Jason Marquis, Expectations, and Reality by Neil Weinberg July 19, 2013 On the surface, Jason Marquis looks like he’s having a good season. He’s 9-4 with a 3.77 ERA for the San Diego Padres and if you’re looking at how well his team performs collectively during his starts and how effective he has been at limiting runs, you might even say Jason Marquis is having one of the best seasons of his career. If you look more closely, however, he’s actually having one of the worst seasons in baseball history. That’s a crazy juxtaposition. We’re used to evaluating players by advanced statistics like FIP, xFIP, WAR, and others, but it’s pretty rare that they tell us something totally different from the basic descriptive stats. Usually we look at a player’s FIP and think their ERA might be due for some regression. We don’t often look at a player’s FIP only to find that their ERA is propped up with toothpicks and Scotch tape. Such is the case with Jason Marquis. His numbers this season are actually quite remarkable. He’s 48th among qualified starters in ERA, good for an ERA- of 104, which is just a bit below average. If you turn to FIP, he’s dead last at 5.70, good for a 156 FIP-. His xFIP is better, but it still remains fourth-worst in MLB at 4.77 to go along with an MLB-worst 125 xFIP-. No qualifying starter’s ERA is outperforming their FIP as much as Marquis’ is this season. Jeff Locke and Jeremy Guthrie are in the conversation, but Marquis is decidedly ahead. Perhaps even more noteworthy is that Jason Marquis’ 156 FIP- is the second-worst number for a qualifying starter since 1901, and the record is within reach at 159. Marquis is essentially having an all-time worst season in terms of strikeout rate, walk rate, and home-run rate when adjusting for park and league average. Yet he’s allowing a very average number of runs in a very respectable number of innings. Only two players in history have had their FIP- outperform their ERA- by more than Marquis’ difference of 52 and they both played before 1910. I don’t mean to belabor Marquis’ 2013 season among the all-time clunkers in MLB history, but rather to simply set the stage for the remainder of this analysis. Marquis is doing a pretty decent job preventing runs, but is doing a terrible job at the aspects of preventing runs a pitcher can most control. What’s going on here? First, let’s consider the Padres defense. By DRS, UZR, and UZR/150, the Padres rank between 15th and 21st in MLB this season. They aren’t a terrible defensive team, but this doesn’t appear to be a club that should systematically deflate their pitchers’ ERA. It’s possible that they are playing amazing behind Marquis and not for everyone else, but that seems unlikely. If you consider the Padres starters as a whole, their ERA is higher than their FIP and have individual starters on both sides of the divide. Additionally, it doesn’t appear as if any of this can be explained by the GB/FB ratio of each pitcher, which might have pointed to a particular aspect of the Padres defense. Sometimes it’s just about the situation, but on the surface it doesn’t look like this is a good explanation either. Marquis allows a .318 wOBA with the bases empty to go with a .380 wOBA with men on and a .306 wOBA with men in scoring position. He’s more or less the same pitcher with men in scoring position as he is with no one on so pitching from the windup versus the stretch isn’t the answer. Let’s look at each base situation. Split PA BA OBP SLG — 294 0.225 0.320 0.395 1_ _ 78 0.422 0.500 0.656 _ 2 _ 32 0.269 0.406 0.385 _ _ 3 5 0.200 0.200 0.200 1 2 _ 39 0.188 0.316 0.344 1 _ 3 15 0.222 0.429 0.556 _ 2 3 10 0.000 0.500 0.000 123 18 0.143 0.222 0.214 Obviously, some of the samples are really small, but notice how much worse Marquis is when a runner is on first base, but not also on second base. Could this have something to do with holding the runner? A couple of possibilities spring to mind. One, Marquis is distracted by the baserunner. Two, having the first baseman holding the runner creates a hole where Marquis often allows hits. Three, the presence of the baserunner and location of the runner cause Marquis to pitch differently in order to avoid the hole on the right side, resulting in pitches that get smashed. I’m not sure if any or all of these are factors, but they are possible factors. If something like this is the case, it’s possible that Marquis isn’t actually as bad as his FIP tells us overall, but rather just really terrible in certain situations and reasonably average most of the time. Some of this timing argument is dispelled if we consider that he’s actually allowing a higher percentage of his home runs with men on base (48%) than league average (40%) so his high walk rate and high home-run rate should be costing him dearly. But they are not. Marquis is pitching like he should allow close to 6 runs per 9 innings but he’s allowing fewer than 4. His strikeout rate is 12th-worst in baseball at 14.7% and his walk rate is easily the worst at 13.2%. Only 1o pitchers have a higher HR/9. Yet he’s right around league average in ERA. Metrics like SIERA don’t rate him any better, as he comes in 2nd-worst at 5.11. He shouldn’t be doing this well. He’s leaving runners on base like he’s Felix Hernandez, but walking guys like he’s Carlos Marmol, giving up home runs like Jose Valverde, and only striking guys out like he’s Bronson Arroyo. He’s getting a lot of ground balls, but he has a low BABIP against. If you look at Lucas Harrell and Jason Marquis, most of the stat line is nearly identical. Name IP K/9 BB/9 HR/9 Jason Marquis 112.1 5.77 5.21 1.44 Lucas Harrell 108.1 5.57 4.74 1.25 Name BABIP LOB% GB% HR/FB Jason Marquis 0.256 79.70% 53.20% 19.60% Lucas Harrell 0.304 70.40% 52.70% 16.00% Name ERA FIP xFIP WAR Jason Marquis 3.77 5.70 4.77 -1.4 Lucas Harrell 5.07 5.37 4.78 -0.3 If anything, based on K/9, BB/9, and HR/9, Harrell should be doing better. But somehow, Marquis is getting a much lower BABIP and a higher LOB% despite getting pretty much the same number of ground balls and having a worse HR/FB rate. They have essentially identical xFIP and Marquis has a worse FIP. The fact that Harrell has a 5.07 ERA and Marquis has a 3.77 ERA defies understanding (the story is the same with park-adjusted numbers). This is likely just one of those small sample size mirage/miracles. Marquis has lost a tick on his fastball this year and his changeup is acting more like a splitter according to Pitch F/X, but nothing appears fundamentally different that would allow him to actually sustain this low BABIP (last year it was .311). Perhaps baseball fans who watch the Padres more regularly can offer some insight into what exactly is the driving force behind his low BABIP this season. If you’re someone who likes to look at FIP, you’re looking at one of the worst seasons in baseball history. If you’re someone who cares more about overall run prevention, you’re looking at an average year. Granted, it’s not uncommon for a player to over or under perform their peripherals over 100 innings, but it is amazing how dramatically it is happening for Marquis. It’s not unusual for BABIP to drive over- and under-performances by 20 or so points in on the ERA/FIP- scale, but what Marquis is doing is beyond the typical variation. For every qualifying season since 1901, the mean FIP-/ERA- differential is around 2.3 and the standard deviation is about 13.8. Marquis’ 2013 season is 3.6 standard deviations above the mean. (For just 2013 those numbers are a mean of 1.3, SD of 17.6, and Marquis is 2.8 SD above the mean) The simple takeaway of this entire exercise is this. Jason Marquis is over-performing his peripherals this season and there isn’t a clear explanation for why this might be the case other than standard variation in BABIP. It’s a perfectly reasonably explanation. Marquis is getting some good fortune regarding where baseballs have been hit during key moments that have allowed his ERA to stay relatively low despite the fact that based on his other numbers it should be much higher. That happens. What is so amazing about this is the degree to which it is happening. We’re all open to the idea that some players will over- and under-perform, but Marquis is over-performing at such a rare level. He’s in the top 0.25% of all over-performances when comparing ERA- and FIP-, which are statistics that control for league average and park effects. When you strip away the context, Jason Marquis’ 2013 season stands out as the third-biggest over-performance in the last 113 years, which includes more than 8,000 individual seasons. Everything I know about baseball tells me Jason Marquis won’t maintain this ERA if he maintains these K, BB, and HR levels, but part of me is really hoping that he does. I like when things make sense and can be easily explained, but sometimes it’s a lot of fun to watch a player defy the odds for no other reason than that the Gods of probability have chosen that player to be the exception that proves the rule. Jason Marquis and Padres fans are hoping he can keep it up. Anything can happen, but as we should note, it usually doesn’t.