Lester, Wainwright, and World Series Repeat Starters

One of the points we’ve been hammering home really all postseason long is that starting pitchers tend to perform worse and worse as a game goes on. It’s far from a dramatic effect — a pitcher the third time around shouldn’t be expected to get completely and utterly bombed — but an effect is there, as pitchers make subsequent trips through the order. Starters become less effective, and so it becomes more and more important to put your trust in a fresh bullpen. It stands to reason part of the effect is pitchers getting a little fatigued. It stands to reason another part of the effect is hitters getting multiple looks at a guy. Hitters communicating with one another about what they’ve seen from a guy. The second and third time, the average hitter might be more prepared to punish the guy on the mound.

Let’s assume that it’s true that there’s a benefit to having already seen a guy once in a game. Maybe it’s not, but that would be quite the discovery. It doesn’t take much of a leap, then, to suggest there might be a benefit to having already seen a guy start in a series. Face a pitcher in one plate appearance, and you might be more prepared in the second plate appearance. Face a pitcher in three plate appearances, and you might be more prepared in the fourth through sixth plate appearances. It sounds sensible, meaning this works as a starting point. It doesn’t yet work as a conclusion.

I got curious about this. In Monday night’s Game 5 of the World Series, the Red Sox will start Jon Lester, and the Cardinals will start Adam Wainwright. It’s the same matchup we got in Game 1, and so I wonder if either team will benefit from the relative familiarity. Lester dominated with his cutter, and maybe now the Cardinals will have a better idea what to do with it. Wainwright didn’t dominate with anything, but now on top of that Boston’s had a bunch of looks. Acknowledging right here that no broader investigation necessarily means anything with regard to this one game in particular, I wanted to learn more about the World Series history of repeat starters.

I went back to 1969, which is my usual starting point given the changes to the mound and all. It’s somewhat arbitrary but it wasn’t selected for purposes of biasing the results. I also looked only at World Series, because I figured teams will be the least experienced against World Series opponents, even given free agency and trades and interleague play. I looked for pitchers who started at least two times in the same World Series. What this gave me was a sample of 152 starting pitchers, from Tom Seaver to Matt Harrison.

Below, a table, showing results from start no. 1, and results from start no. 2.

Start IP BF RA/9 ERA BB% K% HR% BABIP
1 6.3 26.5 4.08 3.63 8.3% 18% 2.5% 0.271
2 6.2 25.8 3.70 3.35 7.9% 16% 2.1% 0.267

In all, we have pools of 152 starters and 900+ innings. I’d rather have samples of 9,000+ innings, but the samples here aren’t too small to learn from. On average, the second start has been a tiny bit shorter than the first start, by a fraction of one plate appearance. Strikeouts have been down in the second start, not insubstantially. But, walks have been down a little, too. As have home runs and hits on balls in play. The most important column is the RA/9 column, and that reveals a pretty big difference. Rather than hitters picking up an advantage, pitchers have been considerably more successful at preventing runs the second time they’ve faced a lineup in the World Series. That’s a drop in RA/9 of 9%. I don’t think that can be completely explained by managers going to the bullpen just a little earlier.

Something that could be a factor is the weather. Probably, it’s been a little colder for the average second start than the first, and colder conditions suppress offense. We’re talking about gaps of four or five or six days in October, and that could help explain the reduced hits and homers. But the general, overall point is this: if hitters get an advantage from having faced a starter one time already, we don’t seem to observe that. What we observe suggests the opposite, and that probably holds true even after you adjust for the weather and the slightly more aggressive bullpen usage.

You might be wondering about third starts within World Series. Unfortunately there’s not much to look at there. We have 31 runs allowed in 62.3 innings, but those are just 62.3 innings, mostly coming on short rest, which is a contributing factor. Most first and second starts have come on normal rest, allowing them to be more easily compared. That’s where we see run production drop off.

Why might things work this way? Well, off the top of my head, while hitters will adjust to the patterns they’ve seen, pitchers will adjust their patterns so as not to pitch identically. Pitchers might also have different pitches working the second time, and I think an important principle is that hitting is reactionary, while pitchers are in control. Hitters have to respond to what pitchers are doing, and if a pitcher senses that hitters are getting the best of him, the pitcher and catcher will change the approach. And as much as hitters might learn about pitchers from a start, it also works the other way around. Weaknesses might be revealed in start no. 1. Pitchers might anticipate how an offense will respond the next time.

What this means for Adam Wainwright and Jon Lester is: very little. They’re their own pitchers, facing unique lineups, and what happens will happen. A lot about the game has changed since 1969, so maybe the data shown here is of little value going forward. I don’t know. What I do know the numbers demonstrate is that there hasn’t been a historical advantage for hitters facing a starter in the World Series for a second time. Advantages within a game, absolutely. Advantages between games, not so much. Players are always making adjustments, and it’s hard to adjust to a player who’s adjusting.



Print This Post



Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.


Sort by:   newest | oldest | most voted
Bip
Member
Member
Bip
2 years 10 months ago

Knowledge can only get you so far. You may know what a pitcher has got, but you don’t know what he’s going to throw next. Some guys have been in the league for more than a decade, and they still can make seasoned hitters look foolish. Hitting is hard.

ian
Guest
ian
2 years 10 months ago

With advanced scouting, I’m not sure if getting multiple looks at a guy is as big a deal anymore. A guy starting 2 games in the World Series is probably among the best starters in the league, and whether that guy has his best stuff going that day is probably more important than whether or not you’ve seen him before.

pft
Guest
pft
2 years 10 months ago

Scouting reports and video don’t replace actually facing the pitcher, not by a long shot.

channelclemente
Guest
channelclemente
2 years 10 months ago

Unless, like Lincecum, you become predictable.

Cus
Guest
Cus
2 years 10 months ago

If by predictable, you mean predictably can’t get ahead in the count.

Hamilton Marx
Guest
Hamilton Marx
2 years 10 months ago

Good stuff as always. Makes sense given the fact that the hitters see about 6 different pitchers in between. I’m guessing there is no effect when a RP faces the same lineup multiple times in a series. Seems like the same concept.

pft
Guest
pft
2 years 10 months ago

Don’t want to get in the way of a good belief, but RP’ers usually only face 3-5 batters per appearance, not the entire lineup. But they may face the same hitter as many as 3-4 times in a 7 game series. Probably not as significant as facing the same hitter 3-4 times in a game, but in the regular season a RP’er only faces the same hitter 3-4 times in a year.

I think the biggest obstacle for RP’ers besides the hitter seeing him more often is that in the playoffs hitters may spend more time watching video and reviewing scouting reports of relievers than in the regular season. Especially the top relievers. More off days and the importance of the games are the reason for that.

An argument against this theory is Rivera was more dominant in the post season than in the regular season. But then he may just be a unique pitcher and his cutter may be more effective in October weather. OTOH, Red Sox hitters saw more of Rivera than pretty much any team and had the most success against him.

Hamilton Marx
Guest
Hamilton Marx
2 years 9 months ago

Are you saying there is a negative effect with RP facing the same hitter more than once in a series? I was saying there is no effect, but haven’t looked at the numbers.

RC
Guest
RC
2 years 10 months ago

It’s interesting, but I’m curious if its maybe just a result of managers having a quicker hook later in the series as we start getting toward elimination games.

Proudhon
Guest
Proudhon
2 years 10 months ago

I know the sample is small in any case, but I’m thinking that in most of the cases you looked at, one of the starts was at home and one on the road. Did you make any adjustment for home-road splits?

Cus
Guest
Cus
2 years 10 months ago

Yeah, this?

Bip
Member
Member
Bip
2 years 10 months ago

One would expect that most of these data points would come in pairs. In other words, if a starter in this set made a home start followed by a road start, you would expect in most cases there is a starter opposite him who made a road start followed by a home start. I would be very surprised the group of first starts had a different home/road ratio than the second group.

pft
Guest
pft
2 years 10 months ago

Lesters H-A splits may be a factor (4.21 ERA -H, 3.09 ERA-R). However, there were 2 Lesters this year, the first one through August 8th with a 4.37 ERA (much like the 2012 and Sep 2011 Lester), and the one after with a sub 2 ERA (include post season).

Maybe he started doctoring the ball, woke up after the Peavy trade (realizing his option may not be picked up), or started drinking Papis shakes (LOL). Dunno. In any event, the season H-A splits may not be significant since he has pitched well on the road since his turnaround.

Bip
Member
Member
Bip
2 years 10 months ago

A whole career’s worth of data is rarely made insignificant by a few months of data.

wee162
Guest
wee162
2 years 10 months ago

I would agree there were two Lesters, but it wasn’t split over that time period really. Lester was good at the start of the season. Then fell to bits in late May till early July. They skipped him for a start during the all star break then straight after that he was back to being good again. He got knocked around in one start against the Diamondbacks in August but that was pretty much it for his bad starts for the rest of the season.

DavidKB
Guest
DavidKB
2 years 10 months ago

I’m interested by this question of whether a starting pitcher’s decline is due to fatigue or batter familiarity. One thought would be to look at what pitchers allow against pinch hitters, though you could argue that pinch hitters get the scoop from their teammates in the dugout so it isn’t like they’re stepping in with zero information. You would also have some sample size problems even in the regular season.

I wonder if there is an effective way to isolate these two variables. It’s a nasty question for statistics folks, but the payoff is considerable.

Lex Logan
Guest
Lex Logan
2 years 10 months ago

I thought I read recently that the decline in pitching effectiveness for starters correlated with decline in velocity. Some pitchers seem to throw harder late in the game; either they may be better candidates for pitching a complete game, or perhaps they lose command, missing more often in the middle of the plate. Either way, it would have nothing to do with batter familiarity, and would be consistent with the world series repeat start data.

Brad Johnson
Member
Member
2 years 10 months ago

As for reasons why hitters do better against pitchers in each consecutive PA, I’m thinking that in-game fatigue probably has a larger effect than times seen. Too bad we can’t talk a team into letting a pitcher throw 5 simulated innings before starting a game (and then doing it enough times so we have a decent sample size).

I know that we’ve tried to measure this using velocity as a proxy, but I assume some pitchers see their command and control suffer as a result of fatigue while still maintaining velocity. As we know, even a very subtle change in location can be the difference between a pop out and a home run.

Adam Guetz
Guest
Adam Guetz
2 years 9 months ago

A basic statistical test, assuming runs are generated follow a Poisson distribution, gives a p-value of .16 (not significant). There is not necessarily anything to explain here.

R code:

in.1 = 6.3*152
in.2 = 6.2*152
runs.1 = round(4.08*in.1/9)
runs.2 = round(3.70*in.2/9)
poisson.test(c(runs.1,runs.2), c(in.1,in.2))

wpDiscuz