Fielding Run Value (FRV) is what you will find on Savant (and Fangraphs) and is the metric I spearheaded.
DRS (Defensive Runs Saved) is what you will find on Fangraphs and Reference, and is the metric spearheaded by John Dewan.
Now, one way to measure a metric is to see how well it can predict the OTHER metric in the year after. I've done this for Catcher Framing for example, where we learned that the Steamer metric actually predicts next year's Savant Framing metric as good as current year's Savant does. In other words, Steamer is value-added, as it is like Savant, and more.
So, this is what I did, and given the incredible layout of Fangraphs, it took me literally under 5 minutes to run the study. I exported everyone with 600+ innings and removed catchers. I turned everyone into runs per 27 outs. I correlated year T to year T+1, matching on player and position. This left me with 635 matched players.
First, how does each correlate to itself? For FRV (that's the Statcast version), it's at r=0.60. For DRS, it's r=0.50. This is a pretty good sign that FRV is better able to isolate the players (though you might argue I haven't proven that I've taken care of parks, so maybe I should look at team switchers... someone out there can pick that up).
Now, how does DRS correlate with FRV? In other words, can DRS explain FRV? That correlation is r=0.38. That's not bad. It shows that DRS sees itself and FRV different enough, though it's still able to explain a good portion of FRV.
How about FRV explaining DRS? That correlation is r=0.40. That's not bad as well. The same explanation holds, though in this case, FRV is able to explain itself to a higher degree than DRS can explain itself, all the while being able to explain DRS slightly better than DRS can explain FRV.
The knockout punch isn't there. It would have been great if FRV would have had a correlation of r=0.50 to DRS (and thereby matching the DRS correlation to itself of 0.50). That didn't happen, with an r=0.40 instead. It would have been interesting had FRV had a correlation of r=0.38 with itself as that's the knockout punch DRS would need. That obviously didn't happen, as it instead had an r=0.60.
So, there's enough here to suggest that both have value, though the value is stronger with FRV.
INFIELDERS v OUTFIELDERS
Now, the FRV method is really an Outfield and Infield method, two separate methods. I suspect that DRS likely has two somewhat distinct methods. So, let's repeat all that, but look at infielders-only and outfielders-only.
With the Infield, DRS correlates with itself at r=0.49, while FRV is at r=0.46. Slight advantage to DRS for self-correlating better. DRS correlates with next season's FRV at r=0.30, while FRV correlated with next season's DRS at r=0.28. Overall, it certainly looks like DRS has a slight advantage. I'd probably call it 55/45 for DRS here. If you want to call it 60/40 in favor of DRS, ok, I won't argue. DRS has two things going for it, one is the DP handling and the other is the little things, the nuances, of playing the infield (like relays, and other subjective calls).
How about the outfield? Well, get ready to get your mind blown here. FRV correlates with itself at r=0.73, while DRS self-correlates at r=0.51. I mean, this is just no contest at all.
But, it's not just that. I'll give you the knockout punch as well. FRV correlates with next-season's DRS better then DRS correlates with itself next-season: r=0.53 to r=0.51. Let that sink in for a bit. FRV knows nothing about DRS, knows nothing about how DRS measures things. And yet, it can predict next season's DRS better than DRS can (for outfielders).
Indeed, DRS can predict next season's FRV almost as well as it can predict itself: r=0.51 for self-correlation and r=0.49 for correlating FRV.
Why does this happen? Because the starting point of the outfielder and how much distance they have to cover is critical, and Statcast can precisely measure this.
In terms of weighting, I'd have to go at least 90/10 for FRV, if not 100/0.
It's clear that in the off-season, my time should be spent much more with infielders, and handing all those extras that I've been putting off. DRS deserves its flowers there.
There are three types of HR saving plays. I will go thru each one, then give you the analogies in the infield, and in other parts of baseball.
The first is the easy one: the presence play. The wall is low enough, say under 8 feet. The ball is in the air long enough. And the outfielder is playing deep enough that they can lightly jog to the spot they need to be. All they really have to do at that point is lift their arm. This is a play that the best or worst outfielder is going to make, whether you are Kevin Kiermaier or not.
The infield equivalent is a 4-3, 5-3, or 6-3 play. All that is needed of the 1B is to get to the bag, and wait for the throw. While the 1B is obviously critical in the play (he is after all getting the putout), the skill required is one that does not differentiate itself among the best and worst 1B. Every 1B will be there.
If the 1B is not there, it was probably because he counted on the pitcher to be there, almost always because the 1B is the one who fielded the ball to begin with. This kind of misplay, the pitcher not covering, happens often enough that it is both embarrassing and not newsworthy.
The way we handle the evaluation of the pitcher not covering is to figure out how often the AVERAGE pitcher does not cover. For the sake of illustration, let's say that the average pitcher fails to cover 10% of the time, and so does cover 90% of the time. (I don't know what the actual number is, maybe it's 5% or 1%, I'm using 10% for illustrative purposes only.) So, when a pitcher DOES cover in scenarios where the out is otherwise assured, he would get +.10 outs. When the pitcher does NOT cover, he would get -.90 outs. The average pitcher in this illustration would get some combination of +.10 and -.90 such that the total Outs Above Average (OAA) is exactly 0. A pitcher who ALWAYS covers would get +.10 for each putout, and if they had say 100 putouts, would have +10 OAA... just by being attentive. Again, this is based on the 90% coverage scenario. If it was 95% is the average, then the fully attentive pitcher would get +.05 OAA times 100 plays, or +5 OAA.
The attentiveness or presence play happens elsewhere on the field. The 3B failing to cover on a steal of 3B, or a pitcher not backing up the catcher on a throw from the outfield. So, you can figure out the OAA of a fielder's attentiveness simply by counting. We don't do this, but we should. That's a gap in the fielding recording.
You can see it elsewhere as well, say the 3-run save. A 3-run save is VERY different from a 1-run save. But, a save is a save in the record books. This is why we don't like saves as a category, because we know a save is NOT a save all the time. A 3-run save is going to be saved say 96% of the time (or whatever the number is). So, a 3-run save is actually going to be worth +.04 "Saves Above Average", while not holding that lead is worth -.96 SAA.
Back to the outfielder: how often will an outfielder fail to camp themselves at the warning track for what would otherwise have been an easy out? Let's say that's 1%. So, if an outfielder in these kinds of HR saving plays simply does their job and holds up their arm without jumping, they will get +.01 OAA. If they don't get there in time or they don't make the out, that's -.99 OAA. On average, this will be 0 OAA league-wide.
Suppose you disagree. Suppose instead you want to give out +.50 OAA on these kinds of plays (high fly, just clearing a low fence). Well, for every 100 plays, you'd give out +.50 OAA 99 times and -.50 OAA 1 time. That's a total of +49 OAA for every 100 such plays. Does that make sense? No. You can't make every outfielder above average.
So, this was alot longer that I thought I'd write, so I'll break this up into multiple parts. See you tonight for part 2.
Last year, I ran a correlation of Framing for Savant and Steamer to next-season's Savant and Steamer. The idea is similar to FIP and ERA, and correlating to next season's ERA. When FIP predicts next season's ERA better than ERA itself, this immediately tells you that ERA has alot more noise and/or that FIP has a more precise accounting of a pitcher's performance. In the case of FIP and ERA, both are true.
With Framing, the conclusion was inescapable: Steamer is better, and Savant is best characterized as Steamer-lite, adding no new information. And this is as expected. After all, Savant uses larger bins, and considers less variables. The only thing that was left to determine was if Savant did anything special that Steamer did not. As of 2022, it did not. I can tell you it DID provide extra information in the past. But after I told the developer of Steamer about this insight, he incorporated a change to close off this gap. There is of course a benefit of having a lite version, or a Naive version. Chasing every last decimal place, every precision point, for extra complexity has a cost to it. That's why having a Naive version is helpful. So, if you had to use one, use Steamer. But if you had to use something simple to explain, use Savant.
FIELDING
Ok, so now I turn my attention to Fielding. Fangraphs makes life very easy, with data from UZR and DRS readily available, side by side with OAA. So, I pulled all that data from 2016-present (excluding catchers). I looked only at players with at least 300 innings for each position, year to year. Since all three methods have their own distinct approaches, one would think that each will be able to provide extra information. Did this happen?
UZR
Let's start with the clear laggard, UZR. Year to year correlation of UZR is at r=0.31. Taking OAA of year T and correlating to UZR at year T+1 gives us r=0.29. This is almost a FIP/ERA situation. OAA knows nothing at all about how UZR is constructed. And yet, using UZR in year T and OAA in year T are almost identical in helping explain UZR in year T+1. DRS is almost as helpful as OAA, with an r=0.27, in explaining next season's UZR.
And what happens if we put all three metrics together in year T, in explaining UZR in year T+1? In that case, DRS almost completely disappears. A weight of 20% UZR, 13% OAA and 2% DRS is what you need to explain UZR in year T+1. So, whatever UZR is trying to do, OAA is able to find it somewhat, and DRS add almost no new information.
DRS
Ok, how about DRS? The year to year correlation of DRS is r=0.42. Using OAA from year T to predict DRS in year T+1 gives us r=0.39. So, again, OAA is able to predict next year's DRS almost as well as current year's DRS. And if we consider all three metrics to explain next year's DRS? Well, that 28% DRS, 28% OAA, and 2% UZR. In other words, OAA and DRS are equally valuable in explaining next year's DRS.
Why would that be? Well, like with Steamer/Framing, OAA has insights that DRS simply does not have available. And those insights have value in explaining the fielding talent of the player, and so, explains next year's performances.
OAA
Finally, OAA. Year to year correlation for OAA is r=0.53, which is much higher than the others. DRS has a correlation of r=0.34 to next year's OAA, while UZR is r=0.27. That number, the 0.27, establishes basically a minimum-type of correlation level. UZR and OAA do nothing the same, though they naturally are based on the same plays. We'd get a correlation based simply on that.
Anyway, what happens when we put all three metrics together? Well, the correlation remains at r=0.53, and the weighting for DRS and UZR are each close to zero. In other words, each of DRS and UZR provide no new information in trying to explain next season's OAA.
Based on this analysis from Hareeb, it would seem to be the case.
If you tried to predict next season's wOBA, you wouldn't use current season's wOBA, but rather the components of wOBA (which of course is BB, hits, HR, SO, etc). While SO has the same coefficient as a batted ball out in wOBA (zero), we'd prefer keeping the SO from the other batted ball outs distinct, so we can treat them as toolsy, and therefore, each will get its own weight, for predictive purposes.
With OAA, on 2+ star plays, this is exactly what I've done. I identified the different components of the plays, and established a value for each component. Now, if I didn't do this carefully, then the components themselves might end up looking random. And the research shows that, at least for Reaction and Sprint Speed, it holds up pretty well. The tougher one is Burst, which is expected. Burst is a bridge between Reaction and all-out speed. So, how I draw the lines between Reaction and Burst means there might be some carryover between them. But also, how I draw the line between Burst and all-out speed also means some carryover between them. Burst is basically the component that is most ripe for this effect, and the research is consistent with that.
In any case, the OAA components work well enough to advance our understanding of fielding.
I recently published this blog post on OAA and Francisco Lindor, and in my followup comment, I made a point regarding Win Shares.
I had earlier sent that article and that addendum to Bill James, who was kind enough to read it, and he responded to it below. I have then included my rebuttal to his response. Everything in quotes is from Bill.
Francisco Lindor is probably the best fielder in baseball from 2016 to 2022.
We can see that using Outs Above Average, at +118 OAA, which translates to 89 Runs Prevented compared to average.
Bill James in his Win Shares system also has Lindor as the player earning the most Fielding Win Shares in that same time period, at 52 Win Shares.
Drilling down
Now, what does +118 outs mean really. Lindor was responsible for 3406 batted balls, of which he converted 2590 of those into actual outs. What would a league average fielder have done? A league average fielder would have converted 2472 of those plays into outs. And 2590 minus 2472 is +118.
Let's break down those 3406 batted balls. A substantial portion of those are routine outs or near-automatic hits. Let's look at the routine plays first. About one-third of the plays are routine. For Lindor, that's 1116 plays. Lindor made 1093 outs, which is an out-conversion rate of 97.9%. An average fielder would convert 97.8% of the routine plays into outs, or 1092. Lindor is a tiny +1 outs above average on the routine plays. Which makes sense of course. On these routine plays, there's not much skill involved. All you have to do is not blow it.
He was also responsible for 485 automatic or near-automatic hits. The way this works is we tag every hit to whichever fielder was the closest to making the play (based on their starting location at pitch release). In some cases, like these 485 plays, there's an almost impossible chance to make an out. The league average fielder makes 0.5% of these plays (or half of one percent). So, we'd expect just two outs of these 485 plays. Lindor however made 7 outs, or +5 more than average. Obviously not a big number, but every little bit counts.
These two categories of near-automatic outs-or-hits accounts for 1601 plays and represent mostly noise. We have another 1805 plays to consider. We can continue breaking up these plays based on their difficulty level. He faced 611 plays where the league average fielder converts them into outs at a 90% to 94.99% clip, for an average of 92.5%. Lindor actually made 569 outs which is 93.1% of these plays turned into outs. This is 4 more outs than we'd expect from the league average.
Plays of skill
The next set of plays is where Lindor starts to build up steam. There were 364 plays where the average fielder converts them into outs at a 85% to 89.99% clip, an average of 88.0%. Lindor on the other hand made 341 outs, and 341 outs out of 364 plays is 93.7%. This is a healthy +21 outs above average. While not routine plays, they are also not difficult plays. And Lindor made them look a bit routine. (Click to embiggen)
In almost every single category of plays, Lindor made more outs than the average fielder. Add it up, and we get +118. The outs to run conversion is at around 0.75 runs per out, so this translates to +89 runs better than average.
Another way to look at it is to focus on those plays that require enough skill to stand out from the average. Focus on bins 3 through 18. The league average SS turned 70% of those plays into outs, while Lindor turned almost 80%. That seems reasonable right? That in plays requiring some level of skill, Lindor would make 10% more plays than the average fielder. If you look at the chart above, you can basically see that.
Since Lindor was on the field for 8240 innings, he played the equivalent of 5.65 full seasons since 2006. And so, +89 runs over this time period averages out to +16 runs per season, which translates to +1.6 wins above average. We are able to identify for every single play, one by one, all 3406 batted balls, where each of these outs advantage comes from. And identifying the outs means we've identified the impact in runs and therefore wins.
The best fielder in baseball is worth about +1.6 wins above average, per season. And this can be explained on a play by play basis.
Here are all the catcher framing numbers for Savant and Steamer, since 2015, min 100 innings. The correlation is an extremely high r=0.92. In other words, they are both coming to very similar conclusions, even though their methodologies are independent of each other.
Is one better than the other? To the extent that I would choose a winner, I would choose Steamer. If we look at year to year correlations of Savant to itself, we get r=0.51. We also get the same correlation if we run a correlation of Steamer to next year's Savant. And we also get the same r=0.51 correlation with Savant to next year's Steamer. Where we get a slight win is Steamer to itself has an r=0.56.
Indeed, when we include both Savant and Steamer to forecast next year's Steamer, the correlation remains at r=0.56, showing that Savant is not including any new information. To that extent, I would argue that Savant is a subset of Steamer, or a Steamer-lite. If we correlate to next year's Savant, the correlation goes to r=0.52, which barely budges from the single metric correlation of 0.51.
These results, at least directionally, is as expected. Savant intentionally keeps the zones larger, while Steamer uses more precise plate location. Steamer might even create different shaped zones for bat-side, implicitly or explicitly including umpire effects. And for all these extra considerations, the gains are fairly modest.
Indeed, if you simply used called strike rate in the Shadow Zone, and did nothing else, the correlation would be just below .90 against Steamer. In other words, just doing the absolute minimum gets you most of the way there.
And so if you are trying to explain Catcher Framing to someone, and to try to convince them that the impact is real, simply quote the Called Strike Rate in The Shadow Zone. Jose Trevino for example led with a 54% called strike rate, against a league average of 47%. Getting an extra .07 strikes on the edge (meaning flipping a ball to a strike 7% of the time), on 2719 pitches, that comes out to an extra 190 strikes. That's a whipping 190 extra strikes. Divide by 8 to get that into runs. So, 190/8 is 24 runs. That's a simple back of envelope calculation. Doing a bit extra work that Savant does, and a better estimate is 17 runs prevented. Doing a bit more work that Steamer does, and they get Trevino with 19 runs. They are all in the same ballpark, however you try to figure it out.
Having earlier looked at it league-wide, the focus now will be to control for the quality of batters and pitchers when shifts happen. I will further limit it to bases empty.
From 2018-2021, the quality of LHH being shifted upon has been about 26 points higher than the LHH not being shifted (ranging from 22 to 30 points). In 2022, nothing really has changed, with a 31 point difference. In terms of the actual performance, from 2018-2021, the wOBA worked spectacularly well on LHH: based on the quality of batters with a wOBA of .347 with bases empty and unshifted, those batters ended up with a wOBA of .321 when shifted. So that's a 26 point drop. In 2022, we have 50 point drop. (This improvement in defense did not carry over with runners on base.) In other words, an extra 24 point gain in performance for the defense, over and above the gains we've come to expect of them.
As for RHH, from 2018-2021, shifts were not really based on quality of batting. The shifted RHH were only 11 points higher than the unshifted. So, whatever reason the clubs had didn't work at all, with an egregious result of 27 points higher with the shift than unshifted. In other words, the exact opposite of the success they had with LHH.
But in 2022, that has changed. The quality of batters being shifted is still the same, but now the clubs are having success, with a 10 point drop in wOBA with the shift, than unshifted. So, that's a change from +27 in wOBA before 2022, to a 10 point drop in 2022. Or a 37 point turnaround. That's an astounding turnabout. They've also seen an improvement with shifting with runners on base as well.
Essentially, the entire drop in run scoring can be attributed to the improved results on shifting on LHH and the spectacularly improved results on shifting on RHH. Whereas RHH shifting was actually INCREASING score before 2022, now in 2022, the clubs have figured it out, and shifting on RHH decreases run scoring as well.
As for why this has happened: I haven't looked yet, but maybe an Aspiring Saberist will get there first.
On June 21, 2018, Mookie Betts of the Redsox made this easy catch (video) on Joe Mauer at Target Field (with RHP Porcello), whereby Betts did not need to move at all, as the top image shows. Where Betts was standing was where Mauer hit the ball. (Click to embiggen.)
On the bottom left is the location of every RF against Joe Mauer at Target Field, in 2018, against RHP.
That image is blown up into the right image. The three times that Mauer faced the Redsox (all on the same day) are shown in red, with the larger red dot being the play in question. The Catch Probability was 99.5% in all three instances.
The yellow dots are every non-Redsox fielders (against Mauer, Target Field, 2018, RHP). The average Catch Probability of all those fielding alignments was 93%, with that far out blue dot at 10%.
Interlude: you might think the dot below the blue one should be even lower, and… well… you are right, for the RF, it is lower. But in that particular fielding alignment, the CF was closer to where Mauer hit the ball if we were to overlay his batted ball on that fielding alignment; and the CF on that play had a 27% catch probability. Hence, the fielding alignment in that particular case is set to 27% Catch Prob (except for the CF, not RF).
Had the Redsox positioned themselves based on a random non-Redsox alignment, but selected from ACTUAL alignments (against Mauer, Target, 2018, RHP), the expected catch probability is 93%, compared to the actual catch probability of that particular instance of Betts v Mauer of 99.5%. And therefore, we credit the Redsox with +0.065 outs for their Defensive Positioning.
And all we have to do is apply this concept to every single batted ball. In every instance, we are comparing the actual single instance fielding alignment (“with”) to the distribution of instances not involving that fielding team (“without”) against that batter, venue, year, pitch-hand.
As of right now, I only have it for the outfield. And as a necessary condition of doing WOWY (with or without you), this means that we can’t use Mauer on the road, since the road site will always have the same fielding team.(*)
(*) Almost! In rare instances, this is not true. As well, this is not true for traded players mid-season.
The last major(*) piece to finally getting a Statcast WAR in place is finally done. I’ll explain the model as to how it works for catchers throwing out runners.
(*) There’s a few minor pieces. There’s always going to be a few minor pieces. This is really a never-ending body of work. But all the major pieces have been identified and modelled.
As with every other fielding metric (Catch Probability for Outfielders, Infield Defense, Outfield Arm), the idea is straightforward: how much time it takes for the ball to beat the runner, and how much distance does the runner have to cover. In other words, the model is not only intuitive, but it exactly matches how all fans evaluate plays.
In all of these fielding plays, the critical decision is this: when does the clock start? And for the catcher, I am setting the start of the clock as when the ball passes through the back of the batter’s box. At that point, we determine the distance of the runner from the target base (2B).
In looking at the data, three considerations immediately popped out. The first is obvious: pitchouts. Pitchouts confer an advantage to the catcher. Since calling pitchouts is not a catcher skill, we want to remove pitchouts from the initial model. (We will bring it back though with an adjustment to evaluate catchers.)
The second one was less obvious, and popped up when I’d see the Caught Stealing rate dropping as the distance the runner had to cover increased. Just watching one play was enough to realize what was happening: delayed steals. While pitchouts are easy enough to spot (and we record them), delayed steals are not so obvious. So, I had to create a rule to identify likely delayed steals. As above, those plays are removed from the model, and re-inserted afterwards with an adjustment.
The last consideration was also obvious enough, once you think about it: when the runner on 1B is the trailing runner, the runner success rate is above 90%. Obviously, catchers either only try for the lead runner, or, they simply won’t even attempt to throw out the trailing runner.
With those plays removed, we can now create a model. Here’s how the actual data (in blue), binned in steps of one foot, compares to the eventual model. As you’d expect, the farther the runner, the higher the caught stealing. This is obvious enough. All we’ve done here is quantified that effect. At what point is a runner 50/50 to being caught stealing? The answer is in the chart: when the runner is ~60 feet from his target base, as the ball crosses the back of the batter’s box. We didn’t know that before Statcast, and now we do.
Roughly speaking, we see the CS rate increases by ~5% for every one foot farther the runner is to the target base. (All other things equal.) Again, with Statcast, now we know the value of each foot. While the chart looks more like 7%, most of the steal attempts are clumped toward the bottom.
Now, you will be thinking, correctly, that Buxton being 55 feet away is different from Molina being 55 feet away. And you’d be correct. So, the model actually uses the speed of each runner to convert the distance of feet to 2B into seconds to 2B. Doing so is also pretty straightforward, since we have the Sprint Speed of each runner. (The Sprint Speed is the speed of each runner on their fastest one second window, of competitive runs.) So, using their seasonal speed, we can establish how much time they have to cover the remaining feet.
What does all this mean? It means we have established the opportunity space that the catcher is facing, removing the pitcher totally out of the equation (since we start the clock after the ball has gone past the batter’s box) and adjusting for the speed of the runner (with additional adjustment for a pitchout). With all that done, all we have left is the catcher.
So who is the best throwing catcher of 2016-2021? That would be Manny Pina. In the data we have tracked, he caught 60 of 155 basestealers, which is 39%. While this figure is 3rd highest (of 68 catchers), his opportunity space (what the pitchers left him, and what runners he had to face) actually had him at the toughest situation to be in of all 68 catchers, with an “expected value” of 20%. In other words, the league average catcher, giving the opportunity faced by Pina, would have them nab 20% of the runners, Pina nabbed 39%. His overall OAA (outs above average) is +30, from 2016-2021.
Two other catchers were on his tail, Salvador Perez (+27) and JT Realmuto (+25). The catcher arms are pretty symmetrical, with the worst catchers at -26, -23, -20.
We’ll have this and alot more (perspective from the pitcher, and runners too) this off-season on Baseball Savant. We’re entering the top of the 5th.
Unlike all the other fielding metrics before Statcast, we know the location of the ball and fielders at all times. What this allows us to do is model reality in a very intuitive manner.
You as a fan can fairly easily judge a fielding play because your eyes can measure distance and time based on your experiences, much like you can figure out when you can and cannot safely cross the street based on your experiences. There is of course that gray area because the "eye test" can only get you so far. You can tell when you can 100% cross and not cross the street safely. But you wouldn't be able to tell for example a 25/75 from a 75/25 situation. But, if you knew the exact distance and speed of the oncoming traffic and you knew how much distance you had to cover and how fast you can cross the street, and had a handy calculator to instantaneously tell you the results, then, yes, you can distinguish between a 25/75 from a 75/25 traffic situation.
So that's where Statcast comes in: we know the exact location of the fielders and the ball, and the time the fielder can get there and retrieve (or miss) the ball. We know how fast the batter can run to first and where he is when the fielder picks up the ball. Basically everything you as a fan are measuring with the eye test in a very intuitive manner we can actually measure and convert into time: will the runner beat the ball or not? This is why Statcast Infield Defense works: it works because we are actually modeling reality.
This chart shows how often an infielder (2B, SS, 3B) makes the play based on how many feet they have to cover. Along the horizontal: Negative is toward 3B and positive toward 1B. Along the vertical: Negative is toward home plate and Positive is behind the fielder. "0" means "0 to 4.99 feet" and so on. The numbers represents the out rate. The green box is the starting point of the infielder.
In this simplified view, I am NOT showing how much time the infielder needs. That's why it looks uneven in some cases. So, you are getting half the view: just measuring distance. And even at that, you can see a pretty strong relationship. Statcast Infield OAA also includes the time the ball gets there, as well as the location and speed of the batter. See the above link for a more complete description.
This is probably the fifth or eighth or twentieth post on this subject since I started one back in March of 2017. Nothing has changed in conclusion, even as I change my approach in the study.
Anyway, in this particular approach, I used 2018-present, with the “baseline” for the batter and pitcher as their performance in bases empty, with the fielders in standard classic formation. Using that, I can create an “expected” wOBA for every batter-pitcher confrontation. I then compare the expected to the actual. So let me cut to the chase, and show you the results during Shift PA:
LHH, expected wOBA .338, actual wOBA .313, for a 25 point drop
RHH, expected wOBA .317, actual wOBA .342, for a 25 point gain
This is pretty much what I’ve been finding each time I study this. Basically, whatever gain clubs are getting from shifting on LHH they are giving back by shifting on RHH.
Maybe some clubs that shift on RHH buck the trend? Well, of the 8 clubs that shift the most on RHH, all 8 of them have a higher wOBA with the shift than we’d expect (based on the pitcher-batters involved in those shifts). Indeed only TWO of the thirty fielding clubs have a better performance with RHH on the shift than expected:
The Diamondbacks (expected .345, actual .330). The D’Backs are the 9th place club in using the shift the most. So, it’s possible they’ve figured it out.
And the other club with the better performance shifting RHH than expected are the Padres (.305 expected, .285 shifted). Except the Padres are dead-last in terms of how often they shift on RHH. In other words, the Padres are being exceptionally careful in deciding who to shift against, and it’s working for them. With so few trials however, it’s hard to tell if it’s just random variation or not.
In any case, the conclusions remains as always: if you have a RHH, never shift on them. Even if you think you have found the right RHH to shift on, you will be tempted to shift on more and more RHH until you end up shifting on too many RHH. The Padres have basically set the ceiling as to how often you should shift on RHH.
We have finally completed the Baserunning and Arm model of taking/holding the extra base. The model is intuitive and matches how a baseball fan processes a play. Let’s take the situation of a batter deciding whether to stretch a single into a double. The runner at contact has a certain number of seconds to reach second base. The fielder at contact has a certain number of seconds to retrieving the ball, and then based on the distance of the throw, a certain number of seconds to getting the ball to second base. The more time it takes the fielder to get the ball in, the higher the probability the runner will try for two (and succeed). The less time it takes the fielder to get the ball in, the lower the probability the runner will try for two (and if he tries, the more likely he will get thrown out).
So suppose the runner needs 8 seconds to get to second base for a particular play. On that same play, let’s say the defense needs 5 seconds to get to the ball, and another 0.75 seconds to release it, another 2 seconds for the ball to travel and another 0.5 seconds to apply the tag. That’s 8.25 seconds. We take 8.25 seconds of fielder time minus 8.0 seconds of runner time, and we have a delta of 0.25 seconds. That’s how much “breathing room” the runner has. Naturally, the runner in question won’t ALWAYS take exactly 8.0 seconds. Sometimes it may be 7.8 or 8.3. Or anything in-between. As for the defense, the fielder might get to the ball quicker than normal or slower than normal. His throw might not reach its peak, and maybe the throw is a bit offline so we need more tag time.
What we have is therefore an S-curve type of probability (a sigmoid function). The more negative, the lower the probability. The more positive, the higher the probability. This is what it looks like for the batter trying for two.
Here we see that when the Fielder Time and the Runner Time matches, the runner will try and be successful about one-third of the time. The more buffer time for the runner, the more often he will try and succeed. The blue line is actual data, while the dashed line is the model.
We’ve identified six different baserunning / arm categories:
Batter going for two
Batter going for three
Runner going first to third on a single
Runner going first to home on a double
Runner going second to home on a single
Runner going third to home on a sac fly
Each type of play has its own model, though they all follow the same principle. The “slope” of the curve is unique to each kind of play, but the structure of the model is the same.
With each play having a probability, we can now compare each runner to the baseline, and figure out how many extra bases they are taking, or how many bases they are NOT taking. As well as how often they are being thrown out. We can combine all that and come up with leaderboards for runners. Here it is since 2016:
Mookie Betts, Kevin Kiermaier, and Billy Hamilton are the best at taking the extra base.
We can also flip it on the other side, and look at leaderboards from the OF Arm perspective, crediting the outfielder not only for throwing runners out, but also holding them to their base. Here’s that leaderboard (more negative is good for the defense) here. Kevin Kiermaier is the leader (as well as Betts and Hamilton also with a strong showing).
You know all those things they say that’s “not in the boxscore”. It’s in the Statcast boxscore, and we’ll be showing the results of that, and shine that spotlight on that “hidden game” of baseball, with Kiermaier its best representative, both as a runner and as a thrower. (We already know that Kiermaier is tremendous as a fielder.) Kiermaier is the kind of player that Statcast does its best to highlight. Eventually, this will make its way to Savant, along with alot more breakdowns, so you can see it by each category and each season.
And what more can we do? Well, plenty. This for example is an Altuve play where he was thrown out by Arozarena. If you are behind the red line, that’s the nogo line. If you are ahead of the green line, that’s the go line. In this particular play, Altuve was thrown out. And we’d be able to show it, frame by frame, in video mode. We’re entering the top of the 5th.
There are 78 RHH who have been shifted at least 100 times since 2016, and whose wOBA with an without the shift has a gap of at least 20 points in some direction. Coincidentally, there are 78 LHH in the same boat.
52 of these 78 LHH lower their wOBA by 20+ points with the shift. And therefore, when a team calls a shift on these 52 players, that’s a “good” call. The other 26 players increase their wOBA by 20+ points with the shift. Therefore, it’s a “bad” call when a team shifts on those players.
In contrast, only 17 RHH lower their wOBA by 20+ points on the shift. It’s a “good” call when a team shifts these 17 RHH players. The other 61 RHH increase their wOBA by 20+ points with the shift. So, it’s a “bad” when a team shifts on these 61 RHH.
All I’ll do here is tally the results. So here are the totals of the good call rate by team, by bat-side. You will notice that 28 of the 30 teams have a good call rate on LHH above 50% of the time. Even the two teams with a good call rate below 50% were very close. In other words, among those batters that they are shift-happy on, they are choosing wisely.
The best team with a good call rate on RHH is 37%, the A’s. In other words, had the league decided to not call a shift on any of these RHH, even those “deserving” of a shift, it would be a gain. In other words, among those batters that the league has targeted for shifting, none of them are making a positive choice.
The bottom on the left is the “hook” spray angle, meaning a RHH where the LF is involved or a LHH where the RF is involved.
The bottom on the right is the “slice” spray angle, meaning a RHH where the RF is involved or a LHH where the LF is involved.
The more the ball is hit toward the lines, the easier to catch.
The other bias I noted is based on whether we have a fastball or curve (meaning the amount of time it takes for the pitch to go from release to plate-crossing). I tweeted about it two years ago:
Researcher Hareeb does an excellent job by jumping way in to look into the various biases. And based on his findings, it would seem the adjustment would be pretty straightforward, one that adjusts the time.
This is interesting, because adjusting the time is exactly how I handle the other biases in the data. When we first rolled out catch probability, there was no adjustments for going back, or considering the impediment of the wall. To handle THOSE was simply a matter of adjusting the time component. To handle the FF/CU issue will be to adjust the time component. And based on what Hareeb is showing, an adjustment to the time component will (probably) be all that is needed.
Just to make an overall point: when it comes to fielding, whether outfielders or infielders, a model MUST be based on distance and time. That's the actual behaviour of the fielders with respect to the ball. If you do anything else, all you are doing is inferring distance and time. Ultimately, you just have to ask yourself: how does this parameter affect the distance and/or time the fielder has? That's the question, that's what the model needs to be.
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…
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.
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.
Recent comments
Older comments
Page 1 of 150 pages 1 2 3 > Last ›Complete Archive – By Category
Complete Archive – By Date
FORUM TOPICS
Jul 12 15:22 MarcelsApr 16 14:31 Pitch Count Estimators
Mar 12 16:30 Appendix to THE BOOK - THE GORY DETAILS
Jan 29 09:41 NFL Overtime Idea
Jan 22 14:48 Weighting Years for NFL Player Projections
Jan 21 09:18 positional runs in pythagenpat
Oct 20 15:57 DRS: FG vs. BB-Ref
Apr 12 09:43 What if baseball was like survivor? You are eliminated ...
Nov 24 09:57 Win Attribution to offense, pitching, and fielding at the game level (prototype method)
Jul 13 10:20 How to watch great past games without spoilers