Finding LOB% Overachievers and Underachievers
There are many pitcher stats that we review and analyze to predict regression or breakout. While there is plenty of focus on BABIP (batting average on balls in play), HR/FB (home-run to flyball ratio) and ERA (earned run average), LOB% (Left On Base percentage), is a stat that I think is sorely overlooked in the fantasy baseball spectrum (and the baseball world at large). For those who need a refresher or aren’t aware, LOB% measures the percentage of base runners that a pitcher strands on base over the course of a season. It is calculated by this simple formula:
An average LOB% for a pitcher will land somewhere in the 70-80% range.
To get a good feel of what a good LOB% looks like, here were the top ten LOB%s in 2017, minimum 50 innings pitched.
|Craig Kimbrel||Red Sox||69||93.9%|
And then we have the unfortunates- the ten lowest LOB% marks in 2017.
|Roberto Osuna||Blue Jays||64||59.5%|
|Brett Anderson||Blue Jays||55.1||60.9%|
So exactly why is LOB% so important? As one can guess, it has a pretty strong correlation (r = -0.76) with ERA. Let’s take a look at Robbie Ray as an example. Without looking at ERA, you’d think he was basically the same pitcher in 2016 and 2017:
- 2016- 3.76 FIP, 3.07 K/BB, 1.24 HR/9
- 2017- 3.72 FIP, 3.07 K/BB, 1.28 HR/9
Dig deeper, however, and you’ll find that there was a major swing in his LOB% between the two years. That swing went from a below-average mark of 68.7% in 2016 (which led to a 4.90 ERA) all the way up to a mark of 84.5% in 2017, helping deliver a 2.89 ERA and all kinds of hype.
We can generally expect pitchers to veer towards the mean of 70-80%, but then in order to get an even clearer picture of this stat the concept of xLOB% comes in. Like every other “x-stat,” xLOB% is the expected rate of runners stranded. The great Mike Podhorzer went in-depth on this in a piece of his at TBT (The Hardball Times, for the unintiated), even going so far as to devising his own detailed xLOB% formula.
“xLOB% = 0.931 – (1B% * 0.877) – (2B% * 1.646) – (3B% * 2.164) – (HR% * 0.377) – (BB% * .380) – (HBP% * 0.640) – (HBP% * 0.640) – (WP% * 0.851) + (K% * 0.137) + (GDP% * 0.401) – (SB% * 0.142) + (CS% * 0.264) + (PKO% * 0.519) + (True IFFB% * 0.351)”
This gave me the spark of inspiration, so I decided to deep dive into some research of my own. Using a lot of the same techniques and abstractions as Podhorzer did, I developed a very similar formula of my own. And here’s what I found, starting with the top ten xLOB% marks.
|Craig Kimbrel||Red Sox||69||93.9%||86.9%|
|Carl Edwards Jr.||Cubs||66.1||78.3%||83.5%|
And then the ten lowest…
|Brett Anderson||Blue Jays||55.1||60.9%||62.9%|
My biggest goal in this endeavor was finding the overachievers/underachievers in LOB%, and looking for regression candidates (as we do with BABIP, ERA, and HR/FB). By subtracting a pitcher’s xLOB% from their LOB%, I found the ten biggest LOB% underachievers among starting pitchers.
|Doug Fister||Red Sox||90.1||63.3%||72.6%||-9.3%|
|Joe Biagini||Blue Jays||119.2||61.5%||70.7%||-9.2%|
And then the biggest underachievers among relievers…
|Roberto Osuna||Blue Jays||64||59.5%||79.6%||-20.1%|
Some interesting names show up in these findings. The first one I noticed was Dinelson Lamet. Looking at some underlying numbers for him, such as a 10.9 K/9 and .207 BAA, one would think he’s better than the 4.57 ERA he put up in 2017. That’s a reasonable assumption, as some of the differences between his LOB% and xLOB% can be to blame. Moreover, Steamer projects him to get a bit closer to his xLOB% numbers (projected 73.5 LOB%), which is right in line with a projected improvement in ERA (4.17).
Another name that caught my eye was Roberto Osuna. 2017 was a one to forget for this young closer, as it seems he was bit hard by the ERA monster (which was one of the causes for his AL-leading 10 blown saves). However, one could easily point to his 59.5 LOB%, second worst rate in all of baseball (min. 50 IP), as the true culprit. His xLOB% of 79.6 indicates that one should expect some regression to the mean next year for Osuna.
Now time for the lucky ones. The overachievers. First with the starting pitchers…
|Marcus Stroman||Blue Jays||201||78.1%||72.2%||5.9%|
|Drew Pomeranz||Red Sox||173.2||80.0%||74.3%||5.7%|
One case that I found interesting was Jon Brebbia. Brebbia put up a very unsustainable 90.9 LOB% last season, putting him 2nd among relievers in my metric. The numbers suggest that he should have been closer to league-average, and so I’ll call bluff on his 2.44 ERA. Steamer pins him at a much more normal 72.0 LOB% for the upcoming season, along with projecting him to consequently regress to a 4.25 ERA.
Now that we know who over-performed and under-performed on their LOB%, it is very important to understand the predictive value in this metric. Once I matched a player’s xLOB% from years past with his LOB% the following year, I found that the correlation was 0.32- neither strong nor insignificant. This is interesting, but to tell the truth, it isn’t the prettiest predictor. I think this is a great avenue for further investigation and refinement, with a chance to turn it into a predictor with a greater correlation.
No matter what the season-to-season correlation is as a whole, LOB% and xLOB% are stats that can be used to identify prime regression and progression candidates.