Projecting batting average

Question

In the glossary, under principles and splits, it mentions that in 2010, "line drives fell for hits 72% of the time, groundballs fell for hits 23% of the time, and flyballs went for hits 14% of the time. However, those percentages are based off BABIP, which excludes homeruns from its calculation; if you include homeruns, the flyball percentage bounces up to 21% of the time".

So I've taken all qualified hitters for 2010, and basically tried to determine a projected batting average.  The formula is essentially [(# of LD * 72%) + (# of GB * 23%) + (# of FB * 21%)] / AB

Now obviously this will not match perfectly, but it should give a pretty good indication on how a hitter should have fared during the season.  My results were not even close.  For example, Josh Hamilton hit .359, but with my formula, it comes out to .273, almost a 9% difference.  I took the average of the qualified hitters averages for the entire league, and I got .273.  I took my average of expected batting average based on the figures above, and got .253.

Is there any reason for this on why it is so far off?  I was expecting results within a couple of percentage points.

 

Thanks

Answer 1

This formula is essentially creating an xBABIP, then converting it to an average.  If you look at Hamilton's batted ball splits that season, you'll see his GB and LD BABIP/AVG are about 50 points higher than average, but his FB AVG is double the league average.  If you take away half of his 65 flyball hits, you get an drop of 33/518 = 64 points in average, which is much closer to your number.  Luck is still greatly involved, so this is more of a projection tool than for retrospect.

Comments

I understand what you're saying and it makes complete sense.  However, for the stat to read that "line drives fell for hits 72% of the time, groundballs fell for hits 23% of the time, and flyballs went for hits 14% of the time. However, those percentages are based off BABIP, which excludes homeruns from its calculation; if you include homeruns, the flyball percentage bounces up to 21% of the time", I'm assuming it's applied to the league as a whole, not each individual player.

Therefore, when looking at the entire league, both averages (actual average and the number I calculated) should come very close.  The actual number of hits landing for line drives, groundballs, etc, and computing an average directly relate, while BABIP should have no influence on what I'm trying to do.

I'm not sure what I'm saying is clear, but it's the best way I can explain it.

Essentially, the overall league average should match my league average calculation because it is using an actual number of hits.

---

Yeah I forgot to talk about the leaguewide stats.  Looks like the .210 average on flyballs is the problem.  When looking at the league stats, flyballs had an average of .236.  Using leaguewide batted ball totals with the new FB average, I get .256, only a point off the true average.  Bunts are left out of this equation, so those bring the average up to .259.  Since not all batted balls result in official at-bats (sac flies), removing those 1301 batted balls bring me to the correct .257 average.  Well, that was fun...