227 runs. If the bases are loaded with one out, a strikeout has a run impact of a whopping -0.789 runs. So, the impact of the event is highly dependent on the circumstances.

We can go beyond the base-out situation though. We can include the inning and score. That’s what Win Expectancy and Win Probability Added address.

We can go the other way, and in addition to the base-out, also include the ball-strike count. In other words, going from a 0-0 to a 0-1 count is good for the pitcher and bad for the batter. A called or swinging strike changes the run potential downwards. But we can go even further and combine the ball-strike count and the base-out situation. And so going from 0-0 to 0-1 with bases empty 0 outs has a run value of -0.04 runs. But doing so with bases loaded and 1 out is -0.14 runs. ​I posted this long overdue chart a couple of years ago. That shows the run expectancy of all 288 base-out-ball-strike states as well as the run values.

These run values is what you see on Savant, like on the Swing-Take Leaderboard.

Thursday, September 24, 2020

Statcast: 2020 models refreshed

At the start of each season, we create models for Hit Probability, Catch Probability, Jump, Infield Defense, Catcher Framing, wOBA, and so on using historical data, for the upcoming season. At the end of each season we refresh those models to match the actual environment of that season. Therefore, after the 2020 regular season completes, we’ll be refreshing all the various models to make them current some time between Monday and Wednesday.

Sunday, September 13, 2020

Scott Karl: the .500 pitcher

Scott Karl was average. His W/L record was 54-56. While meaningless, it’s a nice data point. He pitched in 5 full seasons and one goodbye season, totaling 1002 IP. His ERA+ was 100, which is exactly league average, and his ERA- was 101, a smidge worse than league average. His FIP- was 104, again a shade worse than league average. .500 pitchers have value of course, because the alternative is much worse. But they have no value for the Hall of Fame. No .500 pitcher, no matter how many years they pitch, will ever make the Hall of Fame.

But a great pitcher that pads his career with a .500 seasons? Those for some strange reason DO have value. It’s a padding of counting numbers.

Johan Santana had a 139-78 record. If he finished his career like Scott Karl’s entire career, we’d add 54 W and 56 L. He’d come in at 193-134, with a 3.74 ERA, which is 121 ERA+.

Who is that like? Well, David Cone was 194-126, 3.46 ERA, and 121 ERA+. Orel Hershiser was 204-150, 3.48, 112. How you think of the Hall of Fame all depends on what you think of these pitchers.

You can of course do the same to Cone and Hershiser, and pad their careers with Scott Karl’s career. My guess is that you’d get maybe Dennis Martinez? All of them are worthy in my view.

So to get an appreciation of a pitcher’s impact for the Hall of Fame, and you rely on counting stats, pad them with the stats of a .500 pitcher. While a .500 pitcher has value to a team in any given season, this pitcher has no Hall of Fame value. And so, you are essentially padding with (effectively) zeroes.

Who is John Wockenfuss?

A Bill James reader noted that John Wockenfuss was sixth in the AL in 1980 in WPA, despite only 444 plate appearances.

I started following baseball in the late 1970s, collecting cards, pretty much knowing the names of every player, every year, on every team. Wockenfuss is no stranger to me. Quite the contrary, I could place him immediately. But naturally, this took me for quite a surprise.

WPA measures exactly what it sets out to measures in exactly the way it was meant to. As for John Wockenfuss:

You can sort by the WPA column at the end. And compare to RE24. If it exceeds 1:10, then you know that the performance happened late in a close game.

That top one is +.477 wins added with +1.90 runs added. That would normally mean +.19 wins added, but instead it’s almost +.48. That’s 0.29 extra wins due to leverage.

So, you click the boxscore to see what happened.

And Reference helpfully provided the “top plays”.

And we see he hit a solo HR in the bottom of the 8th to tie the game. Adding 1 run, which should be 0.10 wins in a normal situation is a whopping +0.30 wins here.

***

The day before was his 2nd most impactful game. He was a PH: And in the bottom of the 8th, with 2 runners on, he tied the game up with a 3-run HR. In one shot, he added 0.42 wins.

***

There’s places to use this stat and places to not use this stat. Just like ANY stat in the world. This is a story stat, and this stat tells the story perfectly. You can talk to John Wockenfuss, and say “Remember those two days in September against the Jays when you were on top of the world? Well, we have this stat that will remind people of it forever.”

Sunday, September 06, 2020

Statcast Lab: Why are clubs shifting RHH ?

​Back in March of 2017, I presented preliminary research that showed some massive inefficiency in shifting RHH.

Focus on the wOBA with RHH v Shift, with no runners on, and the 43 point gap.

Now, I have not controlled for the batters, pitchers, fielders. But SOMETHING is going on. Whether only great RHH are being shifted and/or bad pitchers and/or bad fielding teams.

OR, it could also be a massive inefficiency in fielding alignments with RHH, with regards to shifting. Basically, if you put the SS and 3B too close to each other, and you don’t have the 2B covering up enough of the open spot the SS abandoned, you are basically being very inefficient in the fielding alignment.

I have finally controlled for the batters and pitchers. And there’s a 38 point gap since 2015. Here’s what I did:

1. As earlier, I looked only at bases empty situations
2. I tabulated wOBA for each combination of batter+pitcher over the 2015-2020 time period
3. I broke up their matchups whether with a shift or with a standard fielding alignment
4. They had to face each other at least once in each of the two situations over the time period
5. Each matchup was weighted by the Harmonic Mean
6. I found the difference in wOBA with the shift minus without the shift
7. I then figured the weighted average by bat-side, by using each batter-pitcher matchup’s Harmonic Mean

(Reminder: a shift is 3+ infielders to one side of the bag.)

Here’s how it looks by bat-side. For LHH, we have 19,334 weighted PA. With no-shift, the batter-pitcher had a .341 wOBA. With the shift, the SAME batter-pitcher had a .317 wOBA. That’s a 24 point drop because of the shift. These batter-pitchers faced the shift 49% of the time. Shifting LHH is a good thing for the defense.

For RHH: 18,676 weighted PA. With shift: .364. Without shift: .326. Difference: 38 point gain with the shift. These batter-pitchers faced the shift 42% of the time.

I thought maybe some RHH deserved to be shifted, that maybe those shifted over 50% of the time are “obvious” candidates” while those shifted less often are the ones bringing up the average. No dice. Across the board, regardless of how often a RHH was shifted, they had a huge gain with the shift.

My point from March 2017 remains: why are clubs shifting RHH at all?

Runs Per Win in a 7-inning world, Or, What Pythag Exponent now?

​The 7-inning game is throwing us for a loop everywhere. I had to adjust the win expectancy chart. That one was simple: just start the game in the third inning.

Now, what about the runs per win (traditionally 10) or the Pythag Exponent (traditionally 1.8)? Just to confirm those numbers, I looked at all team-seasons from 1998-2019. The runs-per-win relationship was 10.06 runs per win, and the Pythag Exponent was 1.86.

If it was a 1-inning game by the way (or you just looked at extra innings), the Pythag Exponent would be exactly 1. In other words, the relationship of Runs Scored to Allowed would match the Wins to Losses. That is, the percentage of runs you score is equal to the percentage of games you win.

For a 7-inning game, what we can do is look at all games where entering the third inning, both teams were tied. When I do that for the above seasons, the runs per win converter is 9.12, and the Pythag Exponent is 1.57.

If an aspiring saberist wanted, they could come up with the Pythag Exponent and the Runs Per Win converter for 8-inning games and 6 inning games, and all the way down to 1, just by using this simple technique.