Statcast in Ottoneu

Monday, Tom Tango took to twitter, releasing two graphs.

Explanation here: https://t.co/fSoKBGymlH pic.twitter.com/lXpHEkTy7s

— Tangotiger (@tangotiger) June 26, 2017

This might be interesting, especially at the high-end speeds. @pobguy pic.twitter.com/LJcmO7l7RQ

— Tangotiger (@tangotiger) June 27, 2017

Like many of us who benefit greatly from Tom’s work, these got me thinking (specifically the second graph). What we have here is wOBA on each event type plotted against each event’s exit velocity. Like we would expect, the harder you hit the ball, generally, the better outcomes you have. So my thought is, how do we apply this to Ottoneu? While hardly revolutionary, the idea of converting all metrics into Fangraphs Points is an adjustment all players must make.

Typically when we discuss Ottoneu fangraphs points equivalents, we frame these in terms of wOBA. For example, a roughly a .330 wOBA outfielder is a $6-$10 player. This is useful and practical. We know that as Ottoneu is a linear weights based scoring platform (for those playing in points leagues) that there is strong correlation between a player’s wOBA and his FGpts scored. However, how does this compare to statcast? First, let’s start with Ottoneu FGpts offensive scoring settings.

This gives us the raw point totals for each offensive event, however, it would be more useful for us to have this in a slightly different format. For example, when a player hits a home run, what is really happening is as follows:

1.) -1 points for an AB.

2.) +5.6 points for a H.

3.) +9.4 points for a HR.

That leaves us with 14 points in total. This same methodology can be applied to any at bat (not plate appearance). You lose your point for the PA, gain a point for the hit, and gain additional points for anything greater than a single. Not that complex, but to see the new “contact” totals this provides, we can use the following chart and examine the number of batted ball events for 2017 that fall into each bucket.

So for 2017 so far, we have the following. If contact is made, our expected points total is 1.53 points. That may seem a little low – no regulars average 1.53 points – but a large portion of this is that all not batted ball events are excluded, we are specifically looking at contact plays. You make contact, and 2/3 of the time you get out. 1/3 of the time you get a hit. What if we were to look at Ottoneu event scores by exit velocity? 

I have displayed this plot differently than Tom, but it tells a similar story. The y-axis shows the points associated with a specific batted ball event (14pts for a HR, 10.3 for a 3B, etc).  On the x-axis, we can see the range of exit velocities that correspond with each event. For example, the lowest EV home run in 2017 was 88.3 mph (Alex Bregman on May 24). Partitioning the data for lowest exit velocity on each type of ottoneu contact (HR, 3B, 2B, 1B, Out).

Average points per contact on all balls hit over 88.3 mph is 2.88 points per contact.

Average points per contact for all balls hit above 60.3 mph is 1.63

Average points per contact for all balls hit above 43.5 mph is 1.56

So what does this tell us? After after contact made dips below ~88 mph, the associated points per contact drop a bit. It may not appear too dramatic because the charts above display all batted balls hit over the show exit velocity. So perhaps, I should partition the graphs a little bit more, but that is an exercise for another post. I’ve also started adding launch angle and other metrics into this analysis to move beyond the above “most weakly hit batted ball in event x” type of analysis to use more of an average. That being said, I wanted to walk the community through some of the thought processes I was working through.

None of this is revolutionary stuff, but should serve as a brief overview on how different ottoneu points metrics compare with associated exit velocities. This definitely is not meant to be full proof, as I did not include launch angle or spray in any of the charts above. However, it is a start toward taking the statcast data that we have and beginning to see correlations to the Ottoneu points formats we love.