|Handicapping "Think Tank" technical handicapping and statistics|
| ||LinkBack||Thread Tools|
I don't think there's a settled on "formula" for something like this. A lot of veteran handicappers just kind of eyeball it and go with a gut feeling.
The problem is that a pitcher's "win potential" is influenced by a lot of things at once....
The team he plays for...
How well he's suited to his home park...
How many pitches he threw in his last game...
If he's a fly ball pitcher and the wind is blowing out (or in)
The list could probably go on for about 10 more things.
The single best stat I've found for taking out a lot of the pollution is just IP per start. That number by itself pretty much tells you how good or generic the guy is.
7.0 or better: a very strong pitcher
6.5 to 6.9: a star
6.1. to 6.4: acceptable
6.0 or less: probably a guy who's likely to be replaced if he can't get the average up.
It's funny...a guy with a good ERA but a lousy IP per start number doesn't win as much as you'd think. If he goes 5-6 good innings, the caliber of reliever that comes into replace him often blows the game. A guy with an average ERA who can go 6.5 to 7 innings often performs better than you'd think because he gets you to the better arms in the bullpen.
If you're just looking for a simple number that encapsulates pitching...I'd go with IP per start. That one thing by itself carries a lot of weight for telling the story of the guy's performances and prospects.
If you'd like more in the mix:
*Road ONLY earned run average...because the home park can cause misleading numbers in a guy's stats
*KO per 9 innings, because this has the best relationship to long term success according to stathead literature
*Flyball/groundball ratio...because you can line that up with the characteristics of the park he's pitching in
*Pitches thrown in the last start...be wary of anyone who threw 115 pitches or more in their prior outing, with the exception of the big solid veterans who have established they can do it (Randy Johnson, Clemens for awhile but not lately, Livan Hernandez)
Does that help?
I hope I didn't go through all that because someone was giving us a hard time!
Since the thread is in motion anyway...anyone else have ideas on how to "rate" pitchers? Now's the time to get the numbers ready for the season because it willl be here before you know it.
the key for me is recent form versus overall form.in april how much of last year counts?,in june do we completely discard last year and in august how much of april's stats matter.let alone the lefty-righty numbers and what kind of lineup was faced in the prior game and home-road variances etc,etc,etc.how about injurys or the rare horrible game that everyone throws?let the headaches begin.....
I used to drive myself crazy with stuff like that. But, I found one year that using a guy's career number had as much predictive value as any of the short term stuff. So, I went with that because it was a million times simpler.
A lot of those "last start," "last 3 starts," and "last 5 starts" end up regressing to the mean anyway. You might as well just start with the mean and go from there.
If you start with career numbers (ERA and IP per start), then adjust for ballpark and the caliber of that night's opponent, you'll pretty much be in the right neighborhood for the best expectations. The only exception would be when a guy is hurt and trying to hide it...or when a guy is coming off a big pitch count. That's when the shorter term numbers have better predictive value.
I think a lot of the "hot" and "cold" appearances from the numbers in the smaller sample end up being like runs on black or red in roulette. They lead you in the wrong direction.
Also, one bad game can mess up the ERA number and create the illusion of a cold pitcher.
Imagine this scenario:
Game One: 7 IP and 2 ER
Game Two: 3 IP and 7 ER
Game Three: 7 IP and 1 ER
Over the three games, that's an ERA of 5.29 on 5.7 IP per start. That suggests a struggling pitcher. But...he's actually 67% to hurl a very strong game, and 33% to get rocked.
Game Four: 6 IP and 3 ER
This brings the ERA over four games to 5.09 on 5.8 IP per start...but the guys actually 75% to hurl a quality start and 25% to get rocked. I think short term pitching numbers tend to underestimate the probability of a quality outing, and overestimate the chances the guy will get hit hard.
Or...he's got a 75% chance to be a pitcher that's 2.70 on 6.7 IP per start...and a 25% chance of not having his stuff. A "short term" ERA of 5.09 doesn't tell that story at all...and in fact creates the illusion that the guy isn't pitching well.
Something to think about. At the very least using career numbers is a whole lot easier!
dinque and blogguy---
Do you think that the last three starts for a pitcher are still the most useful info in handicapping?? Or do you think that since it is so used its not of any use at all???
Buzz, I dont go to games. I buy all the Directv packages and watch them from the comfort of my own home! I dont like listening to all the fans nonsense at games! I pay for blonde women to come over and have sex with my hispanic hottie maid, and sometimes I get involved to make it a threesome! I like to lay in my pool during the day sipping on drinks that have umbrellas!
Boy Razz...it's been a long time since I thought they were useful at all. I don't think I ever thought they were the MOST useful. The only time I'll give them weight if the IP numbers suggest a big name guy may be pitching hurt.
As I outlined above...I think the numbers can be warped by one bad performance. So...that a guy that's 67% likely to pitch very well can look like he's 100% to pitch horribly because his ERA got mangled in a bad start.
I'll say it this way...half the people I know use the last 3 as a guideline for what to expect...and half the people I know use the last 3 as an INVERSE indicator figuring that if the number's bad it's about to get good, or if the number is good it's about to get bad.
To me...that means they don't tell you anything that really has predictive value. That's just people playing "the hot number" or "the due theory" on a roulette wheel.
The year I did a study...the career numbers were much more accurate for what was most likely to happen on a given day than what happened in the last three. If you start with career ERA and career IP per start...that's going to get you in the ballpark of the best expectation as often or more often than anything else will.
*Start with career ERA and IP per start
*Adjust for the ballpark effects in that game's stadium
I've also spent some time with flyball/groundball stuff. But..it's best to stick with the extremes there (go against EXTREME fly ball pitchers in the bandbox parks...them them in the spacious parks like San Diego and Washington). This year I"m looking forward to that "finesse" and "power" dynamic that's been discussed in earlier threads.
Do you like using the last 3 starts? Have you had the most success taking guys who had been doing well...or going against guys who had been doing poorly? I had really bad luck trying to go against guys who looked like they had been doing poorly. That's why I dug deeper and saw that it was often one bad start causing the trouble. I was going against guys who were 67% to pitch well. Recipe for a migraine...especially if you were tying them to total plays as well...or "first half" five inning lines at places that offered that.
Don't forget the bully behind every starting pitcher you're looking at. Sure, one of the most underrated issues in my opinion.
"Elections are not decided by those who vote but by those who count the votes"
Great point DG.. Too many times the starting pitcher does his job only to have the middle guys and even the closers come in and give it up.
Maybe on the 1st 5 innings it would be stronger.
Let hatred turn into friendship because of your existence.