[Editor's Note: Several BtBS writers and analysts attended the 2015 SABR Analytics conference in Scottsdale last weekend. By all accounts, it was a great event ... basically a sabermetrician's dream. In addition to two of our writers -- Stephen Loftus and Daniel Meyer -- presenting at the event, there were plenty of other panels and presentations, the Diamond Dollars Case Competition, awards for baseball writing, and more. As part of our coverage of the event, we've hosted a version of Stephen's presentation, and we'll be running a few Diamond Dollars cases. But for today, we'd like to present recaps of several presentations, from BtBS writer -- and recent Seattle Mariners hire -- Dan Meyer and Stephen Loftus!]
RP1: Vince Gennaro, "What’s Different About Postseason Baseball?"
SABR President Vince Gennaro kicked off the research presentation portion of the conference with a great session examining what types of hitters thrive in the postseason. We all know it would be too easy to just look at postseason splits -- but before answering Vince addressed what is different about the postseason.
As it turns out, it all comes down to pitching. In the playoffs, Buster Posey can still only hit once in every run through the Giants' lineup, but Madison Bumgarner certainly doesn’t have to wait for every fifth day to make a start. Vince found that the best pitchers in baseball throw a disproportionate number of innings in the playoffs. We need not look further than our example of the Giants, during the regular season Madison Bumgarner threw less than 20% of all Giants innings, yet in the playoffs he threw 33% of all Giants innings! That trend extends beyond this one example. Overall, in the playoffs we can expect a team to lean more heavily on their top three or four starters and best couple relievers. So where does Vince go from here? Now that he identified what is different, he looked at the hitters who can best leverage this imbalance.
If we can expect hitters to face higher quality pitching in the playoffs, then we can try to identify hitters whose ability is less reliant on the quality of pitcher. These are the guys who would potentially be more valuable in the playoffs. While controlling for platoons, Vince found that there is indeed wide variation among players -- some guys feast off of the league's bottom-dwellers and struggle off the premium arms, while others remain pretty consistent despite pitcher skill. Vince cited former all-star Derek Jeter as an example of a guy who is especially resistant to quality of pitcher -- this allowed Jeter to pay less of a price for facing higher talent levels come October. Josh Hamilton, on the other hand -- a guy not known for his playoff heroics -- feasts off of the league's worst arms and is relatively worse against better pitchers.
I question whether or not it would make sense to target these guys for roster construction, but all else being equal it is a really neat find. Vince’s presentation was an excellent example of how one can apply some saber concepts and creativity to a burning baseball question: is there such thing as a good, dare I say clutch, playoff hitter? Yes! -- Dan Meyer
RP2: Jason Wilson and Jarvis Greiner, "A 2014 MLB Case Study in Quantitative Pitch Mapping Using QOP (Quality of Pitch) and GI (Greiner Index) with PITCHf/x data"
This presentation was met with much fanfare prior to the conference, to the point of being written up by Yahoo's Jeff Passan the day before the presentation. So there were many expectations going into the presentation, and I was curious to see what would come of it.
The research presented by Wilson and Greiner hoped to provide teams and public sabermetricians with an objective pitch evaluation scale. To that end, pitches were evaluated on a scale of [0,100] by a pitching coach, and then these values were put into regression in order to identify the important variables to determining a quality pitch. The output of this model was used to create the Greiner Index (a measure of quality of pitch), and this multiplied by the MPH of the pitch gave you the overall quality of pitch score (QOP, already trademarked by the group).
The researchers then gave individual examples of pitches and calculated their QOP score as a tutorial. Further, they looked at Clayton Kershaw's 2014 postseason to see if he was throwing less quality pitches than in the regular season, but this isn't what I want to discuss about the presentation.
While the quality of pitch idea is a very interesting one, I find the execution of their idea a little wanting. Their goal of determining an objective quality of pitch metric seems to me to be slightly undermined by the use of one single pitching coach to determine the coefficients that go into the Greiner Index. Further, I would have been highly interested in seeing more information in the determination of coefficients for the Greiner Index.
However, the biggest question for me is the stripping away of context that this metric entails. Greiner and Wilson stated explicitly that they wanted to remove all camouflage from these determinations of pitch quality. However, to me, camouflage is what pitching is about. Warren Spahn stated that "Hitting is timing. Pitching is destroying timing." Pitch sequencing may hold major key aspects to understanding pitching, and this form of QOP strips this away entirely. Yes, a 92-MPH Clayton Kershaw that's right down the middle may put up a below-average QOP of 2.96, but if it's preceded by one of Kershaw's devastating curveballs, it may not matter.
The applications that are proposed (using QOP to determine injury before it happens, evaluating hitters, etc.), may show promise, but it's entirely dependent on the effectiveness of the metric. Whether or not it sticks will have to be seen, but for now, it does remain an interesting idea. For more discussion of QOP, head on over to Tom Tango's blog where noted names get in on the discussion. -- Stephen Loftus
John Thorn, Pete Palmer, John Dewan, Dick Cramer, John Walsh, "Origins of Sabermetrics Panel"
While not a research presentation, I would like to take a few minutes to talk about this panel, which really brought together a majority of the Mount Rushmore of sabermetric analysis. Personally, I know I wouldn't be here doing sabermetric writing without the 2nd edition of Total Baseball, which was edited by Thorn and Palmer. This book (which I got at a used bookstore for $5), introduced me to linear weights, runs per win, and many other statistics that introduced me to the thought the baseball statistics extended beyond home runs and batting average.
I won't spend too much time talking about the panel, as Brian Kenny already did an excellent job covering the topic. But I do want to key in on one statement that John Thorn closed his comments with.
"I think statistical analysis and sabermetrics, as has been attested to by others at this conference, is a frame of mind. It's a habit. It is not a toolbox. My advice to anyone wanting to break into baseball from an analytical point of view would be: Do not limit your studies to baseball alone. … Learn everything you can about the wide world and think about how baseball might connect to it, or how it might connect to baseball. I think if there is a royal road to success in baseball, it's in knowing a thing or two about the wide world."
This. Exactly this. If there's anything I would say to an aspiring sabermetrician, it's this. I draw inspiration for my baseball work through the work I do with biologists in academia. The writers here at Beyond the Box Score draw from our various careers in our work. Thorn is MLB's historian, Palmer was a radar systems engineer, and Dewan was an actuary. If there's any group of individuals that exemplify this thought of learning from the world, it's this illustrious panel. When SABR releases the video of this panel, I would highly encourage everyone to take time to watch it. -- Stephen Loftus
RP8: Ben Jedlovec, "Trajectory-Based Hitting and Pitching Statistics"
President of Baseball Info Solutions Ben Jedlovec gave one of my favorite presentations of the weekend. Ben gave us in the public research sphere a peek behind the curtain as to what is possible with true batted-ball tracking data. In his presentation Ben used BIS’s batted-ball data in order to objectively assign a batted-ball result without relying on what actually happened on the hit. This lets us look at David Ortiz’s 2014 season and say okay, he hit 35 home runs, but he should have hit 40.
This approach allowed Ben to do some really cool things. For one defense independent batting (DIBS) has much stronger predictive capabilities than linear weights models. DIBS has a year to year correlation of about .65 compared to .52 for linear weights. Within this framework, one can rely on a smaller sample size to detect changes in true talent. Ben showed off the potential predictive capabilities by looking at Jonathan Lucroy’s DIBS numbers from 2013 and 2014. In 2013 Lucroy was worth about 11 runs above average based on his actual hitting results using linear weights. However if one looked to his DIBS numbers we would have seen he should have hit his way to around 21 runs above average. Of course in 2014 Lucroy’s talents manifested themselves in his actual results, and he was worth 21 runs above average based on linear weights. This was even more closely in line with his DIBS runs above average from that year: 24.
Ben’s introduction to DIBS was quite illuminating and made me all the more excited for the day HITf/x goes public. -- Dan Meyer
* * *
Daniel Meyer is a Featured Writer for Beyond the Box Score. Stephen Loftus is an Editor for Beyond the Box Score.