Filter posts by...
Hit_Tracking
Thursday, May 10, 2012
This writer is suggesting so:
And the scout remarked that Posey’s at-bats are dropping off in quality in the later innings. The numbers bear it out. In the first three innings, Posey is a .361 hitter. He has struck out five times in 36 at-bats. In the fourth inning through the ninth, Posey is a .250 hitter. He has struck out 20 times in 64 at-bats. And if you take just the seventh through the ninth innings, Posey has 13 strikeouts in 30 at-bats, while drawing just one walk.
Two points:
1. If the scout says what he did WITHOUT looking at the outcomes, that Posey’s swings “look” slow, etc, then that’s one thing. If the scout however CONSULTS the outcome numbers, then that’s a completely different thing.
2. 5/36 in one instance and then 13/30 in another? Meaningless by itself. Again, combined with the scout’s view (but without the scout seeing the outcomes), then it could be modestly valuable.
For example, let’s say that the scout says that Posey’s got great mechanics, and he expect him to not K alot, that he should K say 3/36 times, and he ends up K-ing 5/36. We might therefore conclude his true talent in early innings is say 3.2/36 K rate.
And if the scout says that Posey has a lousy swing in the late innings, that he looks like he’d K 10/30, and he ends up K-ing 13/30, then we might say his true talent K rate in late innings is 10.3/30 or something.
But, without the scout’s observation, we are forced to conclude that Posey’s true talent K rate in early and late innings is virtually identical.
Again, that’s IF (and a big IF) the scout’s observation is INDEPENDENT of the outcomes. Which is going to be pretty hard to do, since the scout will get biased by the fact that Posey struck out.
Which is why HITf/x will be so powerful, that we’ll be able to tell if Posey is swinging harder than normal, earlier/later than normal (if Sportvision can start the clock on the bat earlier on the swing), etc.
Tuesday, November 22, 2011
A followup to Mike’s terrific piece of the horizontal speed off the bat, this time, with the added focus of the vertical launch angle.
There’s actually plenty of info here, and I can’t comment properly yet until I do a second re-read. There’s also something that seems inconsistent, and I’m hoping Mike can set me straight on whether one of the charts published needs to be updated, or my reading skills need to be improved. I’m hoping it’s the latter.
Wednesday, November 16, 2011
Great stuff from Mike:
Batters have a good deal of correlation between halves of the sample, with a correlation coefficient of r=0.76 with an average of 201 batted balls in each half. That means that we would add 63 batted balls (or about one month’s worth) at league average to the observed average speed for each batter in order to estimate his true skill.
...
Pitchers have fairly good correlation between halves of the sample, though not as good as batters. The correlation coefficient is r=0.48 with an average of 251 batted balls in each half. That means that we would add 269 batted balls (or about three months’ worth for a starter) at league average to the observed average speed for each pitcher in order to estimate his true skill.
Just fantastic stuff, and I’m glad Mike did it, as well as showing the key points, which is the point at which r=.50.
***
I’m not really surprised by the results. The closer you get to someone’s base physical and mental skills, the less observations you need. This is why scouts are so important. And the F/X and Trackman systems are, at their heart, scouting tools.
What we’ve had until recently are outcomes, results, things like OBP and K/PA, etc. What drives OBP and the like are the players’ base skills AND luck. That’s why we infer a players’ base skills by stripping out as much luck as we can figure out. We do this through a Bayesian process (or its equivalent in regression toward the mean). We need a few hundred contacted balls for a hitter, and in the thousands for a pitcher, in order for us to be able to strip out that luck to infer the base skill.
Inside a player’s contacted ball skill is not only the horizontal speed off the bat, but placement as well.
Unseen in Mike’s data is what the horizontal speed off the bat really means. Let’s take a pitcher’s fastball speed. We presume that there’s a high degree of correlation in a pitcher’s fastball speed. I have no doubt that if you do a split-half correlation, you’ll get something ridiculous like r=.99 (really, it’s a question of how many nines) for pitchers who throw 1000 fastballs. So, we can ascertain a scouting observation: we can readily and easily ascertain a pitcher’s underlying true fastball speed.
But, what does THAT give us? He throws really hard or really soft. But, that by itself, still doesn’t tell us how EFFECTIVE he is.
The next step is to correlate that particular base skill, that scouting-level observation, into results. And Mike has given us that:
We see that a player who hits the ball at close to 80mph has a BACON of close to .300, while those who hit the ball at close to the league average (70mph) has a BACON of close to .200, and those at the league low (60mph) is just above .150.
I have to say, all those numbers look pretty low. I guess that’s what happens when you have non-linearity. For example, suppose you hit one-third of your balls at under 60mph, another third at 60-80, and the last third at over 80mph. (Numbers for illustration purposes only.) If it’s under 60mph, you get a batting average of .050 to .150, or say around an average of .120. If you hit it between 60-80, it’s .150 to .300, or an average of .220. And above 80mph, it’s from .300 all the way up to .650, for an average of say .500. That gives you an average of .280, for an average of 70mph. As you can see, the overall average for a distribution around 70mph is way above the batting average at the 70mph point.
Anyway, so what I’d like to see is this: create a DISTRIBUTION for each player, centered around his true talent horizontal speed off the bat, and apply the rates from the above chart (or a more smoothed version actually). This way, we can end up with a player’s true talent BACON, if all we know is his horizontal speed off the bat.
THAT will tell us how valuable knowing his horizontal speed off the bat is.
Thursday, May 19, 2011
Max’s article from a few weeks ago.
(I’m leaving the office in 2 minutes, so I will comment tomorrow.)
Glove-slap: Peter.
Wednesday, April 13, 2011
Great job by Studes. Love all the sleeve-rolling.
Wednesday, February 09, 2011
A pre-cursor to HITf/x by using Greg’s HitTracker to try to estimate the probability of a batted ball being a HR. Love this stuff.
Monday, November 29, 2010
How much is the hit location biased based on where a fielder is normally positioned? The following is a starting point, and as a result, will use a crude estimate.
From 1B to 3B line is 90 degrees. The circumference of a circle is 2PI*r, or PI*r/2 for a quarter circle. If we treat the radius as around 115 feet (meaning a spot somewhere between the 1B/3B bags and the 2B bag), then the distance from 1B to 3B bags, along a circular path is 180 feet. Or, 1 degree = 2 feet. I know it’s not a circle, but, we just need crude approximations.
Also remember that a SS / 2B are positioned around -16 / +16 degrees, where 0 degrees is 2B.
Peter provided us with this data from HITf/x (FX) and GameDay (GD), based on spray angles in 4 degree slices:
Spray FX GD rate
-44 107 61 175%
-40 244 134 182%
-36 253 297 85%
-32 309 389 79%
-28 306 458 67%
-24 316 347 91%
-20 369 382 97%
-16 312 429 73%
-12 339 300 113%
-8 328 218 150%
-4 329 204 161%
0 314 246 128%
4 291 180 162%
8 322 156 206%
12 305 214 143%
16 280 377 74%
20 293 455 64%
24 260 276 94%
28 234 314 75%
32 208 293 71%
36 205 248 83%
40 149 130 115%
44 85 55 155%
rate is FX / GD.
We see that at -16 (meaning -14 to -18 degrees), groundballs are recorded by Gameday far more frequently than HITf/x is recording. We see an enormous bias in the holes as well.
Now, let’s try an experiment. Let’s say that the Gameday scorers agree with HITf/x perfectly on half of the batted ball locations, and are off by 4 degrees (8 feet) on the other half? Let’s start with the 314 balls up the middle (-2 to +2 degrees). Gameday marks 157 of those up the middle, and for the other 157 record half to the left (-6 to -2) and half to the right (+2 to +6).
At the -4 degrees (-6 to -2), HitF/x had 329 balls, of which half Gameday agrees with, and the other half are all more toward the SS side (at -8 degrees).
So, this is what we have at -4 degrees as per Gameday:
164.5 balls that HITf/x marked at -4 degrees
78.5 balls that HITf/x marked at 0 degrees
That’s a total of 243 balls marked by Gameday at -4 degrees under this illustration, compared to the 329 actally recorded by HITf/x… but still a far away from the 204 actually recorded by Gameday.
The same thing happens at -8 degrees: of the 328, 164 are properly marked by Gameday, the rest are marked toward the SS side (at -12 degrees). And we go on until we get to -20 degrees, where the shift happens toward the 2B bag. This is the result of all that:
Spray FX GD rate Tango rate
-44 107 61 175% 53.5 88%
-40 244 134 182% 175.5 131%
-36 253 297 85% 248.5 84%
-32 309 389 79% 281.0 72%
-28 306 458 67% 307.5 67%
-24 316 347 91% 311.0 90%
-20 369 382 97% 498.5 130%
-16 312 429 73% 510.0 119%
-12 339 300 113% 333.5 111%
-8 328 218 150% 328.5 151%
-4 329 204 161% 243.0 119%
0 314 246 128% 157.0 64%
4 291 180 162% 224.0 124%
8 322 156 206% 322.0 206%
12 305 214 143% 305.0 143%
16 280 377 74% 426.5 113%
20 293 455 64% 423.0 93%
24 260 276 94% 247.0 89%
28 234 314 75% 221.0 70%
32 208 293 71% 206.5 70%
36 205 248 83% 177.0 71%
40 149 130 115% 117.0 90%
44 85 55 155% 42.5 77%
My illustration here shows that my model bridges some of the gap. The standard deviation of the original FX/GD is 42%, while the Tango/GD is 34%.
And trying different inputs didn’t make much better difference. If I treat anything between 50% and 75% of the HITf/x data as being perfectly recorded by Gameday, the remain balls in play are about 4 degrees (about 8 feet) biased toward where the fielder is positioned.
I think we can try to construct a more elaborate model, and we’ll probably end up at the following: about half the data from Gameday will match HITf/x, and the other half will be off by 2 to 8 degrees (4 to 16 feet). The amount it will be off will be biased by either where a fielder is normally positioned, or whether a play was made or not, or how much space between fielders (the holes).
This is the framework I’m proposing. Implementations will vary.
Wednesday, September 22, 2010
I can’t see any of this at the office, but it looks interesting.
Tuesday, August 10, 2010
I guess this golf data is from Trackman.
Saturday, August 07, 2010
Cecilia’s blog on Alan’s Mantle’s 565 foot HR.
Wednesday, June 02, 2010
Courtesy of Greg (click to make bigger):
Page 1 of 1 pages
from Amazon
Recent comments
Older comments
Page 320 of 342 pages « First < 318 319 320 321 322 > Last »Complete Archive – By Category
Complete Archive – By Date