Filed under:

# Stop using K/9 and BB/9!

Here's a look at BABIP's effect on the 'per 9' metrics and why you shouldn't use them.

### Start Using K/PA and BB/PA!

For whatever reason, K/9 and BB/9 have both been given a pass by the recent Sabermetrics invasion. It seems that quite a bit of well-informed people still use them, and it's become something of a pet peeve of mine.

I don't mean to offend anyone by telling them they are using the wrong metric, because certainly the differences are often marginal. But I do find it important, personally, and I'd like to illustrate why.

What happens when we use the 'per 9' metrics is that we lose accuracy, because our measurements have become subject to the tyrannical forces of BAbip. As a pitcher allows more Hits per Ball in Play, he becomes less efficient. He ends up facing more batters and getting fewer outs, which consequently means fewer innings. But if he's still striking out batters at the same rate (say 20%) all the while, his K/9 is going to look a lot shinier with those fewer Innings.

In the table below we have 21 theoretical pitchers, Pitcher "A" through Pitcher "U". All 21 of them faced 800 batters and struck out 160 batters while walking 64 of them. Therefore, each one of these pitchers has a K/PA, or K%, of 20.0 and a BB% of 8.0.

Each of these pitchers also allowed 576 Balls in Play. The only difference is between these pitchers is the amount of those 576 Balls in Play that ended up as Hits.

 PIT. BABIP H BIP K BB BF K% BB% IP BB/9 K/9 A .227 131 576 160 64 800 20 8 201.7 2.86 7.14 B .234 135 576 160 64 800 20 8 200.3 2.88 7.19 C .241 139 576 160 64 800 20 8 199 2.89 7.24 D .248 143 576 160 64 800 20 8 197.7 2.91 7.28 E .255 147 576 160 64 800 20 8 196.3 2.93 7.33 F .262 151 576 160 64 800 20 8 195 2.95 7.38 G .269 155 576 160 64 800 20 8 193.7 2.97 7.44 H .276 159 576 160 64 800 20 8 192.3 2.99 7.49 I .283 163 576 160 64 800 20 8 191 3.02 7.54 J .290 167 576 160 64 800 20 8 189.7 3.04 7.59 K .297 171 576 160 64 800 20 8 188.3 3.06 7.65 L .304 175 576 160 64 800 20 8 187 3.08 7.70 M .311 179 576 160 64 800 20 8 185.7 3.10 7.76 N .318 183 576 160 64 800 20 8 184.3 3.12 7.81 O .325 187 576 160 64 800 20 8 183 3.15 7.87 P .332 191 576 160 64 800 20 8 181.7 3.17 7.93 Q .339 195 576 160 64 800 20 8 180.3 3.19 7.99 R .345 199 576 160 64 800 20 8 179 3.22 8.04 S .352 203 576 160 64 800 20 8 177.7 3.24 8.11 T .359 207 576 160 64 800 20 8 176.3 3.27 8.17 U .366 211 576 160 64 800 20 8 175 3.29 8.23

Pitcher "F" allowed just 151 hits out of his 576 BIP, for a final BABIP of just .262. But Pitcher "R" allowed 199 hits out of his 576 BIP adding up to a much uglier BABIP of .345. Because of this dramatic difference in BABIP between the two pitchers, Pitcher 'F' was able to pitch much deeper into games amassing 195 IP, while the Pitcher 'R' just 179.

But with K/9 we end up rewarding the high BABIP for Pitcher 'R' by using his much smaller IP denominator to make our K/9 calculation. He ends up with a K/9 of 8.0 while Pitcher 'F' ends up with a K/9 of just 7.38, despite striking out batters at the same rate!

Obviously, this is a fairly extreme example, as not many pitchers post a .345 BABIP over the course of a full season. But the hypothetical pitcher with a league-average BABIP of .297 in this table posts a K/9 of 7.65, almost a half of a strikeout per game higher than Pitcher 'F' with the low-BABIP. In my mind, this is very much a significant difference.

But, what's worse is that in the case of pitchers who have exhibited a definitive ability to produce weak contact and lower BAbips, we are punishing them for their excellence when we use K/9. Let's take notorious BABIP-Master Mariano Rivera.

On page 2 of Fangraph's RP career leaderboard, we see that Rivera's 8.32 K/9 ranks well below that of, say a "Norm Charlton" and his 9.04 K/9. But Rivera has posted a 23.4% K-rate through 2011 and Charlton a slightly lower 23.3% in his career.

Do you really want to go on pretending Norm Charlton was a better strikeout pitcher than The Sandman?