Sabermetrics Pitching 202: Intermediate Pitching Stats

Previous posts:

Sabermetrics Hitting 101: Basic Hitting Stats

Sabermetrics Hitting 202: Intermediate Hitting Stats

Sabermetrics Pitching 101: Basic Pitching Stats

I was out last week due to a business trip, but I’m back to go to the next level in pitching stats. Luckily, we’ve only got three stats to look at today, but two of them (FIP/xFIP) are worthy of an entire post on their own - but we’ll save that for another day.

---

Left On-base Percentage (LOB%)

Formula:

What does it actually mean: The percentage of baserunners that a pitcher leaves on base (versus scoring a run).

Is this useful? If so, how? Yes. LOB% is similar to BABIP in that it is a good way of measuring “luck” and/or “sequencing” for a pitcher. There is also a small amount of talent tied to this skill.

How can I use it? LOB% is particularly useful in identifying deviations from an “expected” ERA for a pitcher. If a pitcher suddenly has a really low ERA but a really high LOB%, chances are that that pitcher will not maintain that low of an ERA going forward. Inversely, a pitcher with a really high ERA and low LOB% should expect to see his ERA drop going forward.

The league average for LOB% tends to be around 72%. Generally speaking, a pitcher having a “great” season will see a LOB% as high as about 80% over the full course of a season. A pitcher having a “bad season” will see a LOB% as low as 60%. However, these high and low rates aren’t very reflective of skill and are generally not sustainable.

Deeper Dive: Pitchers with higher strikeout rates tend to have slightly higher LOB% rates than average pitchers. Most MLB starters/relievers will sit around league average over their careers regardless of skill. However, below-average pitchers (like spot starters and fringe #5s) will typically post below-average LOB% rates consistently.

Between BABIP, HR/FB%, and LOB%, these three stats provide much of the deviation between a pitcher’s ERA and FIP. Since these stats are all highly regressible and generally uncontrollable by a pitcher, it helps to explain why regressing ERA to a pitcher’s FIP is useful.

Fielding Independent Pitching (FIP)

Formula:

What does it actually mean: FIP is an estimation of a pitcher’s ability to prevent runs (ERA) independent of the level of defensive performance behind them. In other words, FIP focuses on the three main factors in a pitcher’s control - strikeouts, walks, hit batters, and home runs - and estimates the ERA accordingly. The purpose of FIP is to tell you how well the pitcher performed regardless of the amount of runs actually given up.

Is this useful? If so, how? FIP is one of the most useful tools in pitching analysis. While ERA is a more useful tool for measuring a pitcher’s run prevention in the past, FIP is useful because it is a far better tool for predicting how well a pitcher will perform going forward, e.g., if a pitcher has a 2.00 ERA but a 4.00 FIP, chances are that that pitcher will carry an ERA much closer to 4.00 in his future starts.

For clarification, the FIP constant (C_FIP) is a manually-calculated value to align FIP with pitcher ERAs. The FIP constant applies equally to all pitchers in a given season and generally hovers around 3.10. So a pitcher with a FIP below 3.10 will have a fraction component (the (HR+BB+HBP-K)/IP part) that is negative thanks to a very high amount of Ks and low HR/BB/HBPs.

How can I use it? FIP is actually extremely easy to use - thanks to the FIP constant (C_FIP), FIP is aligned to match and scale with ERA. So, you can essentially use FIP as a pitcher’s “predictive ERA” or “ERA talent”. Beyond that, FIP is just like ERA - the lower the FIP, the better. And FIP will match ERA: the league average FIP is around 4.00-4.20, a great FIP is below 3.00, and a bad FIP is 4.70 or higher.

FIP is generally a very broad and reliable stat and is a great first step in analyzing a pitcher. However, there will always be differences between ERA and FIP - especially over the small sample of one season - so it is important to analyze FIP with respect to this. BABIP, specifically, plays a very large role in ERA and FIP differences.

Deeper Dive: If you look at FIP, you’ll notice it uses IP in the denominator, just like ERA, K/9, and BB/9. This allows us to rewrite the equation as follows:

or:

It’s rare to actually see FIP ever broken down into the components like this, but it is useful in showing how HR/9, K/9, and BB/9 directly relate to each other and FIP. As you can see, home runs have a huge impact on FIP (and therefore ERA).

Essentially, the way to translate this is:

Every increase in 1.00 HR/9 = 1.44 higher FIP
Every increase in 1.00 BB/9 = 0.33 higher FIP
Every increase in 1.00 K/9 = 0.22 lower FIP

Expected Fielding Independent Pitching (xFIP)

Formula:

What does it actually mean: xFIP is the same as FIP except that the raw HR rate (HR/9) is adjusted for league average HR/FB%. Essentially, xFIP is the measure of a pitcher’s run prevention skills, independent of defense performance behind him, while adjusting for the amount of home runs he should have given up (based on league average HR/FB%, which is usually around 10%).

Is this useful? If so, how? xFIP is very similar to FIP and almost always better. Most pitchers don’t have much control over their HR/FB% and due to the low-ish amount of flyballs they allow during a season, a few extra (or less) home runs than expected can have a fairly large impact on their FIP and therefore, their ERA. So for a typical pitcher (which will hover around league-average HR/FB% over their career), xFIP can often give a slightly better picture of their true performance than FIP in a given season.

However, xFIP has some drawbacks, mostly with regard to the pitchers that CAN influence their HR/FB%.

How can I use it? xFIP is just as easy as FIP in that is scaled to match ERA, so treat the values and methods that were given above in FIP as you would to xFIP. However, be careful of universally applying xFIP, because some pitchers will be skewed.

If a pitcher has a higher-than-average HR/FB% talent, then xFIP will generally overrate them. If a pitcher has a lower-than-average HR/FB% talent, then xFIP will generally underrate them. Exceptions have to be made for groundballers (due to their extremely low flyball rates) but for most other pitchers, the above qualifications appliy.

In general, xFIP is usually more predictive than FIP in predicting future ERA performance, but make sure to consider the pitcher’s true HR/FB% talent. It usually requires a couple seasons worth of data before we can be comfortable with where they truly are.

Deeper Dive: There are two other things to note in xFIP. First, if you look at the HR term, you’ll actually see that Flyball rate is actually built in to xFIP:

So, extreme groundballers, whom tend to be prone to higher-than-average HR/FB% rates, can still post very low xFIP rates because their FB/9 term will be very very low. The classic example of this is Brandon Webb. Webb had a career 3.27 ERA, 3.50 FIP, and 3.30 xFIP. So, despite the higher-than-average HR/FB% for his career (13.5%), Webby still had a better xFIP than FIP thanks to his extremely low FB/9 rate. This is another reason why groundballers are generally underrated - the immense suppression of flyballs helps them avoid homers and extra base hits in general. So while xFIP isn’t explicitly addressing non-homers, the reduction in flyballs is still very valuable.

The other note in xFIP is that the LgHR/FB% term is reset every year which generally makes the individual season values for xFIP a bit more volatile than desired. However, this also makes it very useful in adjusting for league run environments, which often change quickly.

All-in-all, xFIP is one of the most predictive stats we have for ERA.

---

That’s all for this week, please post any comments, questions, or criticisms you might have down below. See you next week!