Evaluating defense using HITf/x

Colin - I am glad you are going to make it to the Summit this year.  I look forward to meeting you.  Last year I worked with the Sportvision raw footage for 2 games in an attempt to manually calculate the Hit f/x parameters and also to try and estimate spin and hit ball landing location from those parameters.  I was successful in showing that Hit f/x parameters could be calculated from the existing footage, a result that I presented at last years Summit.  I was unsuccessful at the latter task, my conclusion being that there was too much variation in the spin characteristics and not enough input information to be able to make an accurate estimate.  Dr. Nathan and Sportvision have come to the same conclusion.  But everyone is still trying so there may be some further news this weekend. 

You are on the right track with your effort to establish the limits that different spin rates might reasonably put on a hit ball.  However, last year when I calculated spin rates on hit balls by matching their input parameters and the best guess of there landing location using an average of MLB Gameday, STATS, my observations from video, and Greg Rybarcyzk’s observations from video, I found spin rates that were over 3000 RPM, which extends the possible landing points of your graph considerably.  I concluded that until we get full path tracking of hit balls from Sportvision’s proposed future Fielding f/x that we would not be able to pinpoint a landing location from the Hit f/x parameters that is any more accurate than that given by Gameday’s hit locations. 

That does not mean that the Hit f/x parameters cannot be used to improve existing fielding metrics.  My talk for the Summit will be specifically on that subject and I will publish a follow up Skill Based Fielding Metric to my Skill Based Batting and Pitching Metrics that will include the improvements that I will discuss on Saturday.

Just curious, why can’t you just use a metric of where the ball lands and how long it takes to get there?  Wouldn’t that take all thr guesswork out of the whole classification of fly ball, liners, etc? 

It seems to me that with the advanced tracking systems we have we can plot every point on a given ballpark, and also calculate pretty accurately how long a ball takes to land on that area, and from that we should be able to determine how often such a ball becomes a single, double, out, etc.  From there it can easily be deterimed the average value of a ball hit with such a placing/timing ratio and therefore the expected value to the hitter/pitcher/fielder depending on the actual result.

Why worry about spin?  Why not just break it down to the basics and go from there?

I’ll have quite a bit to say about spin at the summit on Saturday.

DaninPhilly - No one is currently mapping where a ball lands. MGL has a project to determine hang time but the information will not be publicly available. Eventually we will have access to both pieces of information, perhaps in the next year or two, and, as you suggest, that will be all we need to know.  What people are talking about here is how to do the best we can with what information we have now.

Peter, for some reason, I remember read an article lately which I thought had said that obsticle had been overcome and some company was actually tracking that information.  If I come accross it, I’ll post it here.

I have suspected that the Rays (and possibly some other teams), have used vectorizing of the flight of the ball, which the game tracking information provides, and not hit charts to realign their outfield defenses. By measuring both the eventually landing site of a hit and time of flight, from multiple years of data at the Trop, they have calculated the position for the outfielders which maximizes the probability of catching any flyball weighted on how much value the hit would have should it drop in. This has allowed them to bring their outfielders in to catch the more frequently occurring, quickly falling single and thus spend less time defending against the less frequently occurring long fly balls. [It also helps that air resistance is quite significant at the sea-level sitting Trop.] While this is fine-and-dandy, the down side is what happened earlier in the year where some very hard line drives where hit over Upton’s head during key moments near the end of the game. Of course the team was crucified in the papers because “that’s not the we do things in baseball.” Kind of like not having a closer or why NFL coaches are too conservative on 4th down.

Based on watching a large number of Rays games,  nearly every game there is at least one very long fly ball.  It looks like it is significantly over Upton’s head and will fall him, but the time of flight is long enough he tracks it down. It helps to have a fast outfielder.

I’m pleased to see baseball start using physics solutions to an applied physics problem.