One of the most commonly asked questions when I venture out into the public at various fantasy baseball events such as Ron Shandler’s First Pitch Forums, Tout Wars, the NFBC and the FSTA Conferences is "What will the next big strategy be?", or if "There ever will be any more advances in strategy? "My answer is most strategy comes from gaining an edge by understanding the player pool better than your opponents. We are presently in between phases. Companies like Baseball Info Solutions revolutionized data collection (specifically batted ball data) and a bunch of really smart people then worked with the data to a point where we are all sick and tired of hearing a player was lucky or unlucky because his BABIP or HR/FB is out of whack due to as small sample size. I don’t mean to belittle this sort of analysis, heck, include me with those that discussed it ad nauseum. But I think we are ready for the next wave of data and the ensuing analysis (I’ll take care of the fantasy application, thank you). This is coming in the form of individual pitch data, improved measuring of defense and how it relates to pitching along with further enhancements to batted ball data. It is one thing to take the next step with respect to fine tuning expected player performance but another to know what to do with it in terms of game theory. So the answer to the question is “yes, new strategies will evolve, but not until we take player analysis to the next level.” And trust me; there are a whole lot of really smart people out there endeavoring this right now.
What I thought I would do for the next few weeks is survey the landscape and share some of the current data I find to be the most useful to the fantasy game. Don’t worry, we won’t forget about BABIP and all that good stuff, but there is some fairly new data available to the mainstream that can be used to help improve our fantasy game playing.
After all that build-up, today we are going to start with a rather simple concept that does not even involve anything new. That said, the topic is rather important and something I admit I have been a bit remiss on discussing over the years.
Everyone is familiar with the metrics K/9 and BB/9, commonly known as strikeout, and walk rates. An alternative, and to some a better means of looking at whiffs and walks is K/BF and BB/BF, where BF is the numbers of batters faced, though some will use plate appearances since it is more readily available. These stats are often referred to as K% and BB%.
The difference may be slight, but it is relevant. Pitchers that face more hitters have more opportunities to rack up a strikeout, so in a vacuum, a higher K% is more impressive than a higher K/9. That said, most of the time pitcher’s are ranked similarly with respect to K% and K/9. Only those that give up or prevent an inordinate number of runners have their K/9 and K% out of sync.
There are two applications of this principle that immediately come to mind – pre-season player projection and in-season evaluation of performance. We’ll spend a little bandwidth on each.
While everyone does projections slightly differently, I like to work off of innings pitched, using historical K/9 to determine the number of strikeouts then using batted ball data to determine what happens the rest of the time. If a pitcher’s BABIP (told ya we wouldn’t forget about it) was exceedingly high or low for any of the seasons used to gauge future K/9, that computation could be a little off. It is going to take a little bit of work to the projection engine, but I feel adjusting to work off of K/BF (or K/PA) will render a better foundation for the pitcher’s expected performance. A like argument can be made for using BB/BF in lieu of BB/9 and HR/BF and not HR/9. The other repercussion of this is many expected ERA formulas utilize K/9, BB/9 and HR/9 as inputs. I am going to have to look into the manner I determine ERA and tweak it to use the batters faced metrics as the data feed.
In season, using batters faced may be a better indicator of how a hurler is performing relative to his history, especially in small samples where base runners allowed, either good or bad have yet to normalize. Actually, as I further think about what I plan on writing as a write it, something else I have long considered important is about to come into play (more on that in a moment).
The idea behind the difference in K/9 and K% with a low (or high) BABIP is as follows: Let’s say a pitcher normally sports a K/9 of 7.0 with a BABIP of .300, basically a league average guy. If his BABIP is low, he is getting more outs than normal in places he normally would not get outs. This should result in a K/9 LOWER than his usual standards. Conversely, if the BABIP is high, then the pitcher needs to find a way to make up for those lost outs, with one opportunity being the chance to face, and therefore punch out the next batter.
Here’s where the proverbial wrench gets tossed into the analysis and actually speaks towards the areas where data collection and analysis can be improved. I am convinced that every pitcher is actually two pitchers – windup guy and stretch guy. I believe most hurlers will exhibit different skills working from the windup versus the stretch. While this is just a hypothesis, it is my contention that one reason some pitchers always seem to over or under pitch their peripherals is the difference between windup and stretch guy. Those more effective from the stretch are more likely to outperform their peripherals; at least that is what my intuition tells me. I have something beyond anecdotal, but not quite conclusive data to demonstrate this as unfortunately, windup and stretch data is not specifically archived. That said, at least for starting pitchers, some assumptions can be made. With no one on, the pitcher works out of the windup. With runners on base but not on third, they work from the stretch. When there are runners on third things get hairy as some starters will go from the windup with a man on third, second and third or with the bases juiced. Using some common sense to delineate windup from stretch, it can be shown that globally, pitchers have superior skills when working from the windup. These include a higher strikeout rate with lower walk and home run rates. Additionally, their BABIP is lower from the windup, but that’s a rant for another day.
Tying this together, it was surmised above that a low BABIP may lead to a lower than usual K/9 since more outs are recorded on balls in play, requiring fewer strikeouts. If my windup versus stretch theory is true, then a lower BABIP results in more work from the windup and MORE strikeouts, perhaps negating the loss from a lower BABIP.
The take home lesson is there is more to K/9 than meets the eye, but there are could be more to K% as well. As with anything, no single metric tells the whole story. But there are going to be cases where K% is more representative than K/9 and be it in site columns, Platinum content or on the message forums, whenever I inclined to discuss K/9, or even BB/9 and HR/9, I can going to take a step back and consider if K%, BB% or HR% may be a better indicator.