Every year we find pitchers who beat or trail the league average in LOB%. This stat is often considered a stat that should return to the league average and as such can be used to predict a return to the mean in pitchers. The expected league average is 72% and was seen again in 2008.
Looking at the numbers, though, you can see that certain pitchers have some control over this and repeat higher or lower numbers every year. Some examples who have averaged over 75% in their career are: Jake Peavy (77.6%), Johan Santana (77.3%) and Scott Kazmir (75.6%). The first thing I immediately noticed is that their ability to strikeout batters must play a huge role, as these guys all have higher K/9 numbers.
They also all have ERA numbers that beat their FIP by up to .25, as the FIP does not take into account any changes in LOB%. If this is the case, we can effectively predict a LOB% and how much that will effect the FIP. This would be beneficial for analyzing young players without much of a major league track record and establishing if they can maintain a higher LOB%.
The first thing I did was try to correlate K/9 with LOB% for all pitchers in 2008, and the r score of 0.19 didn't make me very confident, but this includes a lot of players who pitched a lot of limited innings and luck would have played a large role in their LOB%.
I cut the players down to those who pitched over 160 innings. I reran the correlation and found we did increase our r score to 0.42. This is a medium correlation, but the improvement shows us we are in the right direction.
I thought we could do this again and improve our results by collecting the last three years' data from pitchers who threw over 500 innings. This limited us to 57 pitchers but gave us a good sampling to rerun the correlation. The new r score of 0.68 is again showing us we have an improving correlation and can start to see that a higher K/9 will help a pitcher maintain a higher LOB%.
So far it seems that to ensure a pitcher is reaching his correct LOB%, he needs at least 500 IP. If they have not reached that number, then luck is still playing a large role in their LOB%. To continue confirming this, I ran the numbers for pitchers over 500 IP from 2000-03. The r score for this is 0.51, which is down a bit. It is still a sign of correlation, but not quite as strong.
The next step was to see if we can find a correlation with other pitching controlled stats. My first thought was BB/9 or K/BB. Running this on the three-year sampling size, I found the r score for BB/9 to LOB% to be 0.05, which means there is absolutely no correlation.
Perhaps K/BB would show something, although after that result for BB/9 I was not confident. The r score on the same group was 0.42. This is a medium correlation, but it really just seems to be the BB/9 watering down the K/9.
To look at individual pitchers beating the average LOB%, I set a cutoff of greater than 73% and saw that pitchers with a K/9 over 7.00 had a 52% chance of a LOB% over 73. On the other hand, pitchers with a K/9 under 7.00 never reached a 73% LOB in this 3 year sample size of 500+ IP pitchers.
LOB% is known to be skill related, but it isn't the ability to prevent walks. It's the ability to get outs without the ball being put in play.
My next step is to address how this could affect FIP numbers and pitcher projections. I still value pitchers with K/BB over 2.50, but it seems a 7.00 K/9 adds to the ability to beat projections consistently.
Here is the Graph with the R2 value. This is on SP for the past three years data.
6 comments:
It's a rather short sample size, but Daisuke's two seasons would be another player that supports the Johan, Peavy, Kaz group.
With a K/9 over eight I agree that his LOB% can stay elevated, but with the low BABIP, high walks, low HR/FB and poor GB% he is going to have struggles in 2009.
Since Boston is a savvy statistical team I could see them not let Daisuke think he did as well as he did.
With a FIP of 4.03 I could see him in the high 3's next year as he beats his FIP with an elevated LOB%.
I think Bill James projection is good with a FIP of 3.81 and an ERA of 3.58, but that all hinges on an improvement in his walk rate.
I would be interested in finding out which pitchers fall under the high K/9 category, that consistently fall short of (or rather, fall above %-wise) the LOB% average. First name that comes to my mind as a possibility would be Daniel Cabrera. I have the numbers available, but lack the motivation or time to look it up. Sometimes, those power K pitchers have too many control issues.
Well Cabrera does have a K/9 over 7 the past three years, but it went from 9.55 to 7.31 to 4.75. I don't think we can consider him the high K/9 guy anymore.
A great example is Javier Vazquez. His K/9 for his career is 7.99, but his LOB% is 70.4%, which doesn't seem like much, but over a full career that is quite a few runs allowed. He has always been worse than his FIP and is a case that doesn't follow this model.
Isn't this kind of silly (not said in a harsh way)? Better pitchers will get all batters out more often, including the ones that are up when men are on base. And K's are the out of choice so the better pitchers throw them and the better pitchers get out of more of the jams that they do get into. That's why the pitchers on your original examination of who had a low left OB% were mostly sucky pitchers. What would be interesting is if K pitchers have the ability to get batters with men on out more often then they get all batters they face out. Then you'd have something.
There are two things we found here.
1) 7.00 K/9 seems to be the breaking point for improving LOB%
2) BB/9 has very little to do with increasing LOB%
The question I think we would have to look at first is how many pitchers can consistently strikeout more batters with runners on. Just one year might not supply enough data to eliminate luck.
Post a Comment