Pitching Prospects and K-rates
by Woodman663 on Feb 17, 2012 8:29 AM EST
in Toronto Blue Jays Analysis
Email
PrintI studied the minor league K-rates of successful young big league pitchers the past season, and found that most of them had K-rates over 20% in the minors. That led me to post an article arguing the importance of K% in the summer. However, what I did not do was post the results of those successful big leaguers. Now, I'll take another look at this analysis, and expand the number of pitchers and list the results here in this article. I generally assume the age of 23 or lower to be age appropriate for AAA, with each level below that one year lower. So 22 for AA, 21 for A+ etc. I'm using a minimum of 50 IP for each level.
Of course, which pitchers can be described as successful big leaguers is a bit subjective, but I think the list contains few pitchers that wouldn't be welcome on most, if not all major league rotations. If I omitted or included pitchers you disagree with, please bring it up in the comments.
Rookies in 2006
Jered Weaver - A+/AA (22): 29.4%, AAA (23): 31.5%
Matt Cain - A (18): 29.7%, A+/AA (19): 24.5%, AAA (20): 28.1%
Francisco Liriano - Rk- (17): 28.1%, A (18): 25.8%, A+ (20): 26.1%, AA/AAA (21): 30.9%
Justin Verlander - A+/AA (22): 30.0%
Cole Hamels - A/A+ (19): 38.4%
Josh Johnson - A (19): 16.9%, A+ (20): 20.6%, AA (21): 18.9%
James Shields - A (19): 21.4%, A+ (21): 18.8%, A+ (22): 18.5%, AA (23): 23.8%
Anibal Sanchez - A- (20): 32.6%, A+/AA (21): 28.4%, AA (22): 25.1%
Jon Lester - A (19): 15.7%, A+ (20): 25.4%, AA (21): 27.1%
Adam Wainwright - Rk-/Rk (18): 34.5%, A (19): 26.6%, A+ (20): 23.9%, AA (21): 21.4%, AAA (22): 22.7%
Matt Garza - A (21): 27.2%, A+/AA/AAA (22): 29.0%, AAA (23): 23.7%
Chad Billingsley - Rk (18): 27.6%, A+/AA (19): 28.5%, AA (20): 26.9%, AAA (21): 26.3%
Jonathan Sanchez - A (22): 31.1%, AA/AAA (23): 34.7%
Shaun Marcum - A/A+ (22): 25.8%, AA/AAA (23): 19.4%
C.J. Wilson - Rk/A (20): 26.7%, A+ (21): 16.3%, AA (22): 16.5%
Edinson Volquez - A/A+ (20): 20.4%, A+/AA (21): 24.2%, AAA (22): 25.5%
Rookies in 2007
Yovani Gallardo - A (19): 21.9%, A+/AA (20): 30.8%, AAA (21): 34.9%
Ubaldo Jimenez - Rk (18): 22.6%, A (19): 21.4%, A+/AA (21): 22.3%, AA/AAA (22): 23.1%
Dallas Braden - A+/AA (21): 23.2%, AA/AAA (23): 28.5%, AAA (24): 25.6%
John Danks - A/A+ (19): 25.1%, A+/AA (20): 20.5%, AA (21): 25.2%
John Lannan - A- (20): 13.9%, A (21): 18.6%, A+/AA/AAA (22): 15.0%
Rookies in 2008
Jair Jurrjens - Rk-/A- (18): 21.0%, A (19): 18.4%, A+/AA (20): 19.9%, AA (21): 20.0%
Clayton Kershaw - A (19): 32.4%, AA (20): 24.5%
Johnny Cueto - A/A+ (20): 26.1%, A+/AA (21): 25.3%
Max Scherzer - AA (22): 23.8%, AAA (23): 36.1%
Jordan Zimmermann - A- (21): 32.9%, AA (22): 23.5%
Clay Buchholz - A (21): 28.5%, AA/AAA (22): 34.5%
Ian Kennedy - A+/AA/AAA (22): 29.1%, AAA (23): 25.7%
Gio Gonzalez - Rk/A (18): 24.9%, A/A+ (19): 30.9%, AA (20): 24.3%, AA (21): 30.3%, AAA (22): 24.4%
Jaime Garcia - A/A+ (19): 20.6%, AA (20): 22.0%, AA/AAA (21): 21.0%
Jonathon Niese - A (19): 24.0%, A+ (20): 19.1%, AA (21): 21.5%, AAA (22): 20.5%
Rookies in 2009
Brett Anderson - A/A+ (19): 25.0%, A+/AA (20): 27.1%
Ricky Romero - A+/AA (21): 19.5%, AA (22): 19.6%, AA/AAA (23): 16.0%
Jeff Niemann - AA (23): 26.3%, AAA (24): 21.3%, AAA (25): 23.4%
Tommy Hanson - Rk (19): 26.9%, A/A+ (20): 27.7%, A+/AA (21): 29.2%, AAA (22): 35.0%
David Price - A+/AA/AAA (22): 24.8%
Trevor Cahill - A (19): 26.8%, A+/AA (20): 27.5%
Doug Fister - AA (23): 15.2%, AA (24): 17.3%, AAA (25): 17.3%
Mat Latos - A- (19): 30.1%, Rk-/A-/A (20): 29.7%, A/AA (21): 27.3%
Jhoulys Chacin - Rk (19): 20.3%, A/A+ (20): 22.5%, AA (21): 20.0%
Madison Bumgarner - A (18): 29.9%, AA (19): 16.4%, AAA (20): 17.0%
Analysis
My 20% rule seems to work well, at least for groundballers. Flyball pitchers who are successful in the majors seems to be above 24% most of the time. Also, college draftees seem to post more obscene strikeout-rates than high school draftees. There are, of course, a few exceptional cases. Out of the 41 pitchers listed, 10 had seasons below 20% Ks. However, only 3 never topped 20 in their minor league careers, suggesting that any pitcher who never dominates like that is awfully unlikely to ever become a solid starting pitcher.
Looking at the guys who succeeded despite not getting to 20%, I'll start with Ricky Romero. Ricky is a groundball pitcher who didn't actually get a good percentage of groundballs until his last minor league season, where he did top the 20% Ks-plateau in AAA (only 14.4% in AA that season). He's a successful case of making adjustments to huge effect. Doug Fister is a late bloomer who no-one saw coming, and who recently added a significant chunk of velocity to remain successful (now that SAFECO won't protect him any longer), whereas John Lannan has added as much as 3 mph to his fastball since he came up to the big leagues, but we still don't know if he's actually good or just lucky.
The same question, good or lucky?, can be applied to Jair Jurrjens, who was, in fairness, always close to or above 20% Ks. Josh Johnson added a lot of velocity late, while C.J. Wilson added velocity later as well, and then moved from the pen to starting successfully. James Shields, well I don't know why his pitching skills didn't translate into more strikeouts, but it should be noted that his best season for strikeouts was his last in the minors. Madison Bumgarner was really young (3 years younger than 'standard') for his level when he K'd so few, and he had velocity issues at the time. Marcum and Niese were never far below 20, while Lester quickly and drastically improved after a lacklustre first season.
Conclusion
I'd like to conclude this article by stating that I think minor league pitching stats, specifically the K-rates, are underrated as a tool of evaluation. People, especially Jays fans, often use anecdotal evidence that pitchers can improve by pointing to a guy like Halladay, or Romero, to argue that we shouldn't take minor league stats too seriously. Well, perhaps we should, because the case of Romero is very rare. Depending on what you think of Fister, #1-2 starters just don't produce low K-numbers in the minors. If a guy who is pegged as future top of the rotation starter is not getting Ks, something's wrong. And the odds are always against someone who has to make drastic improvements, like Drabek, Wojciechowski, and Jenkins. Henderson Alvarez (18.7% Ks in AA) might make a lot of you optimistic about the chances of other low-K guys, but even if he does succeed for an extended period in the MLB (and we're all hoping he does), that won't change the fact that it's just very hard to not K a lot of people in the minors if you have MLB-worthy "stuff".
Taken together with John Sickels' recent statement that he now thinks pitching prospects are easier to predict than hitting prospects, despite the high frequency of injuries, we should throw away the old "there is no such thing as a pitching prospect" line. Because there clearly is, even if there might not be such a thing as a "sure thing pitching prospect". With all the pitching prospects AA has added to the Jays' system in recent drafts, watching the K totals from our pitching prospects will be a fun occupation over the next few years.
117 comments
|
2 recs |
Do you like this story?
Comments
Adjustments and coaching
This is very interesting. To me, a great coaching staff (obviously a great pitching coach) at the major league levels and, well, throughout the system really, is very important. Having a coach at the big league or AAA levels that can really teach a prospect a subtle variation of a pitch or even adding a pitch (Alvarez seems to be a good candidate or example going forward) can take their K rate up. 2 pitch starters don’t last long. I like how our guys are learning how to throw quality change-ups. Romero’s has developed into one of the best and his K rate has been rewarded int he bigs. But a lurch in K rate, if no MPH is “found” later on, is likely due mostly to adding something to the repetoire or, at least, refining a borderline pitch.
You've just proven comment advertising works - Rent This Space
I wonder if, too, if success could be predicted at all by K rate movement across levels. That is, do the “studs” or “aces” have a stable K rate no matter what level they play at? Does it go up? Does it go down but less than average pitchers?
What would THEN be interesting is to take those guys who had a lurch in K success suddenly and analyze, likely qualitatively for the most part, what factors contributed to that newfound success. Did they add a pitch into their repetoire? Did they change their fitness routine? Lose weight? Gain weight? Get new coaching? Change divisions? Start eating waffles for their pregame meal?
I volunteer to interview all the “lurchers” if it is in BBB’s budget!!!
You've just proven comment advertising works - Rent This Space
This is an interesting article and a good read
but I don’t think any conclusions can be drawn based on this type of analysis in terms of prospect evaluation going forward.
This analysis is telling us that 76.2% of pitchers who make it based on these criteria have a K% of 20%+ at all levels, and 92.8% have a K% of 20%+ at at least one level. To make a conclusion about whether this stat has predictive value in terms of a prospect making it or not, you need to take all pitchers who meet your age and inning requirements and see if there is a significant difference between the percentage of ones who make it and the ones who don’t, in terms of K%. Intuitively, there would be, but that is not shown in this analysis, so I’m not sure that any conclusions can be drawn.
I will do research like that
Now that Fangraphs’ minor league leaderboards include age.
The positive of this analysis is that it shows the career path of those who did make it to the majors quite clearly.
Of course, “conclusion” is a relative term here in this article, as you’ll never be 100% sure that x% will have K-rate of y% or more. The players change, the leagues change, the coaches change…
Blogging about the Toronto Blue Jays at Bluebird Banter
I’m currently putting together a similar spreadsheet that I did for the AA batters, but for AA pitchers. I think that will provide a better clue to the predictive value of K% (at least, for AA pitchers anyway).
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
I guess what I meant by my comment about conclusions
is that all you can conclude is that pitchers that fit this development path criteria are more likely to have 20%+ K rates throughout the minors. It doesn’t say anything about the probability a pitcher makes it based on their K%, because we don’t know the overall split of all pitchers who match this criteria.
For example, even though there is a much smaller number of pitchers that fit this criteria who have low K% and made it, it is possible (but highly unlikely) that they are the only pitchers that fit this development path criteria with a low K%. If that is the case, then there would be 100% success rates for pitchers with low K% who progress through the minors based on the given criteria.
Now, I doubt this is the case, but I don’t think you can conclude one way or the other without first looking at the failed cases too.
by Playoffs!!!!1 on Feb 17, 2012 11:20 AM EST up reply actions
the one problem with the failed cases
is that a lot of pitchers fail due to injuries, rather than lack of stuff/talent. So that creates a lot of noise.
Blogging about the Toronto Blue Jays at Bluebird Banter
whenever you get the chance
I’d love to see a similar study with K:BB ratios, rather than just K rate. a guy like Hendo might not be a superstar – you need great BB and K rates for that (and getting GBs doesn’t hurt) – but being elite at limiting the free pass, as Fister is, is pretty likely to allow for a great career IMO
another thing, which might have been obvious to everyone but me
wouldn’t high GB rates have a more significant effect on “finesse” (low-K) pitchers than “power” pitchers (high K, high BB)? the “finesse” guys don’t get as many automatic outs through Ks, so they need low-damage BIP whenever possible. of course, there’s selection bias here – guys with low Ks need to have very low BBs, elite GB rates or both to stick in the Majors – but it would be worth a look. has anyone done this already?
K:BB
I think is a poor indicator when used for minor league pitchers. It vastly overrates pitchers with great control but poor stuff.
Blogging about the Toronto Blue Jays at Bluebird Banter
I don't have much spare time anymore
And I don’t see the need/don’t want to do it. I’ve looked at enough historical minor league leaderboards to know that most control specialists just don’t cut it.
Blogging about the Toronto Blue Jays at Bluebird Banter
I think your point has a lot more merit
than woodman’s allowing here.
trying to predict the success of a low BB, high gb-inducing prospect by comparing his minor league k-rates to flyball pitchers who yield twice as many walks is a bad idea.
guys don’t fit into categories like “>20%” or “<20%” there are gradients that are being ignored here.
The point that K/BB overestimates finesse pitchers is true. That doesn’t mean you exclude walks from your study, it means you change how you look at K/BB ratio . . . adjust it so that it makes sense, by weighting Ks more heavily. OPS overestimates slugging average but that doesn’t mean that slugging average isn’t important
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
Jhoulys Chacin, Trevor Cahill, Jaime Garcia, Jonathon Niese
Are groundballers. They were still getting strikeouts. Major league GB% seems hard to predict from minor league GB%. For example, Cecil had awesome groundball numbers in the minors, but not in the bigs. Romero was the other way around.
I’m very well aware there’s a gradient, but 20% seems to be somewhat of a bottom point beyond which the prospect is “in trouble”. Same goes for BB% above 9-11%. However, a good BB% is not so much an indicator of a potential star, while a good K% is.
Blogging about the Toronto Blue Jays at Bluebird Banter
those guys aren't instructive
Garcia and Chacin walked 4+ batters per nine, Cahill and Niese were kind of old for their leagues and also had much higher walk rates than Alvarez.
K-rates are important but there really isn’t a magic number. Derek Lowe and Tim Hudson have had pretty successful major league careers
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
The point that K/BB overestimates finesse pitchers is true. That doesn’t mean you exclude walks from your study, it means you change how you look at K/BB ratio . . . adjust it so that it makes sense, by weighting Ks more heavily. OPS overestimates slugging average but that doesn’t mean that slugging average isn’t important
I’ve heard K-BB being used as an alternative. What are your thoughts on it?
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
Never considered that
Probably better than straight K/BB ratio, which is problematic for the reasons woodman said above. I think this might have the opposite problem, though. I’d rather have a 7 K / 2 BB pitcher than a 9 K / 4 BB pitcher, though if the pitcher is in the minors, the strikeouts might be more encouraging
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
Alse need to consider level
At lower levels, control is less important, since that’s something that develops with time. I mean, better control is obviously more favourable, but you want to see a guy getting Ks, which is generally indicative of good stuff.
At higher levels, you put more emphasis on control – a guy who hasn’t developed control will often get the bullpen tag, but more importantly, has more experience generally and without developing the skill. Ot course, for some guys it doesn’t really click until their 30s (Nolan Ryan, Randy Johnson, both of whom had phenomenal raw stuff so they still got by until they learned to pitch and really dominated)
great point
maybe something like low minors: K%, AA: K% – BB%, AAA: some kind of ratio that weighs strikeouts more heavily?
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
I’m guessing if you were actually trying to do a comprehensive study, correlate component success at each level to MLB success, you would find an increasing emphasis on BB% as you went up. The thing is, you also have to consider age, park/league factors, projection of stuff (a pitcher might two plus type pitches whend drafted, but is working on a third that clicks at some point – or never does) etc…so in practical terms, I’m not sure you can really say “this level, look at this; that level look at this”.
I’ve relatively new to looking at minor leaguers…this time last year, I could have told you who the top couple prospects were, but little about them. I knew nothing about analyzing mechanics, about projecting, about evaluating pitches. I think it’s incredibly important from an organizational perspective, and I’ve launched myself into understanding things. And as much as I love objectively analyzing and relying on data, I really think the data is limited beyond general trends we’ve discussed. The other thing is, the focus for prospects isn’t on dominanting minor leaguers, it’s on development – so a team that knows a guy has a great curveball but crappy change-up might put limits on the CB use and force him to use the change, eve if he gets knocked around a bit.
All this to say, I think you really need stats + scouting to properly evaluate minor leaguers, and I’m wary of saying, at this level look for this, at that level look for that – I think it gives a false sense of confidence about prospects, when it’s really a tremendously complex mix (which is why it’s such a crapshot compared to other sports – and yet, I’d argue, far more interesting and compelling).
Take Josh Banks for example, in 2005 at New Hampshire (age 22/23, so not old for league):
162.1 IP, 145K, 11BB(!!!), 159 hits. 644 batters faced, so a k% of 22.5% and BB% of 1.7%.
So that looks great, heck he looks like an ace in the making. But it was masking very pedestrian – but very developed – stuff. Compare that with Halladay and Carpenter, who had really pedestrian numbers in the minors, but who ended up as aces/#1s (when healthy in Carp’s case).
But generally, sure. When I look at minor league pitching numbers, I pay most attention to K-rate and BB-rate (less weighting for lower levels), and some passing attention to hits, HR, and GB/FB (more to look for anomolies than anything).
yeah, that is kind of how I take the numbers as well
i rarely (if ever) watch minor league players and, although i’m interested in the development of prospects in an abstract sense, i don’t worry too much over the statistics — as you say, what underlies those stats is more important and scouts actually see that.
on the other hand, if we are trying to look at statistics, i do think it is good to have some idea of what to look for (which I think you do as well, from what you’re saying).
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
For sure
Here’s the finer point I think I was trying to make.
We know K% and BB% are the most predictive of any MiLB stats, and to a lesser extent GB%. So definitely, look at these. But as a part of the overall evaluation, there’s so much more and they’re such a small piece, that I don’t it’s meaningful to say, look at K%-BB% rather than K% and BB% seperately, buecause one might correlate a little more.
I'm trying to think of the best way to pose the investigation
we know that high K% pitchers are likely to succeed in the Majors. should we just being looking at the guys whose K% suggests they won’t be successful, or should we be looking at everybody (high K% “superstars in the making” as well)?
hmm
i’d start looking at pitchers who are around 17% – 22% k-rates see if there is any discernible pattern. Those are the pitchers who are on the fringe, so that’s where I’d expect you’d find trends (good control, good gb-rates) emerging
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
fools seldom differ!
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
by jessef on Feb 18, 2012 4:25 PM EST up reply actions 1 recs
You could try binning ranges of K%
say <15%, 15%-20%, 20%-25%, >25% (maybe more fine grained, different ranges). Then look at the distribution of successes / failures in each bin, by defining a series of criteria (WAR, or defined levels of FIP, IP, etc.).
by Playoffs!!!!1 on Feb 20, 2012 11:10 AM EST up reply actions
I like that idea
though it would probably take more work than looking at only the fringe guys, it would also provide some useful information
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
Very good point
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
Like control,
I’d think the usefulness of groundball-rates would scale up in importance throughout the minors. What do you guys think?
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
I’m assuming that pitchers with a pitch repertoire that includes a sinker, splitter, two-seamer, etc. could have above average GB rates in lower levels (that is, if the pitches are effective). It may progress as they move to higher levels with the development of those pitches, but I think GB rate may stay above the league average for each league they play in.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
right, I just mean that
almost any major league-bound pitcher should be inducing a fair number of grounders in the low minors but groundball-inducing tendencies should become more predictive of MLB performance at AA or AAA. I know woodman mentioned that Cecil’s gb-rates in the minors were excellent, whereas he has poor gb-rates in the majors, but I wonder if anyone’s run a larger study on that. Also, Ricky Romero had very good gb-rates in AAA the season before his breakout with the Jays.
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
Anecdotally
From looking at a ton of minor league numbers, but not actually having run any numbers, the ground ball rates fall as they move through the minors. And that makes sense, GB represents weak contact, better hitters, less weak contact.
I’d guess the higher GB rate the better, but again, I’d think it’s because it’s representative of generally better stuff and the ability to induce weaker contact. I don’t know you look at minor league data and tell who will have a high GB% in the majors, though I haven’t run the data. Most good pitchers don;t spend a ton of time in the minors, so there might be a selection.sample problem too. The environments are just so different – in the minors you have some MLB quality guys but a lot of filler, in the MLB it’s all MLB quality guys basically. So if you get a bunch of filler guys to roll over on contact, I’m not sure it’s predictive of anything about MLB tendencies, other than the aforementioned general trend of better stuff = more weak contact generally = more MLB success
while that's true,
couldn’t you say the same about K- and BB-rates? AAAA pitchers could strike out a bunch of non-MLB quality hitters and then stall out once they’re facing quality hitters on the reg?
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
Yeah, I get what you're saying
And it’s completely logical.
I guess what I’m trying to say, and it’s not entirely clear in my mind – and to be clear, basically entirely anecdotal, so grains of salt are in order – it seems to me that K% at any given level in the minors correlates better to MLB K% than minor league GB% does to MLB GB%.
My explanation would be as follows:
A MLB quality minor leaguer will still swing and miss, though definitely less than scrubs. But when an MLB quality minor leaguer makes contact, it’s likelier to be good contact. Whereas the scrub, when he makes contact, is more likely to make weaker contact.
To be clear, that’s a theory. I couldn;t point to any actual evidence to back it up
i think there's something to what you're saying
and i think you’re probably right about scrubs making weaker contact. on the other hand, the high minors are much more highly populated with good hitters than rookie ball, so i’d think that, as you go up in level, pitchers who can really induce a lot of grounders might be able to make that translate to MLB success
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
I think what you’re describing essentially explains why the GB% falls as you move up the ladder. I guess what I’m trying to get at something further – it seems to me that some players have very similar MLB GB% as they did in the high minors. Other guys seem to have drastically lower MLB GB% – Cecil for example, who has turned into a fly ball guy in the majors. Now, maybe it’s a bias in my mind, and maybe it’s just looking at too few guys (batted ball profiles is fairly new, especially for minors), but I feel like there’s a larger force at work. Maybe it’s pitch usage or reporatoire, or something like that.
Yeah, I have thought about it, too
I do think pitch usage has something to do with it — a lot of changeup pitchers tend to give up a lot of flyballs, but I have no idea how to determine what drives them. I think the first step has to be testing the usefulness of MiLB gb-rate against MLB gb-rate, though, as you said, there are so many confounding factors that it’s hard to tease out what’s useful
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
Good changeups are actually
good at getting groundballs. I think a lot of guys with good changeups also use their four-seamers a lot and tend to induce extra flyballs due to the big contrast between change and four-seamer?
Blogging about the Toronto Blue Jays at Bluebird Banter
interesting
that does make sense as a possibility. relatedly, maybe it’s a rarity for pitchers who rely on change-ups to mix in two-seamers?
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
I don't think it's a rarity
But sinkerballers do seem to throw sliders more often as their primary secondary offering. Brandon Webb’s an exception, and of course, so is Henderson Alvarez.
Blogging about the Toronto Blue Jays at Bluebird Banter
do you mean the ability to outperform DIPS?
then yeah, based simply on the assumption that fielders (esp. 2B, SS and 3B) become better the closer the league is to the Majors
I looked into K%-BB%
These were the correlation coefficients I found between the two variables and WAR/150 IP:
K/BB: R = 0.717
K%-BB%: R = 0.781
This was out of all qualified starting pitchers between 2002 and 2011 (n = 283). So, maybe there is some weight into using K-BB over K/BB.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
hmm, interesting
you wouldn’t be able to add gb-rate into the two regressions, would you?
Also, what about using actual run averages instead of WAR/150, which is basically park-/league-adjusted FIP?
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
If I were to use just ERA and R/9 instead of FIP/xFIP/SIERA, and added GB rate to K rate and K-BB, these were the R^2 values that I got:
ERA:
GB+K: R^2 = 0.224
K%: R^2 = 0.356
BB%: R^2 = 0.038
K/BB: R^2 = 0.325
K-BB: R^2 = 0.456
(GB+K)-BB: R^2 = 0.273
R/9:
GB+K: R^2 = 0.183
K%: R^2 = 0.370
BB%: R^2 = 0.039
K/BB: R^2 = 0.341
K-BB: R^2 = 0.472
(GB+K)-BB: R^2 = 0.229
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
So it seems that K-BB may be a better measure to use than K/BB.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
which seems to suggest
that a player can succeed by being elite at limiting walks, even if they’re not also necessarily great at striking guys out
Out of fun
I ran a correlation between SIERA and the two variables:
K/BB: R = 0.778
K-BB: R = 0.928
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
Of course
That’s likely because K% and BB% are included into the equation.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
right
and SIERA way favors K-rates because it assumes high K pitchers will have low babips. Unfortunately, the babip correlation seems like it’s mostly an artifact of high K pitchers inducing a lot of flyballs so SIERA works better than xFIP on the aggregate but can be much worse when looking at individual pitchers
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
But we basically already know this, right?
I mean, basically, from a FIP point of view (holding HR constant), thse two are equivalent:
Player A: 8 K/9, 4 BB/9
Player B: 5 K/9, 2 BB/9
I'm a little confused
how is adding gb-rate into the regression making it less predictive? are you doing it all as one term (e.g., fitting R/9 ~ GB+K-BB) or adding it in as a separate factor? If it’s a separate factor, at worst it just shouldn’t change the model . . . do you mind if i ask which program you’re using?
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
Just using Excel. I did the former, as I don’t have any programs that can do multiple regression analyses.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
ohhhhh, got it
if you send me the datafile, i can run the multiple regression. i’d imagine that adding gb will significantly improve it but who knows
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
How do I send you the file?
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
Sent
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
ok, ran it for Runs/9
i am using r squared values adjusted by Thiel’s method ( http://en.wikipedia.org/wiki/Coefficient_of_determination#Adjusted_R2 )
K – BB: rsq = 0.41
with GB, rsq = 0.45, (p-value for GB = 0.003)
K/BB: rsq = 0.31
with GB, rsq = 0.33 (p-value for GB = 0.04)
with K- and BB-rates as separate factors, rsq = 0.42, with GB, rsq = 0.46 (p-value for GB = 0.003)
This is pretty interesting . . . K – BB is not only better than K/BB but is basically as good as using K and BB as separate factors
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
by jessef on Feb 18, 2012 4:17 PM EST up reply actions 1 recs
I'm curious
If you wouldn’t mind, that is. How much of a stats background do you have? By that I mean, exactly how deep is your background.
I’m familiar with adjusted R^2, but I’d never heard it called Thiel’s method, though that’s clearly the more formal name. But I’ve never taken more than intermediate econometrics (which I dropped, as I was taking it as an elective and an extra class, and it was too much for my schedule at the time to me to devote enough time for it)
that's it?
surprising to me, because your knowledge of the underlying concepts is impressive. I’m in “advanced” econometrics right now (unfortunately it’s time-series and not multivariate, which is annoying for the purposes of baseball) and you seem much more comfortable manipulating your data than I am
Yep
I both the introductory and intermediate clas, I was the only non-major in the room. I actually had to get special permission from the department to take them in the first place.
The comfort manipulating data just comes from working with Excel a lot.
MjwW
I’m not sure if we had this conversation, but did you end up in accounting?
and Business grad here as well. Business definitely helps with the excel side of things.
@VagabondBansal
it bothered me
that there wasn’t an Excel class for econ students (though it’s possible I could have taken one if I looked around)
fwiw
The Excel I learned in formal academic classes was useful, but paled to the Excel I learned in other settings.
I agree
Although it depends on the setting. It is 100% easier and faster to learn in a job setting.
Become a iBanker for a summer and you’ll know your way front and back.
In a classroom setting, I’d say look for an advanced finance course. Depending on your school, the good classes, will have a lot of excel modelling.
@VagabondBansal
don't mind at all
as long as what someone might call formal training in statistics, i have had one graduate course in univariate, one graduate course in multivariate, and one course that focused on using R.
in terms of my background employing the use of stats, my research (hoping to defend this May!) involves community analyses (mostly multivariate ordination with vector-fitting and indicator species analyses) and linear mixed effects model analyses.
anyway, i only referred to it as Thiel’s method there because the values were slightly different than the ones Frag mentioned and I wanted to make sure it didn’t seem like I hadn’t made anything up. I feel like just saying “adjusted R**2” without identifying what method you’re using to adjust it seems a bit silly.
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
What indicator species do you analyze?
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
well, by indicator species analyses
i meant performing analyses to determine which species were significant indicators of specific communities.
that said, my project examines the effects of large scale disturbances on coastal wetlands and i’m looking especially at woody species encroachment by wax myrtle into everglades sawgrass marshes.
i know you had said you took biostatistics . . . are you an ecologist as well?
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
I do find ecology interesting, though.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
another guy in my research lab
has been looking into incorporating some micro-array work on the population level.
personally, I find that I have trouble understanding just about anything I need a microscope to see, though
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
The concept behind microarray analysis can be somewhat difficult to interpret at first, but a lot of information that comes from MA analysis relating to differences in gene expression is very interesting.
I’m assuming the person you are referring to uses microarray analysis to study variations in microsatellites among a population?
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
Sorry if I got too technical there.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
well, his idea is to perform the analyses at the population level
and compare populations to one another. so i guess he was looking into doing them at the species level, really. the basic idea was to determine if there were differences between ecotypes and if those differences could be leveraged for restoration purposes
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
I've always liked your sig
but I saw the episode recently (it’s from the Simpsons, right?) and Milhouse actually says “I’m Tomo Ohka…” not “Tomokazu Ohka”
it is from the simpsons
I saw the episode so long ago that I must have misremembered it . . . maybe I should change it back to “the nigh Mets are my favorite squadron”.
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
This is major league data, no?
The problem I’d see with applying to minor leagues at this point is the quality of the limited BIP data we have. At the MLB level, I mostly trust it (or at least the GB% – the LD/FB dinstiction can be a little dodgey but it’s not terrible).
Sorry
This was meant as a response to jessef – I’m sure GB% matters, but not sure we can use it to study minor league data.
you mean MiLB gb-data from the mid-2000's
or gb-data from today? I’d think the more recent stuff is reliable, no? Doesn’t BIS have people scoring all these games or is that still just MLB level?
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
I'd say today as well
I don’t know exactly who does it in the minors, but there’s a problem with the BIP classifications geenrally.
Take for example a couple of Lansing Lugnuts in 2011:
Marcus Knecht had a BABIP of around .320 – yet a LD% of 9.7%. Marisnick had a BABIP of .380, yet only hit 16.7% LD. Granted, that’s not directly GB%, but I guess I’m just skeptical of relying too much in MilB BIP data f any kind (though you would think the ground ball/“air” ball classification would be okay
right, i have heard that the LD numbers
are suspect (as are the MLB ones, supposedly) but, as you say, it’s pretty easy to tell the difference between a grounder and a ball in the air.
though i guess it is possible that every game isn’t actually being scored and, if so, that could muck the numbers up a bit, too
"Look at me! I'm Tomokazu Ohka of the Montreal Expos!"
I wouldn't worry about the latter point
Every game is tracked batter-by-batter in essentially real time (you can follow the games this way online, they update at the end of every half inning), so the only thing would be if if some data is accidently not getting recorded – which I doubt would be the case.
My optimistic take on this
Given some of the things you’ve said about college pitchers having more dominant K rates and the changes that the exceptions made, it looks to me like repetoire means a lot. Typically, college pitchers have more advanced repetoires than high schoolers at the time of the draft. So when the guys make the adjustment to add more pitches it could turn a decent pitcher into a good pitcher. So there’s hope for Hendo if he gets that slider working, he is only 21 after all.
by T_Mizz on Feb 17, 2012 12:10 PM EST via mobile reply actions
Could you please...
put up the numbers (k-rate) of our prospects like McGuire, Hutchinson, Syndergaard…. that would be great!
Great read!
"Touch ´em all, Joe!"
by jaysfanfromeurope on Feb 17, 2012 1:01 PM EST reply actions
at your service..
Justin Nicolino: 31.6% (A-, bit of A at 19)
Noah Syndergaard: 28.6% (Rk/A-, bit of A at 18)
Drew Hutchison: 28.6% (A/A+, bit of AA at 20)
Mitchell Taylor: 26.8% (Rk at 19)
Adony Cardona: 25.0% (Rk- at 17)
Deck McGuire: 23.5% (A+, bit of AA at 22)
Aaron Sanchez: 23.2% (Rk, bit of A- at 18)
Asher Wojciechowski: 16.8% (A+ at 22)
Blogging about the Toronto Blue Jays at Bluebird Banter
by Woodman663 on Feb 17, 2012 1:34 PM EST reply actions 1 recs
thanks, woodman!
"Touch ´em all, Joe!"
by jaysfanfromeurope on Feb 17, 2012 2:57 PM EST up reply actions
I can understand the scouts view on Deck now
A bit more polished but lacking the upside. He has a strong K% but in line with his level, although, nothing crazy like The Big 3.
I can see him and Hutch contributing to the Jays by September, if they continue to perform well.
@VagabondBansal
I think, much like Alvarez
His success will be determined almost entirely by how his 3rd pitch comes along. If Alvarez can develop that Slider he has top of the rotation potential. If Wojo can develop his change then he could see a big jump in K%.
by T_Mizz on Feb 17, 2012 5:39 PM EST via mobile up reply actions
Looks promising
Until injuries strike. Hopefully not, though.
Blogging about the Toronto Blue Jays at Bluebird Banter
What about
Drabek, Cecil, McGowan, Jenkins?
by T_Mizz on Feb 17, 2012 5:37 PM EST via mobile up reply actions
McGowan: 22-26% range spread out over the minors
Drabek: 19.9% in AA, 29.7% in A+, 18-19 at other levels
Cecil: 26.9% at AA
Jenkins: 18.1% at AA, 15-15.5% at A+, 19.3% at A-ball
Blogging about the Toronto Blue Jays at Bluebird Banter
Those seem about right
I hope this means that Drabek, McGowan, and Cecil can all come back to be solid pitchers.
by T_Mizz on Feb 17, 2012 6:38 PM EST via mobile up reply actions
awesome...
Lets see the rest of the guys on the current jays staff?
Morrow: 21.6% K% (Only pitched 111 IP in MiLB; boy, was he rushed)
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
21.4% if you exclude his stint in 2011.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
Major League Pitchers with <100 IP
Would it be possible to somehow look for all pitchers with <100 IP (or some other arbitrary threshold maybe 111 to set Morrow as a threshold), and to see what kind of MLB performance they had?
@VagabondBansal
I assume that those types of pitchers are either close to MLB ready (eg. Chris Sale) or have injury problems that derail their career.
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
Or pitchers that are rushed tremendously by their parent club(s).
"We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
- Niels Bohr
Sorry, unauthorized hotlinking of copyrighted material not permitted.
Exactly
There are a few different reasons, and ways this would work. But I would be curious to see how it shows up performance wise.
@VagabondBansal
Thanks Woodman btw
For compiling all of this information.
I wanted to pose a question actually. If we look at this from the other side, do you think there a specific K% / B% / K/B / K-B / K% – B%, combination that would help provide a quick eval of a minor-league hitter?
@VagabondBansal
hitters are much more difficult
they’ll do well at one level, then fall apart at the next. Or suddenly have a breakout season out of nowhere. So I don’t dare give you a quick guideline for hitters that is like the 20% “rule” for pitchers.
Blogging about the Toronto Blue Jays at Bluebird Banter
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Related FanPosts on Bluebird Banter
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
Baseball Nation Recent Stories
In Association With
Big League Stew - MLB - Yahoo! Sports
- The Juice: Nelson Cruz matches career high eight RBIs, Pujols stays hot in Seattle
- Benches clear at Fenway as Red Sox-Rays rivalry heats up
- Keeping good company: Albert Pujols becomes fourth-youngest player to hit 450 home runs
- Giancarlo Stanton misses cutoff man, and Logan Morrison is thankful he did (Video)
- Snoop Dogg delivers memorable first pitch at U.S. Cellular Field
