SoxProspects News
|
|
|
|
Legal
Forum Ground Rules
The views expressed by the members of this Forum do not necessarily reflect the views of SoxProspects, LLC.
© 2003-2024 SoxProspects, LLC
|
|
|
|
|
Forum Home | Search | My Profile | Messages | Members | Help |
Welcome Guest. Please Login or Register.
|
Post by pedey on Dec 12, 2013 22:40:04 GMT -5
Doubront went 11-6 with a 4.32 ERA over 162.1 frames this year. These stats lie to an extent though, as he had two relief appearances and got tagged in both of them.
He went 11-6 with a 3.87 ERA and 133 strikeouts in 27 starts over 155.2 frames as a starting pitcher. He averaged 5.77 innings per start.
I have always liked him as a back-end starter. He is a solid pitcher, although he needs to improve on working deeper into games. Although some say he is better suited for the bullpen, he has certainly performed better as a starter.
Considering he improved in his sophomore campaign, can we expect him to improve even more? Is he penciled in as next year's #4 or #5 starter? If he performs very well next year, could he be in line for a team-friendly contract extension that buys out his arbitration years and maybe 1 or 2 of free agency years?
|
|
|
Post by wcsoxfan on Dec 12, 2013 22:51:32 GMT -5
In my opinion:
- He is the Red Sox #4 starter next year unless an unexpected trade occurs.
- His continued improvement likely depends on his dedication to conditioning (which has been questionable in the past).
- I don't expect the Red Sox to offer him an extension due to the conditioning questions. He seems to be like Hanley in that he can be very good when he works at it - but you can never depend on him working hard consistently. Once he gets the extension he may lose incentive and slack off. (I am in no way comparing the talent level of Felix and Hanley)
|
|
|
Post by thelavarnwayguy on Dec 12, 2013 23:46:19 GMT -5
In some ways he is a trade candidate. He's low cost and controllable, therefore more attractive to many teams. We could get a decent return for him. I would have thought Billy Butler straight up. Solving our 1st base problem.
But we went a different direction. We can maybe even trade 2 starters still:
Lester Buchholz Lackey Peavy Doubront Workman Dempster Morales Webster Wright Ranaudo Delarosa Britton
We should dangle some of these guys and move 2 of them. From the group of Doubront, Morales, Dempster, Peavy or one of the prospects.
|
|
|
Post by raftsox on Dec 13, 2013 4:45:16 GMT -5
In some ways he is a trade candidate. He's low cost and controllable, therefore more attractive to many teams. We could get a decent return for him. I would have thought Billy Butler straight up. Solving our 1st base problem. But we went a different direction. We can maybe even trade 2 starters still: Lester Buchholz Lackey Peavy Doubront Workman Dempster Morales Webster Wright Ranaudo Delarosa Britton We should dangle some of these guys and move 2 of them. From the group of Doubront, Morales, Dempster, Peavy or one of the prospects. Morales and Britton aren't starters, De La Rosa might be but probably isn't. I doubt there's a market for the first two as starters. Selling De La Rosa now would be selling for pennies on the dollar. I'm not sure what the anxiety is to trade any of them is for. I doubt Buch gives you more than 20 starts so you'll need someone to cover for him part of the time. Specifically regarding Doubront: he's a solid back end starter making league minimum. If you can get value for him great, if not don't worry about trying to get rid of him.
|
|
|
Post by sarasoxer on Dec 13, 2013 7:49:14 GMT -5
I am more optimistic than most here re Doubront. What I saw in the playoffs was a bit of a different guy. To me he displayed significantly better control and I recall mention being made of Sox personnel then working on mechanics with him (not that they weren't before). If that was not a mirage or merely a temporary upping of his game and if his velocity returns to that of 2012 (there were some upticks late in 2013), I believe this guy could be one of the good stories for 2014.
|
|
|
Post by jimed14 on Dec 13, 2013 10:04:59 GMT -5
A cost controlled decent starting pitcher is exactly what you do not give up unless it's for a huge upgrade elsewhere. Without guys like that making up a portion of the roster, we have no depth and no ability to make moves. 2012 wasn't that long ago.
|
|
|
Post by p23w on Dec 13, 2013 15:22:45 GMT -5
Doubront went 11-6 with a 4.32 ERA over 162.1 frames this year. These stats lie to an extent though, as he had two relief appearances and got tagged in both of them. He went 11-6 with a 3.87 ERA and 133 strikeouts in 27 starts over 155.2 frames as a starting pitcher. He averaged 5.77 innings per start. I have always liked him as a back-end starter. He is a solid pitcher, although he needs to improve on working deeper into games. Although some say he is better suited for the bullpen, he has certainly performed better as a starter. Considering he improved in his sophomore campaign, can we expect him to improve even more? Is he penciled in as next year's #4 or #5 starter? If he performs very well next year, could he be in line for a team-friendly contract extension that buys out his arbitration years and maybe 1 or 2 of free agency years? I'm not sure how much Doubront will improve, but i would suggest that the litmus test will be his IP. I would set the bar at 180 innings pitched. Should Felix attain this in 2014 I think we are all happy campers. I do believe there is room for improvement. Whether this is based on his conditioning or his adapting and maturation to pitching at this level is up for debate. All things considered if Felix is the #4 or #5 starter for the RS there is a very good chance that the team has a slight advantage against an opponent's #4 or #5. I hope Felix gets better. I think he now knows the rules and the expectations of his employer. We'll just have to see how motivated he is to improve his game.
|
|
|
Post by rjp313jr on Dec 13, 2013 16:25:00 GMT -5
I'm high on Felix. He's lacking consistency right now, but he's had long sustained stretches of excellence. The work ethic is a big concern, but if he straightens that out I think he's a strong 2/3 workhorse type starter. In other words, not a lot different from Lester's career.
|
|
|
Post by bmitchsox on Dec 13, 2013 17:13:56 GMT -5
I think he could work his way up to a #2 as well, it all depends on how hard he decides to work at it. We saw flashes of greatness last year, especially in the postseason!
|
|
|
Post by JackieWilsonsaid on Dec 13, 2013 17:19:54 GMT -5
Agreed.
The other big variable is the pen. Guys pitch different when they know in the 6th or Seventh they can air it out and know the eighth and ninth are set.
A healthy pen (after last years disaster we should expect an improvement to the mean) could mean less stress innings and more wins.
I am surprised that we have not been overwhelmed by an offer.
I can't think of a cost controlled lefty with his upside.
Anyone have a comp?
|
|
|
Post by rjp313jr on Dec 13, 2013 18:55:50 GMT -5
Well I'd say Matt Moore, but his upside is way higher.
|
|
|
Post by pedey on Dec 14, 2013 11:20:18 GMT -5
I remember hearing at one point this year someone said, maybe it was Farrell, that "Doubront's stuff is as good as anyone else's". Although he needs to work on control, consistency, work ethic, and pitching deep into games, I don't think reaching as high as a #3 or maybe a #2 starter is out of the question.
For a stretch of 19 starts, from mid-May to the end of August, he posted a 2.99 ERA. He also started only four games in which he gave up more than 3 earned runs.
He certainly has flashed signs of greatness, and hopefully, that will continue.
|
|
|
Post by jmei on Dec 14, 2013 11:24:23 GMT -5
I can't think of a cost controlled lefty with his upside. Anyone have a comp? C.J. Wilson comes to mind as a lefty with all-around good but not great strikeout, walk, and ground-ball rates. But Wilson has pitched four 200+ IP seasons in a row, while Doubront hasn't broken that barrier even once in his career. I agree with p23w that that's the next step for Felix.
|
|
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,944
|
Post by ericmvan on Feb 28, 2014 4:10:03 GMT -5
Some discussion of Doubront in the FG ranking thread led me to look at his game log. Was he really much better for a stretch of 9 starts (June 18 to August 4), or was that just luck?
He had a 2.02 ERA in that 9-game stretch, versus a 4.98 ERA before or after. That's awfully hard to explain just by sheer luck. If you ignore runners left on and just look at guys who scored as a percentage of guys who either scored or were retired, the odds against the differences in percentages being random are 185 to 1. (If you look at guys who scored as a percentage of all BFP, it's 1088 to 1, so that's clearly not the way you want to test such splits.)
Is there any evidence for non-randomness in his BABIP and HR / Contact?
Well, he had 16 starts where he allowed 0 homers, and 11 where he allowed one or more. He also had 16 starts with a BABIP below the league average for LHP of .300, and 11 starts where it was .300 or above.
When you look at the way those starts clustered into streaks, they were both streakier than expected. The home run games had a 11% chance of being that streaky, the BABIP games a 20% chance. So there's about a 2% chance of seeing both stats be that streaky in the same game log. (There was no start-to-start correlation between them.)
What drove the two results was a 6-game streak of good BABIP, which overlapped by two games a 6-game streak of no homers allowed. And those 10 games are the 9 games of his great streak (plus the first of his late-season fade).
The odds of getting a 6-game streak like that in any specified six starts are (16/27)^6, or .04. The odds of getting such a streak somewhere in the course of 27 starts, and hence 22 different 6-game streaks to look at, are, I think, 1-((1-the above)^22), or 62%. (The 22 trials are not independent, but I think that doesn't matter). The odds of getting a streak like that in both stats are that squared, or 14%. And the odds of the 2 streaks overlapping by 1 or 2 games are about 2%.
I think it's pretty clear that he actually did pitch better for that stretch, although he probaly had some luck as well. Next up, at some point, a look at what pitch/fx tells us.
|
|
|
Post by Chris Hatfield on Feb 28, 2014 7:10:54 GMT -5
Honest question: why would you ignore runners left on in that second paragraph? It looks like you did because of the outcome, which seems like it wouldn't be a reason to make a determination like that. Maybe I'm mis-reading you (probably am), but it seems like you're essentially saying the numbers were too good, therefore they must not be usable. If a guy really is pitching better, doesn't that mean he's more likely to strand the runners that do get left on? Alternatively, couldn't a guy be generally "pitching better" because he's not getting as flustered because there's a runner on, or was more mechanically sound out of the stretch than usual and thus pitched better from the stretch?
I also don't get the significance of the combined chance of his BABIP and HRA streaks occurring the way they did, particularly when those streaks didn't coincide in full. Intuitively, it seems like the fact that those weren't the same six games would show they're unrelated to some degree, no?
I actually agree with you that the stretch of starts is probably significant, but I'm just not sure why these numbers say it is. To me the questions wouldn't be the chances of these games occurring in this order/streak under the assumption that the events were independent (which they're probably not, as you even say), but rather things like the effect of the opposing team's strength and whether things like his command or stuff changed (as you say would be the next step here).
Interesting stuff to think about though, as always, so thanks for starting the convo.
|
|
|
Post by joshv02 on Feb 28, 2014 7:48:20 GMT -5
Oh, when in doubt, assume that Eric is trying to demonstrate that some really small subset of numbers are predictive, not that they are "too good, therefore they must not be usable."
Doubront is an above average major league starting pitcher who, as a consequence, can experience a streak of really good games in a row (or interspersed), and, like many, will be defined by the number of those really good games rather than the fact that they exist from time-to-time.
|
|
|
Post by jmei on Feb 28, 2014 11:52:19 GMT -5
Eric: I continue to believe that that sort of odds analysis, undertaken on very small samples, is a flawed method of analysis. It can be useful when analyzing the results of a controlled experiment, but here, you're applying it to data with dozens of extraneous variables that might be confounding. As Chris mentioned above, opponent quality is the biggest potential issue-- how good were the opposing offenses, how good were they versus lefties, were their star players in the lineup, etc. There's also stuff like which ballparks the games were played in, what was the weather like, which defenders did he play in front of (relevant for BABIP analysis), which catcher was he throwing to (Ross has superior pitch-framing skills), and more. When you're analyzing six-game stretches, this kind of stuff can matter a lot.
|
|
|
Post by fenwaythehardway on Feb 28, 2014 14:51:59 GMT -5
Some discussion of Doubront in the FG ranking thread led me to look at his game log. Was he really much better for a stretch of 9 starts (June 18 to August 4), or was that just luck? He had a 2.02 ERA in that 9-game stretch, versus a 4.98 ERA before or after. That's awfully hard to explain just by sheer luck. If you ignore runners left on and just look at guys who scored as a percentage of guys who either scored or were retired, the odds against the differences in percentages being random are 185 to 1. (If you look at guys who scored as a percentage of all BFP, it's 1088 to 1, so that's clearly not the way you want to test such splits.) Is there any evidence for non-randomness in his BABIP and HR / Contact? Well, he had 16 starts where he allowed 0 homers, and 11 where he allowed one or more. He also had 16 starts with a BABIP below the league average for LHP of .300, and 11 starts where it was .300 or above. When you look at the way those starts clustered into streaks, they were both streakier than expected. The home run games had a 11% chance of being that streaky, the BABIP games a 20% chance. So there's about a 2% chance of seeing both stats be that streaky in the same game log. (There was no start-to-start correlation between them.) What drove the two results was a 6-game streak of good BABIP, which overlapped by two games a 6-game streak of no homers allowed. And those 10 games are the 9 games of his great streak (plus the first of his late-season fade). The odds of getting a 6-game streak like that in any specified six starts are (16/27)^6, or .04. The odds of getting such a streak somewhere in the course of 27 starts, and hence 22 different 6-game streaks to look at, are, I think, 1-((1-the above)^22), or 62%. (The 22 trials are not independent, but I think that doesn't matter). The odds of getting a streak like that in both stats are that squared, or 14%. And the odds of the 2 streaks overlapping by 1 or 2 games are about 2%. I think it's pretty clear that he actually did pitch better for that stretch, although he probaly had some luck as well. Next up, at some point, a look at what pitch/fx tells us. Does pitching above his normal level for nine starts actually mean anything, though? How uncommon is it for a pitcher to do that? What's the normal variation in "true talent" for a pitcher in a given season? Without a good answer to those questions, I don't see the value in putting any player's hot or cold streaks under a microscope like this.
|
|
|
Post by teddymonster on Feb 28, 2014 15:36:08 GMT -5
Does pitching above his normal level for nine starts actually mean anything, though? How uncommon is it for a pitcher to do that? What's the normal variation in "true talent" for a pitcher in a given season? Without a good answer to those questions, I don't see the value in putting any player's hot or cold streaks under a microscope like this. Eric, I love seeing your posts come on the board. You're a wizard with baseball stats and I've seen you mine the most amazing data of any poster out there---I've followed you're posts since the old-school SOSH days, back when that board was still brilliant. Having said all that, I'd have to agree with "FenwayTheHardWay". You've done a lot of stats work here, but it's super small sample sizes and you've yet to show that it's even meaningful relative to the streakiness of other pitchers. Beyond that, unlike most of you're other posts, I'm not sure there's anything new or interesting to see here. I think plain old watching Dubront with our eyeballs tells us all we need to know. He has a lot of potential, but isn't consistently dominant, yet.[/quote]
|
|
|
Post by moonstone2 on Feb 28, 2014 17:50:15 GMT -5
Here is how I would do something like that.
First off we want to use a discrete variable. Something a player either did or didn't. Let's take batting average in this case. In every AB the player either got a hit and hence batted 1.000 or he didn't and batted zero. Let's say his true talent level was .300. That means that every event would have a difference from the mean of either .7, (1-.3) or .3, (0-.3). The first difference would occur 30% of the time, and the second occurrence 70% of the time. So to compute the standard deviation you would calculate SQRT(((.7^2)*.3)+(.3^2)*.7))/(n-1)) If we are talking about 50 ABs the standard deviation would be 54 points of BA. Generally I look at events that are at least 2 standard deviations away. So if the player hit .408 over 50 ABs that means there is a 97.5% chance that this event was caused by something other than random chance.
Now. I do see the meaning of the standard deviation quite often misinterpreted so let me correct a couple of misinterpretations:
First off it's been noted that the standard deviation of a players performance over a season isn't normally distributed. That's not relevant. What IS relevant is that a players performance over an infinite amount of ABs will be normally distributed away from his true talent level.
Second. It's also been noted that there can be other factors that that may explain the deviation. A confounding factor. That's certainly possible, but a confounding factor must be one that you wouldn't expect to see normally distributed over an infinite amount of ABs. The weather for instance, the presence of superstars in the lineup, you would expect to be randomly distributed over time and thus already imbedded in the standard deviation computation.
Third. Small sample sizes with extreme outcomes CAN tell you something. Let's say that Archie Bradley or something strikes out 20 out of 30 batters he faces in a game. That would tell you that he has an extreme level of talent even though it is just one game.
Finally with regards to Eric's analysis. An event that occurs 2% of the time doesn't tell you anything definite. You would expect a ball player with a 9 or 10 year career to have one streak that is that far out of line with his level of talent. He's just saying it's worth doing more research to investigate if he was in fact throwing better pitches during this time.
|
|
|
Post by jmei on Mar 1, 2014 12:21:40 GMT -5
Second. It's also been noted that there can be other factors that that may explain the deviation. A confounding factor. That's certainly possible, but a confounding factor must be one that you wouldn't expect to see normally distributed over an infinite amount of ABs. The weather for instance, the presence of superstars in the lineup, you would expect to be randomly distributed over time and thus already imbedded in the standard deviation computation. That would only be true if the sample was randomly selected. Here, the opposite is true-- we chose the sample precisely because the results were significantly different from what we believe to be the median true talent level. What we're really trying to figure out here is what explains these stretches of low BABIP/home runs-- how much of it is Doubront pitching meaningfully/sustainably better, how much might be other variables (like opponent quality, ballpark, defense, etc.), and how much is just variance/luck.
|
|
|
Post by deepjohn on Mar 1, 2014 13:20:51 GMT -5
I'd be interested to know what any "small sample size police" here think of the analyses done here: On the reliability of pitching statsA new pitching metricFrom what I can see there, Doubront's sample size as a starter over the 9 games Eric identified, facing more than 70 batters, may well be large enough to be predictive, given Doubront's reliance on K/PA, LD/BIP and GB/BIP for his success. Eric, thanks for all your great work. Is there any way you could have your own forum, where your work-in-the-commons could be discussed? Right now, the user needs to go to your profile in Members and check all posts by you, to see where or when you've posted. Also, any "small sample size police" could have a caveat emptor.
|
|
|
Post by jmei on Mar 1, 2014 14:11:34 GMT -5
I'd be interested to know what any "small sample size police" here think of the analyses done here: On the reliability of pitching statsA new pitching metricFrom what I can see there, Doubront's sample size as a starter over the 9 games Eric identified, facing more than 70 batters, may well be large enough to be predictive, given Doubront's reliance on K/PA, LD/BIP and GB/BIP for his success. Other research shows line drive rate takes a long time to stabilize. See, e.g.: www.baseballprospectus.com/article.php?articleid=20516www.fangraphs.com/library/principles/sample-size/It's the first time I've seen a study that's drawn such different conclusions re: the reliability of LD%, and I'm not sure why that is (any insight on this question would be much appreciated). However, even if I accept, arguendo, that LD% does stabilize quickly, the streaks identified above are periods of low BABIP and HR/C, which the analysis you linked to concedes are high variance. NB: I think Eric's research is fascinating and very valuable, and I always enjoy reading it. I don't want my criticisms to be interpreted too negatively-- I certainly don't want him to feel like he has to post it somewhere else. But if I do see possible methodological flaws or conclusions that seem too strong, I think it's valuable to point them out.
|
|
|
Post by jimed14 on Mar 2, 2014 13:24:09 GMT -5
Doubront looks pretty good today with his delivery.
|
|
|
Post by redsoxfan2 on Mar 3, 2014 13:54:54 GMT -5
I remember reports of last spring training that Felix came into camp out of shape, I was wondering how he looked in the early goings of spring training? I don't think Will Middlebrooks will be any better for adding 15lbs of muscle, but I do worry about the guys who show up to camp out of shape.
|
|
|