THE BOOK--Playing The Percentages In Baseball

Not too long ago I queried my database at firstinning.com and looked at all hitters making the transition from A+ to AA, AA to AAA, etc. and found that, on average, the player’s line drive rates *increased* after the transition to the higher league even though they were probably facing better pitching.
So I came to the similar conclusions about scorer’s subjectivity, but I’m not sure why anybody should be surprised. Like you said, until folks start measuring hang time we’re better off simply looking at air/ground distinctions unless you want to do complicated conversion rates.

The first version of this appears not to have gone through. Apologies if it duplicates.

a few observations
1) major leaguers hit more home runs because they are older, stronger and better hitters. Their well hit fly balls therefore do not stay in play and drop out of the sample of outfield fly “ball in play” hits, making the outfield fly in-play hit rate drop. likewise, Triple A players hit more home runs than double A, etc. That’s the main reason why the hit rate on fly balls in play keeps dropping at a league level as league classification increases. Since this claim is about scorer bias on the hit type itself, there is no rationale for subtracting home runs.

I only refigured AAA, but the FBH rate is .303 and the ABH rate is .463 if HR are left in. The rates for all levels need to be figured this way and then we can see how striking the differences really are. (Incidentally, it appears that Eric Van worked solely with aggregate data and assumed that all HR are FB HR for the purpose of adjusting his rates.)

2) quality of lighting in night games may correlate to league level as well. If the batter and the fielder can see the ball better in a AA game than a single A game because of better lighting, this increases the batter’s likelihood of hitting a line drive and a fielder’s chance of catching it once hit. That could be why individual LD rates might improve even as better pitching is encountered as the individual changes levels.

3) Others may have already looked at this in detail, but actual quality of competition should not increase in a linear way for each classification level a player advances. Without knowing those relationships, Eric Van’s chart couldn’t be correctly interpreted anyway.

4) Greg Rybarczyk (of Hittracker) posted at SOSH as a followup to Tango that he can record hang times on fly balls to within .02 seconds.  FWIW, I don’t believe that you can be remotely close to that accurate when timing in person instead of from videotape. I also doubt that he can really be that accurate against video, where you have the advantage of a head on view and foreknowledge that contact is going to be made so that the timer can anticipate. I can believe that Greg is consistent with himself to within .02 as he repeatedly times the same hit, but that is not the same thing as being consistent with reality to within .02 seconds. The error bar on in person hand timing should be .20 seconds or greater. Still will be great to have it, but there’s no merit to reporting hand-timing to the hundredth of a second.

"That’s the main reason why the hit rate on fly balls in play keeps dropping at a league level as league classification increases.”

For what its worth, this effect holds even when looking at all bals in play, as I have.

I think the lighting quality issue is interesting I personally doubt this will explain much of the differences because otherwise why would the rates be so similar in the smaller stadiums of the lower minor leagues?
It’s something that should be proven to be false, however.

Chris,

not sure what you mean, on either count. Can you elaborate?

The LD% for single A (so atlantic + midwest) is 15.0 out of all balls in play, growing to 17.0 in aaa and ~20% in MLB. I don’t have any quarrel with this aspect of Eric Van’s chart. In physical terms, the difference in solidity of contact between a line drive and other hit types is small. If I am able to follow the flight of the pitch better and get the head of my bat 1/16 of an inch closer to the center of the ball, I probably improve my LD rate by another 5-10%.

GB hit % for 2006, by classification
A: .246
AA: .232
AAA: .238
MLB historically (recently) has been close to .230
If that’s part of what you’re talking about for all balls in play, that’s not a clear effect.

My response was related to your claims about why the “outfield fly in-play hit rate” drops. The hit rate for fly balls drops as you move up levels even if you include all batted balls and HR.

Regarding the lighting effects (your point #2), I suppose someone should just look at how the batted ball types are distrivuted during day games/night games at each level. That should give some insight into whether or not lighting plays a role in the league level differences.

Chris,

you are right that the hit rate on outfield flies generally drops as classification increases, but of course I was critiquing Eric Van’s statistic which omitted homeruns. My claim looks defensible that this omission is “...the main reason why the hit rate on fly balls in play [his statistic] keeps dropping ... as league classification increases”, since it accounts for roughly half of the fluctuations:

MLB ..AAA ....AA .....A+ .....A ....A- ....R+ ...R
0.43 0.461 0.467 0.472 0.488 0.497 0.499 0.493 ..GB%
0.15 0.142 0.141 0.144 0.148 0.153 0.142 0.148 ..Pop/Air
0.19 0.170 0.150 0.127 0.136 0.143 0.150 0.132 ..LD%
0.65 0.584 0.489 0.392 0.458 0.503 0.538 0.438 ..LD : FB
0.23 0.238 0.232 0.243 0.246 0.233 0.249 0.230 ..GB hit%
0.73 0.737 0.735 0.753 0.733 0.713 0.746 0.739 ..LD hit%
0.28 0.303 0.322 0.353 0.330 0.312 0.357 0.365 ..FB hit%
0.13 0.224 0.256 0.297 0.274 0.270 0.300 0.333 ..OFBIPH

[2006 MLB is included for completeness, although values are variously estimated from prior years or supplied from other 2006 sources.]
The second to last line is outfield flyball hit rate with homeruns included, and the last line is Eric Van’s version. With homeruns added back in, there is a much milder decline from AA to AAA to MLB, and there is a decline from rookie ball to low A, but there actually is an increase from low A to A to high A.

My version of the fly ball hit statistic doesn’t demand substantial scorer bias as an explanation. Those differences seem within a range which could be explained by park effects (including lighting) and differing balances between hitter quality and pitcher quality in different classifications, perhaps due to differing aging trajectories, different rate of adaptation to wooden bats, different adlection rates into higher leagues due to differing injury rates for pitchers and hitters, differing availability of scouting information in the higher leagues both in terms of how to pitch to hitters and how to position fielders ...

We wouldn’t expect scorer bias to have much impact on classification of popups or groundballs; and indeed popup rate (pop/pop+ld+OFfly) looks pretty constant accross all classifications. Ground ball percent drops as classification increases - hitters therefore become better at hitting either line drives or fly balls or both as the classification increases.

Note that the hit% both for groundballs and for line drives is basically very consistent accross all classifications. The consistency in line drive hit rate is not inevitably contradictory to Eric’s hypothesis that balls which aren’t really line drives are getting scored as line drives, but I would have expected the hit % on line drives to be dropping if balls with more catchable trajectories were being scored as line drives at higher levels. You would have to believe that there are other effects or scoring biases which just happen to balance out this particular supposed scoring bias.

The line I haven’t discussed in my chart is LD:FB ratio, which increases, though irregularly, as classification increases. I have no problem believing that this is because of a selection bias in favor of advancing hitters who can hit actual line drives.

So I don’t see anything about line drive rate or line drive outcome which justifies the hypothesis of significant scorer bias. To me the sole eveidence suggesting such a bias is decline in hit rate on outfield flies, which I see as a mild decline - in the range explicable by park effects and most optimal defensive positioning due to greater investment in scouting.