It’s both good and necessary to have context.
In 1979, one of my least-favorite ballplayers ever, Doug Flynn, drove in 61 runs despite a .582 OPS while batting eighth for the 99-loss Mets. That seems like a lot of context. But we’re missing at least one vital piece of information. How many opportunities did Flynn have? He had 174 PA with RISP. Is that a lot? It ranked fourth on the Mets that season, which is more than you’d expect from an eighth-place hitter on a lousy team.
David Pinto’s RBI% calculator shows Flynn had a 14.92 RBI% that year, directly behind offensive dynamos Aurelio Rodriguez, Bake McBride and Dane Iorg, and 136th overall among those who had at least 100 runners on base that season. That seems to indicate that Flynn’s elevated RBI total was due to having a lot of chances, rather than any special skill in that situation.
Is that enough context, though? Do we need to know each of the RE24 situations that Flynn batted in that season? Do we have to know each individual pitcher he faced? Do we need to know how many calls the home plate umpire got wrong in each PA of his? Perhaps any and all of those things would be helpful to know and would get us closer to the right answer.
But, ultimately, players are not robots and do not perform the same each day or each situation. For whatever reason, Flynn in ’79 performed significantly better in RISP situations than he did overall. And there are dozens of things like this which might explain why a certain player did better or worse than expected in a particular situation or a particular game.
Our tendency is to add more and more variables to try to hone in on the right answer to whatever the question is. And this has the potential to be either helpful or detrimental. Making up something completely – does it help to know that ’79 Flynn was 3-4 against RHP in night games with a runner on second base against pitchers who already threw 75 pitches?
At some point, you need reliable sample sizes for your conclusions to mean something. And the hope is that a sufficiently large sample will negate the flukes or the less-than-ideal conditions that inevitably occur – like Flynn driving in 61 runs or Brandon Nimmo trying to play thru a neck injury or Alex Torres being brought in to face a LHB in a key situation despite a history of reverse platoon splits.
Keeping all of that in mind, let’s look at how the Mets fared against Sandy Alcantara Friday night.
Alcantara is one of the best pitchers in the game and the Mets scored four runs against him and chased him from the game after five innings. With that as our context, we can say that it was a good outing for the Mets’ hitters, perhaps even a great one. We can add even more context, that Alcantara came into the game with a 1.81 ERA and had pitched at least seven innings in 13 of his previous 14 games. He’s really, really good.
But, as he’s not a robot, Alcantara was not nearly as good last night as he’s been the great majority of 2022. When he gave up back-to-back extra base hits to Nimmo and Starling Marte in the second inning Friday, he was clearly not the same guy who pitched a complete-game shutout on June 8. Was the context of Alcantara in 2022 more important than the context of Alcantara on 7/29/22?
We can’t possibly know how a player is feeling or how he will perform in any particular game or moment. With that caveat – is it better to view things thru the prism of the individual, in this case one where he had 144.1 innings pitched or is it better to look at a sample of all players, one at least 10,000X bigger?
Maybe Alcantara was under the weather. Perhaps he got bad news about a good friend or family member and was pitching with a heavy heart. Or it could be something completely different that caused him to have an off day. Hey, it happens – even to someone as great as Alcantara
Because of the potential for any number of things to influence one particular game, my opinion is that the best way to approach the question of how good the Mets’ offensive performance was last night is to use the universal sample, rather than the individual one. And the universal sample is the RE24 mentioned earlier, the Run Expectancy matrix for each of the 24 base-out scenarios.
Let’s use the RE24 from FanGraphs, which if memory serves is a Tom Tango creation using results of multiple seasons. Here are the Mets’ run expectancies while Alcantara was in the game. In innings they did not score, we’ll use the highest RE and for innings they did score, we’ll have multiple RE to consider.
1st inning – 1st and 2nd, one out – RE 0.908 runs
2nd inning – 1st and 3rd, two outs – RE 0.471 runs
2nd inning – 2nd and 3rd, two outs – RE 0.570 runs
2nd inning – 3rd, two outs – RE 0.413 runs
3rd inning – 2nd, no outs – 1.068 runs
4th inning – nobody on, nobody out – 0.461 runs
5th inning – 1st and 3rd, one out – 1.140 runs
That works out to a total of 5.031 runs and they scored four, so they left a run on the bases. And it’s likely that this simple way of looking at things underestimates the actual RE in innings were runs actually scored. In the fourth inning, the Mets scored on a solo HR. If we counted this, the RE for the inning would be higher than one because there was a 0.095 RE after the HR. So, instead of the 0.461 RE given above, a more detailed one would have it at least 1.095 RE.
And a few more things to consider. The RE24 assumes a 4.15 run environment. MLB has a 4.34 rpg environment, so all of these numbers should be a tiny bit higher. And if you think the individual and the team should count more than overall sample, the Mets have a 4.67 rpg average.
Finally, if you want to look at teams and individuals – who drives the individual battle, the pitcher or the hitter? We’ve established that Alcantara is great. But Nimmo, Marte, Pete Alonso and Francisco Lindor are not exactly chopped liver. If you look askance at the RE24 because it includes the results of a Thomas Szapucki – should we not have the same revulsion that includes the results of a Travis Jankowski, too?
Among pitchers and batters who qualify for the leaderboards, no pitcher allows HR at the rate that Aaron Judge hits them. In fact, his HR/FB rate is 36.3% while the worst pitcher gives them up at a 17.5% clip. Hunter Greene has the highest ISO among qualified pitchers with a .238 mark while 18 different hitters have a higher ISO, including Judge at a .372 mark. When Alcantara gives up four doubles, a triple and a homer in five innings – shouldn’t the batters get some credit?
No pitcher strikes out batters at a rate that Brandon Marsh whiffs, which is 36.5% of the time. Xander Boegarts has a .387 BABIP, which is higher than any pitcher’s mark in the category. Why should a pitcher be the sole driver of run expectancy?
It’s certainly easier to focus on one variable (pitcher) rather than nine variables (hitters) and while we prefer simplicity if that gets you close to the “right” answer – it’s far from a slam dunk that focusing solely on the pitcher at the exclusion of the batters works in this matter. That’s why I prefer to use the universal RE24.
No doubt Alcantara is a great pitcher but just the same I’d like to give some credit to the Met bats. It is likely that he was just slightly off on some pitches but the one that Marte smoked was all about Marte getting there. The Mets also seem to chase less which runs up the pitchers pitch count and hastened his departure. Doug Flynn hit .243 in 1979 so he averaged almost 1 hit in 4 AB. Plenty of chances to knock runs in if they are on base in front of him. That he got 61 though is amazing given that it was a terrible team. He was also 4th in RBI on the team with Lee Mazilli and Richie Hebner knocking in 79 and john Sterns 66. I did think Flynn would be better as one of the solid players coming here in the Seaver trade fiasco.
Thanks for the interest analysis – as always.
Seems like the excess of extraneous data confuses & obscures rather than sheds light.
But it is helpful to know the number of PA w/ RISP. (I’d love to see that qualified, since, say, man on 3rd, nobody out, infield back, is much different than man on 2nd, two outs: there must be numbers derived from the “run probability matrix” that would sharpen the blunt tool of RISP opportunities.
Another helpful bit of data, which I’m surprised isn’t here, would be the slash line for those ABs.
And that would be enough for me to give us a rough (but better) indication of what’s going on that purely getting RBI totals or counting PAs with RISP.
Applying the average run expectancy chart against Alcantara’s start yesterday shows than an average pitcher would have given up 5 runs. But Alcantara is not an average pitcher, because he’s better at stranding runners and preventing runs, so he buckled down and only gave up 4.
That’s not “leaving a run on the table”, as you put it, that’s an above average pitcher being better than average.
Sandy Alcantara or Jacob deGrom’s personal average RE table is a lot better than the league average RE table which is also better than Thomas Szapucki or Scott Rice’s personal average RE table.
Outstanding point Name. Hat tip. But certainly a nice bit of work Brian. I wonder if the Mets weren’t as good as they are, would it have been only two runs? I brought up Nido’s base running mistake yeaterday, for example, and how Marte saved him.
Interesting read, thanks as always. I do agree with Name in that Alcantara is far above average, so kudos to the Met bats for delivering against one of the best.
Easy on Scotty Rice there, he of career HR rate of 0.3 per 9 innings. For some context, his arm may have be the most abused in Met history and he may have had a record pitch count in what good old Mickey Calloway eloquently referred to as dry humps courtesy of Terry Collins.