Bellos's Promsising Splits

Brayan Bello had two huge splits: four days rest vs. five or more, and day vs. night.

Both of these are about rest. Extra day, extra hours of sleep.

All of the following numbers are xwOBA adjusted for opposing team's offense over the season, vs. RHP. 

The translations of xwOBA into rotation slots (on an average team) has the top 15 pitchers as aces (because we wouldn't call then aces if there were enough to go around), and then groups of 30, which leaves the last 15 guys as needing replacement. This system happens to put the line between #3 and#4 exactly at MLB average, which seems right, right?

 

------

.360 (7 G, 154 PA) 4 days rest

.303 (21 G, 514 PA) 5 or more days.

.360 is a somewhat below average #5 starer, while .303 is a slightly below average #2.

A further breakdown:

.296 (15 G, 365 PA) 5 days rest

.313 (6 G, 149 PA) 6+ days

.296 is a great #2 , while .313 is a good #3.

-----

Like Day vs. Night (because it is!)

.353 (10 G, 220 PA) Day

.296 (18 G, 448 PA) Night

Similar numbers, eh? Here the bad split gets promoted from "somewhat" to "slightly" below average #5.

It's kind of cute that the best conditions for each of the splits differ by just .000.

But this doesn't tell you that much. You want to break it down 6 ways, 3 x 2, with the 3 being short, normal, and long rest. I mean, what if all his games with extra rest came at night?

Which is exactly what happened. And that's where it gets really interesting.

On the 6th Day He Rested (damn, again?)

To begin with, after the ASB he had 8 days of rest, and had an xwOBA of .191. Remove that, as you should, and his longer-rest goes from .313 to .337, which is worst #4 starter.

What to make of this? It's very likely that his experience with 6 days rest was minimal, right? Which means he had no between-game throwing regimen for it, and likely did what the team told him to. And it didn't work.

But you'd like to hope that he worked on that regimen over the season. We're not looking for a statistically significant result with just 5 starts, but a trend in the right direction would be a good sign. So here are the starts:

.389 (5/17)

.389 (5/30) (Really!)

.310 (6/18)

.321 (7/26)

.293 (9/27)

OK, I set you up. If you just number the starts 1 to 5, and use that as the dependent variable, that trend is statistically significant, p = .035.

It could still be random, of course (1 chance in 28). But it sure looks like Bello, who by all reports is both smart and a dedicated and had worker, figured out a routine that worked for him. 

 

Now, if we were to find another such improvement, we'd be that much more confident that he does this, right?

-----

Bello had 7 of his day game starts on 5 days rest (starts 1 through 4, and 7 through 9), and 3 on four days. Here are the 7 rested game results:

.371 (4/14)

.375 (4/23)

.380 (4/29)

.270 (6/5)

.354 (8/24)

.218 (9/4)

.260 (9/10)

Even with the August hiccup, that is also statistically significant, p=.048.

Every article online about work schedules for MLB players references night games only. But my guess is that starting pitchers arrive by 10 AM or earlier for day games. If Bello goes to bed after midnight, that's going to be tough for him.

I will  spare you the complete explanation of sleep bed-times, and just note that most night-owls make things worse by exposing themselves to blue (other than red, really) light, in the evening, which fakes out the brain into thinking it's still daytime. If someone told him about this after April, and he started a project of finding how early he needed to go to all red-light, that would explain what we see. August could well be, "hey, that really worked, maybe I don't need to start that early."

(There are other way to get extra sleep, of course, but the pattern works with those as well.)

The four night games on 4 days rest are very interesting. Overall he's .306, somewhat below average #2. BUT ...

.355 (5/4)

.279 (6/23)

.421 (8/29)

.199 (9/15)

This is also hone, road, home, road. That's .386 at home and .239 on the road.

The odds of the results happening by chance are 1 in 10, but if you look at the average event in each of the two sets, it just misses significance (p = .057).

His kid arrived too late to explain the first home game, but of course there are plenty of reasons why behaviors might be different (in either direction) home vs. road. I think this is probably for real, but they if go to a 6-day rotation it may well not happen again.

----

Which are his Worst and Best situations.

Annihilation. .495 on 7/19, .367 on 8/12, .514 on 9/20 = .445. This suggests that he's still a bit sleep-shy for day games, but it's not an issue unless he's also lacking rest both ways. Again, they should keep him off short rest as much as possible.

    

2) Immediate Upside ...

It's just 8 games, with 205 PA.

It's .280, which is an average ace.

Obviously he's not that good (wait for it ... wait for it ... yet). Two ways of looking at this:

Given that he had splits that were both good and bad, it makes good sense to regress the .280 to the mean, and that's .295 (by my metric), which is the best #2 in the game. That's credible for his 8 ideal games, I think.

And obviously, if you took the best split from every pitcher who has them, .280 is likely not extra-special. But that just stresses the importance of getting your pitchers into the situations where they're at their best.

The first pair is before the ASB and the second one is after.

Split  G   xwOBA G   xwOBA
Night-5  5   .265 3   .303
Day-5  4   .341 3   .290
Night-6+  3   .359 3   .268
Night-4  2   .312 2   .300
Day-4  0 3   .445
He was .312 in the first half, and .289 in the second excluding the death combo, which he never had to face in the first half. The .265 in his most common situation would have ranked him 3rd in MLB if sustained for a whole season, so that was likely due for regression. The resulting .303 (still a bit below average #2) was more than offset by improvements in the others. Meanwhile, the 3 ultra-tough games boosted his overall number to .317. So there's .028 of illusory struggles.

His adjustment for opponent hitting went from adding .004 (meaning the opposition was .004 below average) to subtracting it, so that adds another .008.

But both of those pale in comparison to his luck and defensive support on balls in play. In the first half his wOBA was .295, so that's .017 of good luck / defense. In the second half it was .373, an amazing .056 of unearned badness. That's another .073 of mirage decline (relative to the first half), for a total of .109!  

And yes, wOBA - xwOBA is almost entirely luck and defense. If anyone disagrees, we can talk.

Of the 72 pitchers who had 100+ PA end with a changeup, Bello ranked 8th with a .210 xwOBA -- but none of the 7 guys with more success approached his 626 pitches and 172 PA. (Next best was Nick Martinez with 504 /141.) It's well-established that the more you throw a good pitch, the less effective it becomes, so it's likelier than not that Bello had the best change in MLB.

A good thing, because that was his only pitch that was above average. That he still finished with a .315 -- a somewhat above-average #3 -- was impressive for a rookie with one reliable pitch.

What, you say? Even his sinker --a glorious pitch to watch, when he has it working -- was subpar? Yes : .343.

But I gave it away, didn't I? When he had it working.

In his 11 starts from 5/10 to 7/19, he had a .280 in 98 PA. But in 5 earlier and 12 later starts, he had a .408 and .371 (46 and 120 PA).

(At present I have no plans to break down all of his pitches. Ha!)

Consistent execution of your pitches, for both command and movement ... that is very big. It sure looks like Bello has resolved his day / night problem, so this is his final frontier. And man, if I were Andrew Bailey I'd be over the moon with excitement at helping a talent this large achieved that step.