Playoff Baseball is Not All About the Pitching
One of the reasons I really like having InstaGraphs is that it gives us a more natural place to link out to other good stuff around the web. It’s more natural to have a link in a blog designed for several hundred word posts than it is to have one where we’ve trained you to expect 1,000 words. And often times, the article I’m linking to doesn’t really need supporting commentary.
Like this one, for instance. Ben Lindbergh tackles the question of the advantage of having elite starting pitching in the postseason. These kinds of studies have been done before, and the conclusion isn’t revolutionary, but there are still a lot of people who think that you can dramatically skew the odds in October by loading up on high quality pitching, but the data keeps pushing back against that idea.
The home run paragraph of Ben’s piece:
So why doesn’t the quality of a team’s top three starters or its ace register as significant? For one thing, the differences between teams are compressed in the playoffs, relative to the regular season: Teams with terrible staffs don’t make it to October, so the gulf between the best- and worst-pitching playoff teams isn’t as stark as we’re used to seeing during the season’s first six months. Perhaps more importantly, there’s more than one way to win baseball games, and even under an expanded playoff format, teams don’t get to October without doing something well. A team with an inferior pitching staff often makes up for its weakness on the mound by being better on offense.
Read the whole thing, and keep the takeaway in mind, because you’re going to be inundated with “pitching wins championships” cliches for the next few months, even though the reality is that good baseball teams win championships.
Dave is the Managing Editor of FanGraphs.
As a Phillies fan, this is knowledge that I’m painfully aware of.
And as a Braves fan… well, same.
If pitching on short rest leads to poorer results, on average, then maybe the idea of going 4 deep with aces is to attain the opposite – better than normal results by way of extra rest?
The problem I feel is the piece explains why there is no data backing up the myth, however the piece then goes on with an aura of “this myth is false” which is just as wrongheaded because they have no proof showing that it certainly is wrong.
The piece doesn’t say there is no data backing up the myth. It says the data has found no correlation between good starting pitchers and/or aces doing better in the postseason when controlling for that team’s regular season record. Data was found. It did not back up the myth. It, in fact, disproved the myth.
I mean really, who needs pitchers anyway with Big Donkey on the mound?
I keep on seeing how BP retired the “Secret Sauce” but the reason it was not predictive was because the “Secret Sauce” was being used incorrectly, not at all what the book’s study found, I feel.
The Secret Sauce (and to be fair to BP, it was the study’s author who made the mistake) was never meant to rank playoff teams against each other in the playoffs, though that would be nice. Re-read the chapter, nothing about ranking the teams against each other.
Where the ranking should have been happening is inserting each 20XX team into the full database of teams, and seeing where each team ranked within the history of all teams. The book looked at the top 10 teams in ranking and found that it was a powerful formula for the top ranked teams. I would bet most teams today are falling into the 100’s in terms of rank, meaning that these teams, playing each other, it’s random who wins. It’s like watching two .500 teams playing each other, one will eventually have to win that game.
I tried to research the numbers for each of the Top 10 teams and compared the 2010 Giants to them, and (I’ll admit it was hard to decipher all of BP’s numbers) it appears that this team would have ranked up there all-time. So I still believe in the formula. I wish BP would just make the full database of data available on Google Drive for everyone to play with, especially since they have officially given up on it. Then we can see for ourselves how much more likely it is that top ranked teams win in the playoffs.
What I expect to see is a curve of World Series probabilities based on relative ranking. The Top 10 obviously would be in the 100% probability of making the playoffs, and pretty close to 100% winning. Then the probabilities will fall, which we could group into ranges.
Then the way the Secret Sauce article should be written is discussing where each team falls in the historical database and what the probabilities of the World Series is in that rank. So maybe one team might rank high in historical perspective and thus appear likely to make (or even win) the World Series. Or mostly, the teams are all mediocre in historical terms, and random luck will determine who wins the World Series.
Trying to find the “Secret Sauce” was too high a goal I think. I think we all know that there is just too much randomness involved for that to ever work 100% of the time (and as the saying goes, that’s why we play the games). But I think that this study’s result can at least give us a good indication of the likelihood potential for each team, and that would still be valuable, even if it is not definitive.
Here’s where I think everyone is missing a nice Secret Sauce. People focus too much on overall WAR, positing that everyday players are more valuable than pitchers. What they forget is that in the playoffs, there are only 5 to 7 games. And pitcher’s WAR/start is much much higher than that of regular players, making their influence on each playoff game that much greater, on a per game basis. For example, in 2013, Puig averaged 0.050 WAR/game (roughly, 4.9 divided by 96 starts is 0.051 but he played in another 8 games) but Kershaw averaged 0.255 WAR/game, which I would bet is greater than summing up the rest of LA’s lineup’s WAR/game. A pitcher’s influence in any one game is huge compared to any one offensive player.
They are also forgetting a stat from yesteryear that I think is a key to the playoffs: quality starts. QS is, rightfully so, not all that good given what we know today about baseball, but Baseball Forecaster/HQ brought the QS into the sabermetric lexicon with their PQS or Pure Quality Start, creating a saber-based indicator of a quality start. 6+ IP. IP>=H. K>IP-2. K>2*BB. HR<=1. Each true statement earns a point, sum up the points to get PQS from 0 to 5, auto zero if under 5 IP. Pitcher has an equivalent to a quality start if his game PQS is 4 or 5 (called Dominant or DOM start), and I've found that good pitchers have 40-49% DOM, great pitchers 50-69% DOM, and elite 70%+.
Giants in 2010:
Lincecum 64% DOM (flip of coin)
Cain 67% DOM (should win)
Sanchez 48% DOM (should lose)
Bumgarner 50% DOM (should win)
Lee 64%
Wilson 45%
Lewis 66%
Hunter 18%
We know without data that there are pitchers who regularly pitches QS. We know without data that a team is more likely to win a game when they get a QS from their starter. Just to be sure, I did the PQS data for the past 4-5 playoff runs and, yeah, QS generally win against non-QS around 80% of the time (obviously, QS v. QS is .500). And looking at the pitcher historical records, certain pitchers tend to get a larger percentage of QS than others, the best pitchers.
So if you want to maximize your chances in the playoffs, you want to stock your 4-man rotation with pitchers who have very high percentage of QS. It's no guarantee, as the Phillies learned in 2011, but if you got a bunch of starters with 60-80% QS going up against teams with 40-50% QS, you have certainly maximized your chances.
In addition, a lesser known fact of BP's study on Billy Beane's playoff statement that begat the Secret Sauce is that VORP for Top 3 Starters (as well as for all starters) had higher correlations, not quite the same as for Closer WRXL but much higher than that for Pitcher's strikeout rate or FRAA (these are the three metrics chosen for their secret sauce). Among the Defensive metrics (Pitching and Fielding), VORP for all had the third highest correlation, VORP for 3 had the fourth highest (per Figure 9-3.1). Not sure why the authors never followed up on that one, but I feel that this would have made their metrics that much stronger an indicator (and again the Secret Sauce is an indicator, not a definitive conclusion, much like team record).
FYI, Beane's recent A's teams have not only not led in strikeouts but they have been, even worse, below league average. And we know how he feels about closers, he picked up a nice one in Johnson (people justified the cost in that it reduced the arb pay for young players, but Doolittle is going to get a huge payday now with Johnson doing so poorly that he got DFAed already).
Forgot to look at the closer. I think it is a mix of closer and non games, but Jansen averaged 0.035 WAR/game in 2013, which is nearly as much as Puig contributed in a game on average, and I would bet if you add in one more reliever (Paco at 0.020 WAR/game), the two relievers contributes slightly more than Puig’s per game contribution. So relievers might not contribute as much per season as a hitter, but in terms of impact on a game, the closer and set-up guy contributes more than their top hitter.
Also, apologies, but all these data is bWAR not fWAR, but I would not think that there is a huge difference.
OK, here is the fWAR equivalents:
Puig: 0.038 fWAR/game
Kershaw: 0.200 fWAR/game (still roughly 5 times the WAR)
Jansen: 0.029 fWAR/game
Paco: 0.009 fWAR/game (together equals Puig)
Doolittle was signed to a 5 year, $10.5m contract before he got the closer job.
OK, so Billy got that one covered.
Still, he spent $10M for basically nothing in 2014, that takes a big chunk out of the money he saved with his young relievers vs. arbitration.
The thing that’s not being addressed is that in a 5 game series, your MVP hitter will get his usual percentage of AB/game, but your Cy Young pitcher will get a higher percentage of IP/game than during the regular season (assuming he pitches games 1 and 5).
This format allows for better pitchers to have more of an impact while better hitters do not.