Marlins not fooled by Nolasco’s ERA

Our organization gets a lot of props, some of them well placed, and others not so much. I have a feeling like our organization may be behind on certain things that are prevalent in baseball front offices and that are helping small market teams win. I have my ideas of what the Marlins do and do not do well, and it is not as rosy a view as one may like.

Nevertheless, hearing these words from Joe Frisaro made me feel pretty good about our organization.

After 2010, Nolasco will be entering his third and final season of arbitration. The Marlins have already spoken with Nolasco’s agent, Matt Sosnick, who also represents Johnson, about doing a multiyear deal next offseason.

The condition is if Nolasco has another solid season, the squad would be receptive to locking up the right-hander long-term.

That has to be some promising news coming from the Marlins’ camp, especially given the strange year Ricky Nolasco had last season.

In some organizations, a situation in which a young starter posted an ERA in the 9′s early in the season, even one who eventually finished the year in the 5′s, would not be close to meriting a multi-year contract. This would especially be so if that pitcher had only one good season worth of track record. But in the case of Nolasco, as we all know, the poor season that was 2009 was not as poor as one might imagine.

Ricky Nolasco K% UIBB% HR/FB% (FG) FIP ERA
2008 21.6% 4.2% 10.6% 3.77 3.52
2009 25.1% 4.8% 11.0% 3.35 5.09

Of course, you could have seen much of this at FanGraphs, but it bears repeating. The things that Nolasco had the most control over were very stable through 2008 and 2009. In fact, his 2009 FIP was better than his 2008, indicating (if you buy the premise of FIP, which I do, mostly), that he improved in 2009.

Can that be said? Not necessarily. There are things that go into pitching that are not included in FIP, of course. I was worried that Nolasco might have an issue pitching out of the stretch, and he certainly performed worse this season out of it (more research on that as I continue to figure out my SPSS statistics package). But I can safely say that Nolasco did have what appears to be an insane bout of bad luck during the 2009 season, and barring an equally poor run of bad luck (or something inherently wrong with the way he pitches out of the stretch or with runners on), Nolasco is due for a more typical performance in 2010.

And that apparently is what he’ll need to earn a contract. The question in my mind is in how the team is determining what constitutes “performance.” If the team is measuring the same things smart Marlins fans are seeing, including (I hope) defense independence, than a solid performance like CHONE is projecting (160 innings of 4.16 ERA, with an underlying FIP of 3.68) should earn him a three- or four-year deal if he is interested. However, if the Marlins find that performance as merely average and not worth the price of admission due to ERA or some other peripheral stat while not digging underneath to view his performance, I will be sorely disappointed.

At this point, I don’t know what the team is looking at. When the club decides to keep Jorge Cantu for $6M, it makes me think that they are not valuing the right stats and numbers. At the same time, I believe that pitching is something that can be better scouted and coached, and that if the Marlins scouts see that Nolasco is still pitching well, I should hope that they can ignore the results and focus on those good inputs. It’s clear to most of us fans that after Nolasco came back from his Triple-A stint, regression to the mean hit in a big way and he began pitching much better. Perhaps the scouts and the organization saw that as well and thus considered him someone worthy of multiple years.

Or maybe not. Let’s just hope the team has a department where stat guys work and are digging through more than just’s stats page. Heck, it doesn’t even need to be FanGraphs. Baseball-Reference will do, folks.

Next Marlins Game View full schedule »

Tags: Miami Marlins Ricky Nolasco

  • Pingback: / archive » Overnight Hot Stove Notes: Phillies, Marlins, Blue Jays, Mets, Mariners, more

  • jrhana

    Interesting take on Nolasco.

    Same take I have by just watching him pitch. I imagine the Marlins are aware of sabermetrics but also watched him in the game against Atlanta.

    Came across a couple of FIP references

    Apparently for fip all that matters is Homers, Ks and walks

    I would be more impressed if they would use a park specific constant rather than a MLB constant. Some parks are much easier to hit a home run in than others.

    And fip pretty much discounts the value of a ground ball pitcher.

    So really just another number to look at no better and no worse than the old ones IMO.

    • Michael Jong


      FIP assumes a measure of defense independence, which is mostly correct (pitchers do have some control over BABIP, but the spread in that talent appears to be very small). So FIP only calculates run totals for HR, K, and BB+HBP. However, balls in play like ground balls are still in the calculation. It’s just assumed that a pitcher is at the mercy of his defense, and as a result, the net run total for a ball in play is 0 runs above average. In comparison, a home run is worth 1.4 runs, a walk is worth .3 runs, and a strikeout is worth -.27 runs.

      Good point regarding park-based constants. The solution to this is that, while they use a league constant, you can easily park-adjust FIP using park factors, so parks don’t become a big problem.

      The reason why we consider FIP better than ERA is because of its defense independence. ERA has other factors, some in control of the pitcher (pitching with runners on vs. out of the stretch) and some not in control (defensive play). FIP takes a lot of those things out, but I personally prefer my pitching stats defense independent. Also, FIP is generally a better predictor of future performance than ERA, as you may expect.

      Finally, if you’re interested in a defense-independent stat that does include batted balls like ground balls (but not necessarily singles/doubles/triples), try tRA, which is a linear estimator like FIP but includes grounders, flyballs/line drives, and popups in addition to K/BB/HR.

      Hope that gave some more insight on FIP. As for Nolasco, well, he has nasty stuff, as many fans can attest to. I’m happy the Marlins office can see that as well. We’ll see how the extension talk turns out.

  • jrhana

    Thanks for the detailed and cordial reply. I see that I mistakenly left out one of the references I found.

    (13 * HR + 3 * [BB + HBP - IBB] – 2 * K)/(IP) + 3.20,206286

    So I would point that it really has nothing to do with run totals. The only runs that come in are those from the homeruns.

    I see the sabermetrics seems to be dominating the baseball threads so (with misgivings) I will be one by one studying them and making up my own mind as to how valid they are (again on a one by one basis).

    I still feel there is a place for the intelligent observer of baseball. You would say that means it’s all subjective, but I would say that these formulas are themselves derived with a lot of subjectivity. I am sure you have been over this argument a million times so no need to dwell on it forever.In addition statements like Also, FIP is generally a better predictor of future performance than ERA, as you may expect. rmeain to be proven IMO.

    That said I plan to spend more time over here as you seem like a true Marlins fan and a serious student of the game. And I intend to take close looks at your formulas.

    At least until the season starts when I will be spending a huge amount of time actually attending the games.

    Now I will take a look at tRA.

    And oh yes one comment is that I hope the Marlins continue to be extremely selective with any long term contracts.

    Also if you get a chance, please EMail me, I have an idea for next season.

    • Michael Jong


      The numbers listed are all based on linear weights, or run values per event. To get to the coefficients shown, the numbers are multiplied by 9 (then divided by innings pitched, kind of like ERA). The run value totals are based on the average change of run expectancy found in actual game data. In other words, drawing a walk on average increases the number of runs scored by the end of the inning by 0.3 runs.

      Here’s a great primer on FIP, written by Kincaid, very explanatory.

      The research on FIP’s predictive capability has been done, I just didn’t want to put too many links on the page. Here’s one such article written a respected colleague, Colin Wyers.

      That being said, you are correct to call me out on that summary claim. The article shows that the RMSE for FIP is lower than ERA in predicting future ERA, but not by a whole lot (0.2 runs/9 better, significant, but not amazing). There are reasons for that, particularly tied to things that FIP strips away that may be more in control of the pitcher. Whether you choose to use it is, of course, on you.

      Thanks for the kind words on the blog. I am attempting to be thorough with my research (when I do it) and commentary, and I think the best way to do that is to embrace the current saber-techniques while at the same time keeping an eye out for new things and always questioning. I don’t generally like to use things without reading about them thoroughly as well.

  • jrhana

    Michael I have been spending the morning reviewing various sabermetric formulas including tra and wOBP.

    You may not like this one, but I did come across wpa. Maybe you have discussed it recently so I hope I am not wasting time here. Like all of these sabermetric formulations (in fact like all baseball statistics new and old) it certainly must have its limitations. Maybe it just does describe the past without predicting the future.,209597

    And it does provide objective confirmation about what we have been saying about Cantu versus Uggla in the clutch.

    • Michael Jong


      I’m glad you’ve been checking out some of the readings. I recommend all of Alex’ “Everything You Always Wanted To Know About” series, they’re all great pieces.

      On WPA, it’s a great stat at what it does. Basically, it tells a story of how games unfold, and it gives credit to players based on the events that occur when they come up. WPA is based on run expectancies (we talked about that yesterday) and how run scoring would affect how often a game is won.

      You are right in that WPA would show that Cantu was more “clutch” than Uggla last season. I absolutely agree with you on that. My premise is not that he wasn’t more clutch in the past, but rather that we can’t expect him to be more clutch in the future. WPA is a great storytelling stat, but one thing it does not do well is predict the future. So much of what goes into WPA is dependent upon situations that individual players have no control over (who’s on base, when they come up, what the score is when they do come up), so predicting that similar outcomes will occur in the following season would be impractical. As I mentioned before on FishStripes, Uggla and Cantu saw a vast difference in runners on between 2008 and 2009, and that factor alone would affect the WPA of Uggla and Cantu. And that’s just one factor, not counting position of runners, score, and inning.

      Having read reviews of the research on clutch performance (and the latest research from The Book), clutch talent seems to small to predict, so we just can’t predict what kind of effect we’d see between Cantu and Uggla knowing what we know.

      If you’d like to check out some stuff on the clutch research in sabermetrics, I do have some recommendations. Tom Tango (who works occasionally as a consultant for teams, currently consulting for both the Mariners and Blue Jays) did this study in 2008 called the Great Clutch Project. He also discusses the recent debate here, where he cites work done by Andy Dolphin in their book, The Book, Playing the Percentages in Baseball (great read). Finally, Phil Birnbaum goes through Dolphin’s work and explains and critiques it here (search for “Doesn’t “The Book” study pretty much settle the clutch hitting question?”).

  • jrhana

    I see I should perhaps have been using wpa/li. This happens to give the way I would have ranked these guys subjectively. Ramirez, Cantu, Coghlan, Uggla

    I suspect that often the statistic is cherry picked for effect (which of course is human nature).

    I have been looking for the exact formula for both wpi and li to no avail. All I find are vague subjective descriptions.

    • Michael Jong


      Yes, WPA/LI would have been better. This is basically game-state specific linear weights. What you do to calculate it is basically what it implies; you divide the WPA for all of the events by the leverage index (LI) of the event. As a result, you get each event’s worth if the leverage (the relative importance of the event) was considered average.

      This sort of implies what you’ve said before, that Uggla’s major events have come in low-leverage situations. In those situations, the value of a HR, a single, or a walk have very little relative difference, as they will not likely change the odds of winning. If Uggla hits a homer in the ninth with a 6-2 lead, it’s worth less than the average home run. Similarly, a walk and a home run in a bottom 9th, tied game, bases loaded situation is worth the same, so players don’t get extra credit for the walk off grand slam (even though homers in general are worth more).

      Most of the time, your WPA/LI and linear weights runs (the wRAA from FanGraphs) are going to be very similar; note how Uggla’s 1.26 WPA/LI wins are close to his 14.2 wRAA (1.42 wins using the conventional conversion of runs/win). Cantu’s is really far off, suggesting that maybe he faced far fewer low leverage situations (his average LI was 1.12, average is 1). I’ll look into what this means.

      Here’s the stat’s innovator, Tom Tango, on the topic. Make sure you check out the comments as well. He says that after a few seasons, he’d trust WPA/LI, but at this point comparing Uggla and Cantu would probably be best using context-neutral stats like linear weights.

  • Mark

    Even if they can’t pay him the money they should get him to sign something because he would be able to reel in something very good in return. The FIP was terrific last year and he’s a power pitcher on top of that. In the second half of last season he was better than Josh Johnson and it wouldn’t surprise me if he was the true ace of the staff this season.

    • Michael Jong


      You’re right on that point. Locking up Nolasco in a three-year deal would buy out two years of free agency if it were done after 2010. At that point, he’d be holding a lot of value if the two free agent years are somewhat below market value as likely expected. It definitely would be good idea, but I think the Marlins want to see a solid performance first, unlike the so-called inconsistency of 2009. I hope they know what solid is (crosses fingers).