FanGraphs Baseball


RSS feed for comments on this post.

  1. How is game score calculated? Were there any other game-to-game statistics that you considered?

    Comment by Larry — July 19, 2011 @ 2:10 pm

  2. Very cool. What is Burnett’s standard deviation? Who has the highest and lowest standard deviation? I will make very crude guesses off the top of my head of Halladay and Wakefield.

    Comment by Max — July 19, 2011 @ 2:10 pm

  3. Here is a good Game Score explanation by the best sportswriter on the planet.

    Game Score calculation, from the article:
    • Start with 50 points.
    • Add a point for each out, and two more for each inning completed after the fourth.
    • Add one point for each strikeout.
    • Take away two points for each hit, four points for each earned run, two points for each unearned run and one point for each walk.

    Comment by Max — July 19, 2011 @ 2:14 pm

  4. I’m not so sure you should use “average game score.” The average is biased towards outliers, and since we are talking about wild inconsistency, he is obviously subject to those. What happens if you use “median game score”?

    Comment by Sitting Curveball — July 19, 2011 @ 2:36 pm

  5. You need to take the ratio of the SD to the average gamescore. If Ubaldo has an average GS of 50 and his SD is 17, that’s huge compared to someone with an average GS of 75 and an SD of 17. You can’t look just at SD alone.

    Those numbers are made up, but you get the point. Cameron misses the glaringly obvious math, in a post about math.

    Comment by Telo — July 19, 2011 @ 2:45 pm

  6. While this is true, doesn’t using the six guys closest to Jimenez in WAR help reduce the effect?

    Comment by Max — July 19, 2011 @ 2:47 pm

  7. Yes and no. GS is based heavily in ER, where fWAR completely ignores it. They are pretty different. Plus, the idea is that you just need to see those numbers, even if they are very similar. Just seeing the SD is like showing us half of the picture.

    Comment by Telo — July 19, 2011 @ 2:49 pm

  8. Great Article, I was wondering if there is a type of GameScore for relief pitchers, as it would be interesting to see. Has anyone heard of one and if so how does it work?

    Comment by duncanbishop1 — July 19, 2011 @ 2:49 pm

  9. No, we’re not getting the whole picture. But you don’t actually think the differences in the average game scores of those seven pitchers are going to be meaningful, do you?

    Comment by Ben Hall — July 19, 2011 @ 2:59 pm

  10. When Burnett is criticized for being inconsistent, it’s really a within game inconsistency. As a Jays fan, I watched it for a couple of good but maddening years. Burnett more than anyone I can remember watching seems to pitch really well and still give up 4 runs a start because of one bad inning when suddenly all the hits bunch together. Lots of streaks where he looks great, and then it falls apart for 20 pitches, and then right back to being awesome.

    Completely dominant and yet gives up 3-5 runs, that’s what made Burnett so frustrating despite being good. I’m guessing Jimenez is inspiring similar frustrations. How to measure – % of innings runs are scored against, compare to pitchers with similar ERAs – do runs tend to bunch for Jimenez, or is it just perception? Maybe it wasn’t even reality for Burnett in his Jays days, but it did seem that way…

    Comment by test — July 19, 2011 @ 3:03 pm

  11. “Inconsistent” is the polite word that people use to describe a player who they think is supposed to perform at a high level, who in actuality fails to perform at that high level.

    Game-to-game performance is not the issue.

    Ubaldo is “inconsistent” because for whatever reason the first half of 2010 is the image that has stuck. So when he pitches less well (even though he couldn’t possibly maintain a 1.15 ERA) people are disappointed, yet insistent that somehow he’s going to get back to the promised land one day.

    No one is out there saying Ryan Roberts in “inconsistent” simply because he had one pretty good half of a season … because that’s not the image that stuck for him.

    Comment by David — July 19, 2011 @ 3:07 pm

  12. The guy pitches in an amusement park, where variance in performance is magnified.

    Move him to a more forgiving park, and he’d show similar variance to most top pitchers.

    Comment by puffy — July 19, 2011 @ 3:08 pm

  13. An interesting post, but a valid point by Telo. The other thing Dave could do is tie those standard deviations to a known statistical distribution, probably chi-squared. Stata or some other canned software program can do this really easily. The result would tell you whether the differences are significant or could be explained by chance. Of course, since you’re posting this content for us for free, I’m not complaining, just offering suggestions …

    Comment by Taylor — July 19, 2011 @ 3:09 pm

  14. Well, AJ’s standard deviations shouldn’t be too big now that he seems to suck every start…

    How about month to month though? What if Ubaldo goes through stretches at the top of his range and then again at the bottom? say five to seven great starts and then five to seven mediocre or worse starts. This is a type of inconsistency that is not measured in the article above. I have no idea how this would come out, but it might say something about inconsistency.

    Comment by mark — July 19, 2011 @ 3:35 pm

  15. Exactly. There is this phenomenon in the world of sports, and it goes like this:

    – Man, I bowled a 175, but man if I would have just converted those three spares I would have had a 225.
    – Man, I hit .300 this year but I was 25 for 50 in half my games and only 5 for 50 in the other. If I’d have only gone 15 for 50 in the bad games I’d have hit .400!
    – I shot .400 from three-point range this season, but that counts two games where I was 0-for-5 and 0-for-6. If you take them out I would have shot .500!
    – Man, the star running back had five games over 100 yards rushing, but if he’d averaged 5.5 yards per carry in all games, and not just the games he got 100 yards in, he would have had over 1,000 yards on the year!

    There is a tendency in observers (and athletes) to act as though the whole of the set of their data should be shorn of the results they don’t like. You know, so if Jimenez gives up 5 runs in 4 innings, he’s “off” that day, but if he gives up 1 run in 8 innings, that represents his true ability.

    Comment by J.E. — July 19, 2011 @ 3:35 pm

  16. Standard deviation is sequence agnostic while the perception of inconsistency (for better or worse) is not sequence agnostic.

    If he traded some of the bad 2011 games into his early 2010 he would be seen as a more consistent pitcher, yet the standard deviation would remain the same.

    Comment by Nick44 — July 19, 2011 @ 3:36 pm

  17. this.

    Comment by joshcohen — July 19, 2011 @ 3:44 pm

  18. You should try complaining. It’s moderately more entertaining. Everyone wants a nemesis, right? I’m actually paid by FG to provide a foil to the heroes in beige and green and to inflate comment totals.

    Comment by Telo — July 19, 2011 @ 3:45 pm

  19. Was wondering that, too (re: Burnett). Since he stunk so profoundly last season, I can’t help but think that his standard deviation would be pretty low actually. He was consistently awful.

    Comment by chuckb — July 19, 2011 @ 3:45 pm

  20. I agree with the premise that Ubaldo isn’t especially inconsistent as Olney would have you believe; it’s just that this argument of that thesis is half assed and sloppy at best. As most things Cameron, they leave much to be desired in the way of mathematical and logical rigorousness.

    I have no idea whether or not the difference between the GS of those pitchers would be sig, since I don’t see them in the article! Plus, why use those just those pitchers? Is that supposed to a big enough sample to convince us?

    Comment by Telo — July 19, 2011 @ 3:52 pm

  21. Wouldn’t that mean we would want to use average? Since we are trying to see if his numbers are influenced by outliers (inconsistent starts).

    Comment by Frank Costanza — July 19, 2011 @ 3:53 pm

  22. Also a very good point.

    Comment by Telo — July 19, 2011 @ 3:59 pm

  23. On this note, instead of separating performance by starts, what about by months, or half seasons. Inconsistencies can occur in different time periods. I’m guessing there’s not a lot of predictive information in “past inconsistency” anyway, but at the very least, Ubaldo might get into month or two-month long “grooves” moreso than other pitchers. Versus just start to start variation.

    Consider the following 6-game patterns by ER:


    Both sets have the same standard deviation, but they are different “forms” of inconsistency i would argue.

    Comment by Large Numbers — July 19, 2011 @ 4:03 pm

  24. But his far outliers skew the average away from the “middle” and more towards the outliers. This would reduce the deviation, even though there is less “consistency”.

    Comment by Sitting Curveball — July 19, 2011 @ 4:04 pm

  25. Yes, I agree with the comments talking about sequence. You’ll often see a pitcher hit the skids for a few consecutive starts, and this is less likely to make him appear to be inconsistent than a pitcher who mixes in poor starts between good ones. I don’t think we can measure game-to-game consistency very well without looking at the games in sequence.

    Another issue is that game score is not taking into account the widely variable quality of opponents. Whether you’re pitching against the Padres in PetCo or against the Red Sox in Fenway Park is likely to have a large effect on your game score.

    Comment by Jon — July 19, 2011 @ 4:07 pm

  26. I partially agree. His 14 straight great starts in early 2010 shows consistency. I would say that the most inconsistent pitcher imaginable would be one who can’t string together two similar performances to save his life (although I suppose that would be a different form of consistency in and of itself). A better (yet imperfect) measure of (in)consistency would be something like RMS of game score differences between consecutive starts.

    Comment by novaether — July 19, 2011 @ 4:19 pm

  27. Don’t forget to contextualize numbers as well, if a player returning from injury or illness takes a while to return to form it isn’t the same as random ineffectiveness.

    Jimenez was having puss drained from his gripping point in the last weeks of ST and was visibly bleeding on opening day. He said that the thumb cut/infection hurt his arm strengthening regimen in spring training. The recent results would seem to indicate he is returning to form, as he had lost an average of 5MPH off of his four seamer in april and now it is closer to 3MPH for the season

    Comment by Gordon Brown — July 19, 2011 @ 4:49 pm

  28. Amen. He’s pitched pretty well in that park over his career, although this year he has really struggled in Coors. That happens. He may not be quite as sharp as he was at the beginning of 2010, but that happens as well.

    I’d take him in a heartbeat, especially given his salary and age. Stick him in almost any other park half the time and perception would change quite a bit.

    Comment by razor — July 19, 2011 @ 4:49 pm

  29. from my quick excel calculations from 2010 to his last start his avg game score is 48.58

    the standard deviation is 17.73

    Comment by Jon — July 19, 2011 @ 4:56 pm

  30. I see, thank you. Now why use game score if you are comparing Ubaldo to his WAR peers? The two do not seem to mesh. Game score looks kind of like how you score pitchers in fantasy points leagues… Instead of this method wouldn’t it be better to examine his game-to-game K%, BB%, GB%, FB%, LD%, O and Z swing%, HR/FB%, BABIP, and swinging strike %? Of course also noting that each is based on a very small sample and occurs against different lineups in different parks.

    I don’t really get how this would be the preferred method to measure consistency.

    Comment by Larry — July 19, 2011 @ 4:57 pm

  31. I’m not sure how if he’d had poor games sprinkled in among the dominant games of last year, that would make him more consistent. It sounds right that Burnett and Jimenez swing wildly from end to end, but it doesn’t fit the facts. They are both streaky more than they are game-to-game inconsistent.

    Additionally, to an observer nothing is indicative of “badness” like wildness, because to witness a pitcher who has lost his command is to be unable to believe the outcomes are in the pitcher’s control. In the cases of both Burnett and Jimenez they have had stretches of games where they’ve walked a lot of batters, and this prejudices observers.

    Comment by J.E. — July 19, 2011 @ 4:58 pm

  32. And if you were wondering AJs highest gamescore was 79 (8ip 0 runs 1 bb 4k)

    and his lowest was 10 (3.1ip 8er)

    Comment by Jon — July 19, 2011 @ 4:58 pm

  33. Is Prince Fielder a threat to your potato chips?

    Comment by Bronnt — July 19, 2011 @ 5:22 pm

  34. Your bowling analogy presumes that excuses need to be made for players that have an inability to be consistently good. It has nothing to do with shooting a 175, and saying the bowler is better because he should have picked up 3 more “easy” spares…

    The better bowling analogy would be to say that Ubaldo shoots with a 220 average, and has a knack for rolling streaks in the 280+ range more than most typical bowlers that carry a 220 average. His issue is that his excellence is countered with a knack of shooting in the 170 range more than most guys who carry a 220 average. Does that make him a typical 220 talent bowler? I don’t know because it’s not a clear definition of the kind of ability he has, as you look at another guy with 220 average who never shoots below 190 and has never rolled over 260. What about the bowler that has never shot anything other than a 220? They are all still very different kinds of bowlers. Factor in age, experience, etc… and you start to get lots of varying opinions about what that 220 really means (past, present, and future).

    1. Ubaldo isn’t old yet. Pitchers can get worse, but pitchers can get better, and in a game where most evaluators like to wager on a player’s ceiling over their floor, I think it’s perfectly reasonable to find opinions that believe getting closer to the ceiling is still possible, which is why many continue to wait for Ubaldo to finally prove he’s the better Ubaldo version any time he steps on the mound. At 27, Ubaldo is still perceived as having a bit of a “prospect-like” tag on him still.

    2. I think some will always wonder about that stigma of being a Colorado pitcher… What he could be like if it that variable wasn’t a major part of his game. It’s not to say that it should make a difference with his production, but I’m guessing it’s definitely on the minds of some.

    3. I myself wonder what someone like Don Cooper could do with a guy like Ubaldo. There’s still a lot of speculation as to what expectations could be, and while all of this is brutally intangible… who cares… Ubaldo might just be another AJ Burnett, but I wouldn’t yet say that he isn’t more capable of what the aggregate suggests.

    Comment by baty — July 19, 2011 @ 5:30 pm

  35. It would only really noticeably affect the data if he had a disproportionate number of excellent and terrible starts. Seems to make sense though.

    Comment by Pat — July 19, 2011 @ 5:48 pm

  36. The best sportswriter on the planet? Personally I think Posnanski’s act gets old real fast. He knows his stuff, but he serves it up with a heaping side of rambling, overly nostalgic schmaltz. Bit too much hero-worship and sepia-toned memories of his Rust Belt childhood for me.

    Comment by Arthur Xavier Corvelay — July 19, 2011 @ 5:49 pm

  37. Yeah, but the point was to figure out if he had a disproportionate amount of really good and terrible starts.

    Comment by DavidCEisen — July 19, 2011 @ 6:48 pm

  38. No the point was to see if he had a lot of starts which deviated from his average performance.

    Comment by PhillyPat — July 19, 2011 @ 7:27 pm

  39. I actually, strangely, think both sides are correct here – and I’ll try to explain it in a simple way

    2.50 FIP / 3.46 xFIP – March/April 2010
    2.64 FIP / 3.29 xFIP – May 2010
    4.09 FIP / 3.84 xFIP – June 2010
    4.21 FIP / 4.37 xFIP – July 2010
    2.54 FIP / 3.42 xFIP – August 2010
    3.06 FIP / 3.51 xFIP – Sept 2010

    Its seems extremely consistent to me for 2010 outside the June 23th to July 24th stretch which saw him fall apart for some reason – he even gave up 6-7 Runs in 4 of those 6 games. (otherwise, he gave up 3 or fewer in 24 of 27 starts, and as we can see from the FIP/xFIP, his ratios were obviously consistent). I personally don’t think a (literally) one month stretch warrants an “inconsistent player” label, and that one month is the only real outlier he provided last season.

    5.53 FIP / 4.48 xFIP – March/April 2010
    3.63 FIP / 4.33 xFIP – May 2010
    3.04 FIP / 3.20 xFIP – June 2010
    2.08 FIP / 2.32 xFIP – July 2010

    This shows a bit of inconsistency, but we know March to early May he was still hurting. Otherwise, he seems to be performing consistently with last years performance. Could be said to show a consistent “becoming healthy” trend.

    *I know “other then his unbelievable July” could also be added in there too (which ironically, Inconsistency is rarely mentioned when a player outperforms themselves every so often – people inconsistently only claim “inconsistency” when you occasionally do worse then they hope a couple times a season, but forget the word completely and instead love it when you outperform expectations over a small stretch…) But he also only has 3 July starts so far, so we don’t know what the end result will ultimately be here – just that it looks like its likely to follow the 2.50-3 FIP/3.25-3.75 xFIP model he has seemingly consistently provided the last 1.5 seasons.

    Overall, he seems extremely consistent to me in Month to Month performance, which matches the Year to Year course mentioned in the article.

    BUT, if you look at specific game logs I think you see a very different pitcher then shows up in the Month to Month or Year to Year lines

    6 IP – 8 H – 1 BB – 6 SO
    6 IP – 7 H – 3 BB – 7 SO
    9 IP – 0 H – 6 BB – 7 SO
    7 IP – 5 H – 2 BB – 5 SO
    6 IP – 2 H – 2 BB – 6 SO
    7 IP – 4 H – 2 BB – 13 SO
    That’s only the first 6 games of 2010 – and he follows a similar pattern the entire year. That is, hit&miss on Hits-Allowed, Walks and Strikeouts. And I know this will obviously vary for everyone, but it does seem a higher variance for Jimenez to me. On any evening it seems he might give up 2 hits, or 10. He might walk 1, or 5. He might K nearly 2 per inning, or maybe only 3 all day. (which can be seen in a game like the 5/26/2010 where he went 8 IP, 6 H, 1 BB and those 3 K)

    Now, he rarely gives up 3+ runs, and somehow manages to remain extremely consistent ratio-wise over 5-6 game stretches, but game to game the way he gets it done seems to vary quite a bit.

    Comment by Joe — July 19, 2011 @ 7:34 pm

  40. Analysis with game score?

    Game score?

    Comment by Garrett — July 19, 2011 @ 8:31 pm

  41. “His 14 straight great starts in early 2010 shows consistency.”

    IMO, this is paradoxically false and precisely what Dave is right to argue against using year-over-year xFIP data. Many posters above make this argument quite well.

    Comment by joshcohen — July 19, 2011 @ 10:17 pm

  42. SD/MD is a pretty good measure of relief effectiveness, as are WPA, and WPA/LI

    Comment by jim — July 19, 2011 @ 11:12 pm

  43. I think when they say, “He’s inconsistent,” what they mean is, “He got really lucky last year and he’s been really unlucky this year, but we don’t understand that.”

    Comment by Ari Collins — July 19, 2011 @ 11:17 pm

  44. Should probably read, “he’s been a bit unlucky his year.” He’s not underperforming his peripherals by all THAT much.

    Last year was crazy, though.

    Comment by Ari Collins — July 19, 2011 @ 11:17 pm

  45. This article does nothing to disprove his (in)consistency

    Comment by Wing — July 19, 2011 @ 11:29 pm

  46. Last year anyone who watched Ubaldo knows the turning point came in a cold rainy game against Toronto, no way that game should have been played at all but Ubaldo lost his fastball command and struggled for a good month/month and half to find it.

    Comment by Mr. wOBAto — July 20, 2011 @ 12:38 am

  47. burnett was 4-0 last year and then sucked, this year started 4-1 and then tanked.

    starting to become the norm for aj

    Comment by dave — July 20, 2011 @ 3:20 am

  48. I agree…isn’t game score just a made up statistic with no mathematical basis…

    Since the comment section seems to be going into mathematical handwringing….Does game score have a normal distribution?

    Comment by wahooo — July 20, 2011 @ 6:32 am

  49. These are the articles that are the reason I come here…great stuff!

    Comment by bcp33bosox — July 20, 2011 @ 10:57 am

  50. Do yourself a favor go sort road FIP for the last 4 years 08-11 Jimenez is tied with Felix for fifth behind Timmy, Doc and Lee. That seems pretty consistent or is at least a large enough sample size to determine a pitchers value.

    Comment by Gordon Brown — July 20, 2011 @ 11:12 am

  51. This might go on for a while, and I’m not a statistician, so feel free to find a better way to do this, but here goes. To try to solve the sequence issue, I tried looking at the change in WPA between starts for a bunch of these players. I used WPA only because I couldn’t find Game Score on the Fangraphs pages, but hopefully it approximates what we’re trying to get at. There’ll be more situational noise here than with Game Score, but that might help approximate the perception issue. Also this is just a comment, so deal with it.

    I looked at Jimenez, Haren and Wilson (the two biggest outliers in the OP), Hamels and Lincecum (two roughly-the-same-as-Jimenezes from the OP), and Burnett.

    Avg. WPA: .081
    Avg. Change betw. Starts: .015
    Ratio: .186
    St. Dev.: .257

    Haren: .064 / -.011 / -.181 / .257
    Wilson: .064 / -.002 / -.024 / .258
    Hamels: .101 / .006 / .056 / .246
    Lincecum: .049 / -.001 / -.026 / .290
    Burnett: -.032 / -.005 / .150 / .254

    So there’s actually a lot going on here. Once again, the Standard Deviations are clustered fairly close together; Jimenez, Wilson, Haren, and Burnett are all roughly similar, and Hamels (.246) and Lincecum (.290!) are the outliers.

    But if you look at average WPA swing between starts, Jimenez jumps to the front of the line. He has the largest average swing (.015) by a considerable margin. He also has the largest ratio of swing-to-average-WPA, though because his WPA is so good, his lead here is much smaller.

    Curiously, Lincecum has the lowest average WPA swing between starts and the highest standard deviation.

    Burnett, meanwhile, isn’t so much inconsistent as he is terrible. He’s the only player here with a negative average WPA. His average swing isn’t so bad, but the ratio of swing-to-average-WPA is the third highest, largely because his average WPA is so abominable.

    Two final things. I counted up everyone’s “good” (positive WPA) and “bad” (negative WPA) starts. Jimenez has the most good starts (34) and the fewest bad starts (18), and therefore has the highest ratio of good/bad. Everyone else is pretty close together, with 32 or so good starts and 20 or so bad starts, except Burnett, who’s at 25/28.

    However, I also counted “Great” and “Terrible” starts, by arbitrarily deciding that anything >.300 was “Great” and anything <-.300 was "Terrible". There are better ways to do that, but whatever. By this (again, pretty arbitrary) measure, Jimenez has 7 terrible starts, which account for 38% of his total "bad" starts. That's the highest Terrible Ratio of the bunch (yes, even worse than Burnett). His Great Ratio (35%), meanwhile, is almost identical to Wilson and Hamels, pretty much in the middle of the pack.

    So. After all that, I think it at least becomes easier to see why Jimenez is considered inconsistent. He does tend to fluctuate more between individual starts than other similar pitchers, even if his standard deviation is middle-of-the-road. Moreover, when he has a bad start, it's more likely to be Really Bad, which magnifies those starts in people's minds and makes him seem more inconsistent. And, of course, he started the sample period with a 14-game Good streak, the longest streak of any of these pitchers in this entire period (the next best was Haren and Lincecum, tied at 8), which also magnified perceptions.

    TL;DR: By some measures, Jimenez might actually be a bit more inconsistent than his peers. But he's still really good.

    Comment by Rex Manning Day — July 20, 2011 @ 12:06 pm

  52. This used to be his problem. He has admitted as much that during games he loses concentration and lets his arm slot drop. Then he gets smacked around and he refocuses and pitches well.

    Last year his problem was that at his age he doesn’t have elite stuff anymore. Yet he’s still basically a two pitch pitcher and he can’t always throw his breaking ball for strikes. So if he’s not blowing the fast-ball by you he’s in trouble. Early this year he showed a commitment to using his change-up more. Recently not so much…

    Comment by Preston — July 20, 2011 @ 12:25 pm

  53. I think “luck” is the most overused term on FanGraphs. Ubaldo wasn’t unlucky early this season; he was wild and his velocity was down. Now his velocity is back up to the 96 range (from 91-93 early in the year) and he’s commanding his pitches. So he got hit hard in his early starts and has turned it around. That’s not luck, that’s skill.

    Comment by Trotter76 — July 20, 2011 @ 12:38 pm

  54. By the way, if the Rockies trade Ubaldo they are incredibly dumb. By WAR the closest comparisons are Felix Hernandez, Clayton Kershaw, Dan Haren, Tim Lincecum, C.J. Wilson, and Cole Hamels, some of the finest pitchers in the game. Meanwhile, Jimenez signed one of the worst contracts you can imagine a couple years back, 4 years, $10 mil total with 2 club options:
    09:$0.75M, 10:$1.25M, 11:$2.8M, 12:$4.2M, 13:$5.75M club option ($1M buyout), 14:$8M club option ($1M buyout)
    The average pitcher in the group above is making just short on $9 million this year, including Kershaw’s half mil because he hasn’t even had his first shot at arbitration. Take him out and the average goes to $10.45 million. After investing heavily in Tulo and CarGo last year, it makes no sense to trade your ace when you’ve got him locked up for 3 more years below market value.

    Comment by Trotter76 — July 20, 2011 @ 12:49 pm

  55. One last thing, regarding streakiness v. game-to-game inconsistency.

    Jimenez only has one streak of 3 or more “bad” games in a row in this period. On average, he has 2.8 good games in a row and 1.6 bad ones, both of which are pretty much in the middle of the spread of these pitchers. The outliers are Hamels, who averages 3 good games in a row, and Lincecum, with 2.1 bad games in a row.

    If you confine it to actual streaks (3 or more games in a row), the pack tightens even more. Jimenez has the longest average good streak at 6 games, but that’s distorted by his 2010 start. He only has 4 streaks of 3 or more good games in that period (Haren and Wilson have 6, and Hamels has 5). Everyone else averages 4-5 games for their streak length.

    As for bad streaks, Jimenez only has 1 of 3 games or more, and it lasted 4 games. Haren and Hamels also only have one qualifying bad streak each. Burnett only has two, of 5 and 3 games. And only two of these players have had consecutive Terrible games: Lincecum (two in a row) and Burnett (3 in a row, at the end of his 5-game bad streak).

    So if anything, streakiness is even less to blame for the inconsistency meme. The only real outliers are Jimenez’s giant good streak, and Burnett’s Very Bad streak in 2010. Otherwise, neither player seems to be measurably “streakier” than the other players in the sample, particularly not Jimenez.

    Comment by Rex Manning Day — July 20, 2011 @ 12:54 pm

  56. @Novaether I don’t think there is really a “right” answer mathematically here. The idea I was trying to point out is that having a bunch of great games in a row can create a psychological “anchor point” that can get disrupted when one sees the inevitable bad game. If the sequence is interrupted, it never causes any cognitive dissonance.

    The AV Club recently tried to use this idea as a test of musical greatness with their 5 album theory. I don’t necessarily but the arguments but it is an inkling that this idea is out there in the mists.,59098/

    Comment by Nick44 — July 20, 2011 @ 12:57 pm

  57. Most of the times I think it involves good performance being the “anchor” that people remember, I agree – unless you are talking about Tony Larussa’s assessment of the performanace of Colby Rasmus.

    I believe that is the counterexample.

    Comment by Nick44 — July 20, 2011 @ 1:00 pm

  58. Absolute Value for the average between starts maybe? Without doing that your good swings and bad swings should cancel out and leave you with the tails / number of games setting your average.

    Comment by Nick44 — July 20, 2011 @ 1:06 pm

  59. True, writers should just stick to listing a player’s WAR and then discussing whether or not they have been lucky/unlucky on balls in play.

    Comment by Steve — July 20, 2011 @ 1:22 pm

  60. How has he tanked?

    when he was 4-1, his ERA was 3.90.

    Since that point, it’s 4.30.

    Obviously he hasn’t been great, probably not even good, but “tanked” implies something far worse than he has pitched.

    Comment by Steve — July 20, 2011 @ 1:28 pm

  61. excellent stuff.

    “So if anything, streakiness is even less to blame for the inconsistency meme. The only real outliers are Jimenez’s giant good streak, and Burnett’s Very Bad streak in 2010.”

    does this mean you think this is a definitional issue more than anything?

    Comment by joshcohen — July 20, 2011 @ 1:39 pm

  62. I propose a near moratorium on the word “lucky”. It certainly has merit, but it is certainly overused as well. How about a phrase like “intangibly positive/negative” or even good/bad without not quite knowing why”, haha.

    Comment by baty — July 20, 2011 @ 2:28 pm

  63. My own impression is that great-“stuff”/mediocre-or-poor-control pitchers are more likely than others to get the “inconsistent” label, whether or not they’re actually any less consistent on a year-by-year or start-by-start basis. When I think of guys I’ve heard labeled as inconsistent (or whom I’ve thought of myself as inconsistent), it’s guys like Jonathan Sanchez, Brandon Morrow, Yovani Gallardo, Dice-K, etc. Jimenez and Burnett fit that mold too.

    Maybe it’s because strikeouts and walks from pitchers elicit such extreme, opposite emotions from baseball fans. When a pitcher is striking a lot of guys out, we are in awe of his dominance; when he’s walking guys left and right, we want to pull our hair out. So, high-K, high-BB guys are perhaps more likely to create the impression of being inconsistent, whether or not they actually are.

    My own impression could be wrong of course. It might be interesting to track the usage of “inconsistent” in the baseball media, and see if these kinds of pitchers really do get stuck with that label more than others do.

    Comment by DavidJ — July 20, 2011 @ 7:00 pm

  64. No, the point was to see if his “results fluctuate wildly from start to start.”

    Comment by DavidCEisen — July 20, 2011 @ 7:50 pm

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Close this window.

0.296 Powered by WordPress