Note: All stats referenced go up to but not including the Minnesota Twins series, beginning on 7/25.
After watching Derek Holland poop things up again against the Angels on Wednesday, I felt compelled to take a pretty amateurish statistical look at his inconsistency. I perceive a common complaint made about Holland (although of course, it might not actually be a common complaint) – apart from his erratic start-to-start form – is regarding his tendency to give up big run-scoring innings. So I went back and tracked each start made by the Rangers’ five main starters this season, marking down the actual number of runs they gave up on each instance where they allowed a run-scoring inning.
Of course, the method I used had an obvious flaw – as it didn’t take into account the possibility that a reliever may have allowed inherited baserunners to score. However, I’m hoping that such a problem isn’t enough to completely hinder at least a general attempt to look at how things break down, in terms of the actual number of runs each pitcher gives up every time they go out to the mound and end up allowing at least one run to score.
As it goes (with tracking up to the end of the Toronto series), on 38 occasions has C.J. Wilson gone out to the mound and given up at least one run (in terms of either letting it be driven in off his pitching or having a reliever lose an inherited baserunner). Holland has done this on 32 occasions; as has Colby Lewis. Matt Harrison has given up at least one run on 23 occasions, while Alexi Ogando has done so 18 occasions. Now obviously the reason why C.J. leads the staff in this particular counting statistic, is because being as good and handsome as he is, he has pitched more innings (147 – 17.2 more than Colby Lewis’s second place inning count of 128.1) than anyone else. Likewise, the fact that the number of times in which a pitcher has given up a run-scoring inning correlates rather closely to the number of innings they have pitched, applies to the other members of the pitching staff.
After doing the proper tallying up of runs given up on each instance of a ‘run-scoring inning’ from a pitcher, I produced histograms, which are shown below:
From this data, I took any instance where a pitcher gave up four or more earned runs in an inning, and considered such observations to be evidence of them having allowed a "big inning".
On this count, Holland was found to have given up a "big inning" five times from the 32 instances where he has allowed at least one run in an inning. On the other hand, C.J has not allowed more than three baserunners to score in any one inning this season (and even so, he has allowed three earned runs in an inning on only one occasion) – which is pretty incredible. For Colby, it’s three times from 32 instances; for Harrison it’s two times from 23 instances; and finally for Ogando, it’s twice from 18 instances.
Simply converting into percentages, Holland does come out ‘on top’ by this measure, having allowed four or more earned runs in an inning 15.6% of the time. Ogando is next, at 11.1%. Of course, it shouldn’t be surprising that nobody in the rotation is giving up big innings as constantly as Luis Mendoza, but I wouldn’t say that the 15.6% put up by Holland is all too low either. Of course, without league-wide tracking of such a thing, it’s rather difficult to say whether or not Holland is particularly culpable of giving up a big inning – but he definitely is so, in the context of this particular staff.
Quickly, if the criteria for "big inning" is reduced to giving up three or more earned runs, things change just a little – it becomes 8/32 for Holland; 1/38 for Ceej; 6/32 for Colby; 6/23 from Harrison and 4/18 from Ogando. With this, Harrison assumes the title of "pitcher who most often lets things get out of hand, to the anger of LSB". Even so, giving up three earned runs in an inning is less annoying than giving up four or five.
Part 2 – Random statistical scattershot on Derek Holland’s runners on base/RISP splits:
Examining the available splits on Fangraphs, or at least splits which I am capable of producring (and of course, anyone who reads this can easily peruse the information by themselves), there are few things which I found worth noting regarding how Holland pitches with men on or with men in scoring position, compared to how he pitches with the bases empty. Make what you want out of these particular observations, in terms of connecting it to Holland’s slightly-greater tendency to give up 4+ earned runs in any given inning.
Looking at his general FIP splits, Holland actually has a better FIP when pitching with men on than with bases empty (4.25 to 4.03); however, his FIP when pitching with men in scoring position is worse, at 4.93 compared to his bases-empty FIP of 4.25.
Of course, the differences in FIP between splits are formed by the difference in Holland’s strikeout and walk rates. When the bases are empty, he strikes out 7.57 hitters per nine innings and walks 4.43 hitters per nine innings. The strikeout rate falls to 6.05 per 9 with men on, and 5.65 with men in scoring position. Likewise, his walk rate falls to 2.07 per nine innings with men on, and 2.83/9 with men in scoring position. As a result, his K/BB ratio improves from 1.71 when bases are empty, to a pretty good 2.93 with runners on and reasonable at 2.00 with runners in scoring position.
From Baseball Reference, league average K% falls from 19% with bases empty, to 17.6% with men on and 18.1% with runners in scoring position. Meanwhile league average BB% is at 7.6% with bases empty, increasing to 9% with runners on and 11% with runners in scoring position.
So compared to the rest of the league, Holland actually does a nice job in terms of improving his K/BB when it comes to pitching with men on/in scoring position.
His lowered strikeout rate (meaning more balls put in play), however, coupled with the fact that his BABIP allowed in situations with runners on/in scoring position is not lower than his bases-empty BABIP, does mean that he has allowed a higher batting average against – at .281 – with runners on base or in scoring position, than he has when the bases are empty (.266).
Holland’s GB/FB profile also changes when he pitches with runners on base/in scoring position. With bases empty, it sits at 1.61 (with a GB% of 49.8% and a FB% of 30.9%). With men on, it falls to 1.07 (GB% drops to 40.6% and FB% increases to 37.8%); with men in scoring position, it falls a little less precipitously but still by a fair bit down to 1.14 (GB% of 43%; FB% of 37.6%). There are also changes in HR/FB% in conjunction with this FB% increase, rising from 9.4% with bases empty, to 11.8% with men on base, and 14.3% with men in scoring position (which might explain the jump in FIP).
What’s the result of an increase in batting average allowed and HR/FB% (which might or might not be luck) with men on/in scorer’s position? BR’s splits have Derek Holland’s OPS allowed with men on at .792; his OPS allowed with men in scoring position is at .829. These two figures are well above his .727 OPS allowed with the bases empty. League-average for these three splits goes as follows - .723/.722/.701. Not a big difference, I know, but remember that it’s based off a very large sample, which probably makes it okay to suppose that the given gap is enough to explain that pitchers in general tend to allow a higher OPS with men on or men in scoring position.
What does all of this mean? Holland’s results are certainly not as good when runners are on or in scoring position – which could mean that he does actually dig himself into holes which he then struggles to get out of. His K/9 and BB/9 splits suggest that with runners on or in scoring position, he pounds the zone more often. As a result of this approach, he significantly lowers his tendency to issue walks in these situations, but probably also throws fewer quality strikes and more meatballs which are more likely to turn into homeruns.
I probably need to explore these three particular splits for the other pitchers on the staff, but in general it seems that Holland does worse with runners on or in scoring position – although that might apply to most starting pitchers out there, since they have to deal with things like pitching in the stretch, or checking runners, and so on. It’s possible that if there could be pitching splits on how a pitcher performs after X earned runs have been allowed in a particular inning, there could be something else worth finding.
Part 3 – more graphs:
This stuff is more just a visual look at how the five main Rangers starters have performed from start-to-start, in terms of earned runs allowed. I marked a straight black line where the Y-axis is at three – I designated three earned runs to be the line between having a ‘good’ start (<3 earned runs allowed) and having a ‘bad’ start (>4 earned runs allowed).
Make what you want out of these graphs, but even here I still find Holland's things annoying.
Derek Holland does have a tendency to give up a lot of earned runs in an inning - at least moreso than anybody else on the pitching staff, if not in comparison to the rest of the league's starters.
This tendency could possibly be explained by the fact that Holland doesn't do as well for himself when men are on base, but then again maybe that's how all starting pitchers go about their business.
C.J. Wilson is really really good and it would suck for the Rangers to lose him in free agency.
I have very limited experienced in finding the splits I need/Fangraphs needs more splits. I'm leaning towards the former being true.