clock menu more-arrow no yes

Filed under:

Talking Sabermetrics: More On Pitchers and Runners On Base

New, 8 comments
 Pitcher Jordan Lyles of the Houston Astros takes the field.  (Photo by Bob Levey/Getty Images)
Pitcher Jordan Lyles of the Houston Astros takes the field. (Photo by Bob Levey/Getty Images)
Getty Images

Last week, we talked about the new stats for pitching dependent components, and speculated on what it might show about the Astros' pitchers' tendency to perform better on a Fielding Independent Pitching (FIP) measure than ERA or simple runs allowed per 9 innings. This also pointed to the problems Lucas Harrell and Jordan Lyles have experienced with runners on base.

Beyond the Boxscore's James Gentile presented two articles last week on BABIP with runners on base and pitchers' success in inducing groundballs with runners on base. Both topics tie in nicely with the Astros' starting pitchers' overall negative "wins" in in the "LOB-Wins" stats (see the previous Talking Sabermetrics article for more explanation). First, I'll summarize Gentile's conclusions, and then look at how the topic may relate to the Astros' starters.

In the first BtB article, Gentile examined the 15 worst BABIP seasons with runners on base during the last 20 years. These pitchers had a BABIP as much as 130 points higher with runners on than with the bases empty. Not surprisingly, almost all of them had an ERA worse than their FIP. Interestingly, the list is composed of all kinds of pitchers, from the elite, such as Roger Clemens, Greg Maddux, and Felix Hernandez, to the less renowned, like Glendon Rusch, Casey Fossum, and Luke Hochevar. Here is the take away from this article: nearly all of the pitchers on the list had major regression in their BABIP-ROB in subsequent seasons, with a similar regression of their subsequent season ERAs in the direction of FIP.

The second BtB article is aimed at testing whether pitchers are more likely to induce groundballs with runners on base with less than two outs. The analysis found no evidence that pitchers induce higher groundball rates in those situations. Undoubtedly, pitchers usually want a GIDP in those situations, but the hitters are aware of that too and presumably try to avoid hitting a groundball. Despite the general conclusion, a few pitchers appeared to sustain higher groundball rates in DP situations. Two of the top three pitchers over the last 10 years at increasing their groundball rates in those situations are former Astros: Billy Wagner and Brad Lidge.

I will now turn these two topics in the direction of Astros' starters Bud Norris, Lucas Harrell, and Jordan Lyles. Harrell and Lyles are among the worst (i.e., most negative) major league pitchers at LOB-Wins. Norris' weakness has been BIP-Wins more than LOB-Wins. Norris' negative BIP-Wins is associated with a high overall BABIP. Perhaps we can shed additional light on why all three starters have ERAs higher than their FIP.

The table below compares overall BABIP with BABIP-runners on base, BABIP-high leverage, and BABIP-bases empty. Also shown are Groundball Rate with runners on base, high leverage situations, and bases empty.








BABIP-runners on




BABIP-hi leverage




BABIP-base empty








GB%-runners on




GB%-hi leverage




GB%-base empty




LOB-Wins is associated with the timing/sequence of pitching outcomes. So, a higher BABIP during critical high leverage situations can lead to negative LOB-Wins. High leverage situations involve a relatively small number of innings which can significantly damage a pitcher's ERA. Jordan Lyles' BABIP in high leverage situations is among the five highest in the major leagues. Therefore, Lyles' very high BABIP in high leverage situations (as well as runners on base situations generally) provides a tidy explanation for an ERA which is 0.6 runs worse than his FIP. The good news is that the unfortunate timing of Lyles' BABIP is susceptible to regression. As the BtB article indicated, pitchers in Lyles' situation frequently experience significant ERA improvements in subsequent seasons.

Double plays help a pitcher offset the impact of BABIP with runners on base. Neither Lyles or Harrell induce higher groundball rates with runners on base or in high leverage situations. (That's not to say that Lyles and Harrell have a poor groundball rate in high leverage situations--in fact, both pitchers induce high groundball rates regardless of the situation.) Although Norris generally is a fly ball pitcher, he has exhibited a higher groundball rate in high leverage situations. It's possible that Norris relies more heavily on two seam fastballs with runners on base in order to increase the chances of a double play.

Harrell is an interesting case, because his weakness in LOB-Wins is not caused by a high BABIP during high leverage situations. In fact, Harrell's BABIP declines significantly (relative to bases empty) during high leverage and ROB situations. Harrell's LOB-Wins underperformance can be traced to other causes, starting with his peripherals. With runners on base, Harrell's K rate declines and his BB rate increases. Harrell's K:BB ratio declines from 3.31 with the bases empty to 0.69 with runners on base. Harrell's HR/9 rate also increases with runners on base. As a result, compared to his bases empty performance, Harrell's OBP-against and OPS-against is 47 points and 100 points higher, respectively, with runners on base. Harrell's LOB-Wins also isn't helped by the fact that 6 of the 7 errors behind him occurred with runners on base.

Bud Norris' overall BABIP is relatively high, suggesting that his ERA will improve in the future, assuming normal regression. However, like Harrell, Norris has benefited from below average BABIP with runners on base. This might indicate that an improvement in Norris' BIP-Wins will be partially offset by a regression in LOB-Wins. But, unlike Harrell, Norris' peripherals improve with runners on base. Norris' K:BB ratio increases from 2.32 with the bases empty to 2.72 with ROB. A similar pattern emerges in comparing high leverage and low leverage situations. Thus, Norris may apply a true skill in suppressing runs during high leverage situations.

I am guessing that Norris and Lyles will incur improved results as their ERA normalizes in the direction of their projected FIP. This may not be the case for Harrell. However, the difference between Harrell's ERA and FIP is relatively small. Potentially any positive or negative regression will be modest in Harrell's case.