King Felix, unique
I’ve been kicking around what to say about this, but… the new PECOTA cards are out, and if you look at Felix Hernandez… well, let me take a step back.
To over-simplify, PECOTA takes a player and uses a bunch of criteria to find similar players in baseball history to compare them to, and then looks at what those guys did the next year. So in doing this, it comes up with a score on how well it did finding similar players.
From the glossary:
Similarity index is a gauge of the player’s historical uniqueness; a player with a score of 50 or higher has a very common typology, while a player with a score of 20 or lower is historically unusual. For players with a very low similarity index, PECOTA expands its tolerance for dissimilar comparables until a meaningful sample size is established (see Comparable Players).
Historical anamoly Ichiro (who the system predicts will be out of baseball in 2010) gets a 19.
Felix gets a 1. 1.
That is the craziest group of “comparable” pitchers I’ve ever seen. It’s like the names were plucked with the assistance of a bingo caller.
He’s listed at 170 lbs on the PECOTA page. Being on tv must really make him look heavier.
And yet it grabbed 20 of them. Someone like Roger Clemens (similarity score 2) gets 8 comps and then PECOTA gives up.
For some fun comps, compare Roger Clemens to Jamie Moyer. They basically have the same group of guys in a slightly different order.
Ichiro’s score is artificially high. What these are really saying is how confident PECOTA is in its projection. A low similiarity score means that PECOTA is well aware that it has no idea what it’s doing. By giving Ichiro a higher score, PECOTA thinks it understands Ichiro.
As a counter-argument, I point to PECOTA’s previous lists of comps for Ichiro, which included such luminaries as Luis Polonia, Tony Womack, and Damaso Garcia. Clearly PECOTA’s been all over the map when it comes to Ichiro. It’s confidence seems unfounded.
I think the comparables aren’t bad – it’s just tough to get ‘accurate’ comps when your criteria are pitchers who’ve had MLB success before turning 20. You’re basically taking the entirety of that set, and not breaking it down by GB or FB pitchers, strikeout guys, finesse guys, etc.
That said, his comps are better than Ichiro’s, which contains a group of scrubs. I’d take Blyleven, Jim Palmer and Milt Pappas any day. Kazmir and Ankiel have to be there, though I’m surprised at Gooden’s absence. I’m just amazed at how much better Felix’ cup of coffee was than the other guys like Blyleven. Couple that with Felix freakish GB rates, and yeah, maybe he should be historically unique.
That said, wouldn’t you expect this number to shoot up as he gets more innings (and gets older, of course)? We’ll see where it settles in when he’s 25-26.
I just don’t get how Ichiro is less unique than King Felix.
There are more position players than pitchers in the DB, for one. Also, there are a few Ichiro analogues out there, though they’re either slap-hitters of the dead ball era (Wee Willie Keeler) or guys who finally made it to the bigs at 28.
Felix’s analogues are almost entirely guys who pitched as teenagers (Blyleven), recent sensations (Greinke), or superphenoms that bombed (Ankiel). That’s extremely scattershot.
If PECOTA choked on matching Ichiro, then it swallowed a gallon of Crazy Glue by comparison when trying to match up Felix.
Again, though, we should point out that “20″ is a really low score: that’s not an indication of the system’s confidence at all. “50″ would be. 20 is still a huge disclaimer
The guys they NEED to be using to comp with Ichiro are guys like Tony Gwynn, Heinie Manush and Paul Waner- fast corner OF’ers. (If they threw in his J League numbers and adjsuted them for MLB, I bet they’d come up as comps.)
And my comps for Felix would include Tom Seaver. The problem is Tom Seaver came up a bit later and didn’t have a partial season…
Here’s one that occurs to me – a hard-throwing strikeout pitcher who broke in relatively young:
Felix 2005 (selected)
Age 19
W/L:4-4
IP 84.3
K/BB: 77/23
ERA (unadjusted): 2.67
STF: 53
Player A (cup-of-coffee year, 1966)
Age 21 (prime: 27-37)
W/L: 3-3 (career high: 27)
IP: 52.0 (career high: 346 at age 27)
K/BB: 25/18 (career best: 310/87)
ERA (unadjusted): 3.12 (career best: 1.97, career total 3.22)
STF: 4 (career best: 39)
Player A is Steve Carlton, who probably doesn’t show up as a comp for Felix because he’s a lefty. I am amused that PECOTA rates Felix’s stuff at age 19 significantly better than Carlton’s in his prime.
Derek/whoever:
I’d like to preface this by saying, I am HUGELY excited about King Felix. I think he’s going to be a great player for the Mariners if they don’t destroy his arm, and if they manage to keep him in Seattle. Neither of which has anything to do with what I’m about to ask, though.
How much of stuff like this is due to the fact that we still have a relatively small sample size of results for King Felix? Does PECOTA and such take that into account somehow?
Or am I mistaken, and we have a larger sample size than I think?
Minor league stats count.
Okay, that’s what I was thinking but I wasn’t sure. I don’t know that much about the mechanics of PECOTA.
Is there some sort of a translation that happens to take out the tendencies of the league that the stats are from, or does it normally even out over time?
Perhaps I should, instead of asking here, try to google for a good explanation of how PECOTA works.
The first choice, and that’s probably a good idea.
Last season, I did a personal study of pitchers in the last 35 years to attempt to find someone like Felix, but it was to no avail. Kevin Brown was really the only one in the statistical vicinity, but he was so much older when he started being successful. Other than Brown, Felix’s success was wholly unique in the pitching realm of the last 35 years worth of data. The real question is, does that uniqueness also make him the best? Here’s my old blog entry on the subject (although the blog is long since defunct).
http://shortporch.blogspot.com/2005/08/king-felix-best-pitcher-in-last-35.html
Sexson’s card is pretty interesting. His similarity score of 28 seems lower than it was last year, as I recall. His 10% EQA is .279, so even if almost everything goes wrong, he still looks like a solid bet to be an average hitting 1B.
The fairly weird thing, to me, is that it has his defensive value decreasing as his offensive value increases. It probably doesn’t really mean that much, since BP’s defensive statistics seem rather unreliable, but I think that’s the first time I’ve seen that on a PECOTA card. (Okay, after looking at a few more cards for this in particular, I guess it’s not that uncommon, but it did surprise me a bit.)
I think that’s because historically weaker hitters have been slicker defenders, and PECOTA is looking at historical comps. So, the 90th percentile projecion is if the player looks like the best of his comps, while the 10th percentile is if he looks like the worst of his comps.
The worse of his comps were probably better with the glove than the best of his comps.
This is all just speculation, though.
Regarding Sexson’s defense, it might be playing time, as plate appearances increase along with the percentiles.
Something I noticed about Felix’s card. PECOTA projects him at 55% GB rate. Felix pulled a 68% GB rate in SEA last season. If he somehow maintained a GB rate in the 60s, you think the jump in projections from 05 to 06 was big? He’ll project to be the greatest SP ever by this time next year.
“Regarding Sexson’s defense, it might be playing time, as plate appearances increase along with the percentiles.”
I thought about this at first, but his defensive runs over replacement (or maybe average, not exactly sure which they use there) winds up going from a positive number at 10% to a negative number at 90%, which made me think it wasn’t just a playing time effect. I think Evan’s probably got the right idea with historical comps tending to be either mostly hit, less leather or less hit, mostly leather.
#11, I assumed that’s part of why his sim score is so low, and why he’s so highly touted. Has there been a pitcher quite like him in AA and AAA, at such a young age. Utter dominance, and then to go to MLB and do more of the same. I think most people agree, combined with scouting (what ammount scouting does it take, There were nights I’d personally never seen stuff that good). To me the questions are, How much does his GB/FB take a hit (or do people start figuring him out, a little)? Does his control problems reappear? Does it get to his head? Does he stay health? If he doesn’t run into those issues, he’s going to dominate, like nothing we’ve seen. We had Randy, and that was otherwordly to watch. Felix (looks to be) might give us more of that. Even with some BB/9 added to his stats, he’d still be an Ace because his pitches are so good. My opinion. I dont see much alarm (barring injury), and see lots of bright spots (even in ’06). Best case is best of the very best, worst case, a good pitcher.
Barring injury is a pretty big caveat. He is a young pitcher in the M’s organisation – he’s already facing pretty long odds on the injury front.
Agreed fully, but he’s soooo good!!!!
Anyone know much about Chaves program? The one thing I like the most about Leo Mazzones philosophy is “Back off” on the effort and pitch more often. I’ve always loved that philosophy for any physical activity. It train the nervous system better than long rest periods, it minimizes muscle and ligament rebuilding, allowing for more activity. I’ve followed Pavel Tsatsouline’s strength training principles for years, applying it to all kinds of physical endeavors (hiking!!!). It makes so much sense, and there is no reason it shouldn’t apply to baseball. If you practice what you do physically, but not to fatigue or failure, and do it very often, you’ll do it better, with less risk of injury. I’ve noticed many of the players with long careers practice daily, although not universally. Roger Clemens and Edgar Martinez come to mind. Edgar swung at balls constantly when he wasn’t playing. Ichiro reminds of similar conditioning. He practices ALL his plays every day. that makes sense. He makes sure his body is fit for THOSE activities, and strengthens the neurological connections to his muscles. I think that’s a big big reason Mazzone has been so successful with so many different types of pitchers, at least making/keeping them healthy.
Also, notice many good pitchers who remain good and healthy lay off the heat until they “dial it up”. Schilling comes to mind. Randy Johnson did a lot more of that this year than what I remember “back in the day”. It makes sense, if you can get away with lower velocity, do it, when you can’t, bump it back up.
Anyway, my two cents, but I think it’s interesting to think about
Throw more often. Not pitch more often.
Throw more. Pitch less.
Ichiro out of baseball by 2010? That’s only 4 years away. He will be only 36-37 years old. No way.
Ichiro can play till he is 40 at least.
I suspect that means that PECOTA thinks that Ichiro will lose his speed within the next 4 years and without his speed, his hitting style is toast.
I suspect that PECOTA projections for 2010 are to PECOTA projections for 2006 as next week’s local weather forecast is to tomorrow’s local weather forecast. Or, in SAT notation:
PECOTA 2010 : PECOTA 2006 :: Next Week’s Weather : Tomorrow’s Weather
Maybe the “1″ is meant as #1, not as a similarity score. j/k
26: They don’t do analogies on the SAT anymore.
I would think one somewhat similar comparison might be Brandon Webb, except that Webb has a few more control problems and has only been in the league himself for a few years.
It’ll be interesting watching their careers tho.
I don’t know much about the PECOTA system, but looking at Felix’s card (and taking the numbers at face-value), should we be surprised to see a BABIP figure of around .290 for his 50% scenario (the prospectus dictionary defines this as “a typical BABIP”)? I would think that with all this talk of Felix being an extreme groundball pitcher, M’s have a good infield defense, etc., that the figure would be significantly lower in even the 50% case………can anyone help?
Just some random comments/responses:
1. I think PECOTA handled Felix pretty well. Mark is right that at some point it gives up and has to start picking off random 20-year-olds who were somewhat okay pitchers in some area or another, just so that it can get a decent enough sample size. But the pitchers at the top of the list are weighted more heavily, and they seem like the right guys.
2. Gooden didn’t make it because of the way that PECOTA handles player ages. Gooden was an “old” 19 in 1984, whereas Felix was a “young” 19 last year; the difference is enough that it shifts Gooden’s seasons forward by a year. But he should be near the top of the charts next year.
3. Groundball pitchers give up higher BABIPs, all else being equal. That Felix is projected to have a BABIP of league average or a little better in spite of that reflects the M’s good defense.
4. The groundball numbers that we use for the PECOTAs (and in this year’s book) are different than what you’ll see elsewhere, in that they include linedrives and popups in the denominator. Q&D you should add about 10% to our numbers to get the “right” figures.
5. Felix’s listed weight, though comically inaccurate, doesn’t have very much influence on his projection. Height is more important for pitchers (though weight is more important for position players). Besdies that, it gives up on trying to worry about something like weight when it’s having this much trouble finding comparables.
6. I don’t want to get into the Ichiro debate.
“They don’t do analogies on the SAT anymore.”
Yeah, I forgot about that. It’s a shame, too, the analogies were kind of fun. It’s nice to be able to have a little fun while your academic abilities are being poorly evaluated. I think it’d be nice if they added 3-4 significantly harder questions on the math section, too, to help separate the top 3-5% more accurately.
“3. Groundball pitchers give up higher BABIPs, all else being equal. That Felix is projected to have a BABIP of league average or a little better in spite of that reflects the M’s good defense.”
To add to that, even if Felix is allowing a few more groundball hits, those aren’t going to hurt you nearly as much as the flyball hits (just ask Eric Milton) and even if there are a few more guys on base thanks to seeing eye singles, Felix’s groundball tendancies should help increase the chances of a double play (just ask Carlos Silva). Not all BABIPs are created equally.
Interesting, Nate. Good stuff.
Uh, I hate to point out blasphemy, but Ichiro!! can’t be pleased with being called Ichrio. One of the few typos I would ever gripe about.
And all hail King Felix, he is currently the highlight of my life as a professional sports fan.
btw, the M’s site has their Outfield round-up today, with lots of love for Ichiro! before moving on to Reed & Ibanez
[long link]
Or maybe I would, after all.
Felix might contend for a Cy Young?
In other news, Seattle likes coffee, it’s cold in Siberia, and the guys who invented google have a little bit of money.
Kurkjian must have really been hurting for column fodder.
I assume that refers to this link? On a similar theme, Tom Verducci includes El Rey as one of his potential breakout pitchers in SI
“it’s cold in Siberia”
what are you talking about? It’s only 68 below with the windchill in Yakutsk right now.
Windchill is for wimps. If you’re dressed for it, it’s only the air temperature that matters.
Saying “this is how fast you’d freeze if you went outside without a coat” – which is basically what windchill does – is like saying “this is how fast you’d suffocate if you went into space without an air supply”. True, but pretty pointless.
Evan, as long as any part of your skin is open to the ambient air, wind chill is a factor. Even if it’s just your nose, convection will ruin your day!
I would assert that being dressed for it presupposes that you not have any exposed skin.
If it’s below -40°, and the coldest part of your body isn’t your feet, you dressed poorly.
getting (somewhat) closer to the original topic, if you haven’t seen Venezuela’s unbelivable win in the finale of the Caribbean series, check it out:
http://mlb.mlb.com/NASApp/mlb/mlb/events/mlb_winterleagues_2006.jsp
You know what’s great for protection against the element? High quality clothes with fashionable USSM designs! Buy one or more today!
you know, I hear that the Organic Cotton is lightweight for summer comfort or winter layering….
Nice one, Derek.
As my father always says, “If you always stay warm & dry, you’ll never be wet & cold.”
[ot troll]
and my favorite tautology
“Ireland at war will never be at peace”
I guess that’s OT.
Anyone know if they’re planning a “Felix home starts only” 16 game plan for season tickets. It’d be a hot item.
#50–and my favorite tautology “Ireland at war will never be at peace†I guess that’s OT.
yah, but I sure could hear it stated as a Fairly Obvious Fact.
it’s less tha a week for pitchers & catchers…. sigh. hey, the MLB spring training page now has an Opening Day countdown! oh, and an AL West overview, with Tom Singer projecting the M’s in 2nd place in the division….
msb- I am surprised that Singer doesn’t see better things for Texas. Though its nice to see Seattle not relegated to the basement automatically.
The team with obvious weakness is the Angels. Their offense has some big holes (basically everyone except Vlad).
I’m not sure I’d be so down on Texas, and I also don’t think Washburn and Everett are positive additions and Singer seems to.
Still, if things break right, the M’s probably do have a shot at second. After Oakland, the division looks weak.
Note that Singer, in his AL East overview, predicts Toronto to win that division. So I’m fairly confident I don’t trust his analysis.
From Singer’s article…
Moyer and Washburn should be a potent left-left punch around King Hernandez, especially at Safeco.
Moyer’s 2005 xFIP: 5.17
Washburn’s: 5.01
Their FIPs are pretty close to each other as well. I’ll be thrilled if they both end up with Moyer’s 2005 ERA (4.27), combined. While that would be a big improvement over the Franklin/Sele debacle of 2005, I’m not sure it wins a division, or gets you in second place.
xFIP normalizes HR/FB by park, right? What about hits and runs for FB in play? Doesn’t safeco affect? I belive the THT annual points this out. They should have lower ERA’s than even xFIP (which park neutralizes the HR/FB if IIRC). Doesn’t mean they’ll be good, just pointing that out.
I’m personally expecting ~4.5-5.00 for most of our rotation, with the backend being 5+. Better than last year, but mostly because of the King. Maybe a little bounce back from Joel, but not to his peak, to his expected performace. Now if they had gotten some LF defense …