Huston Street on (Imperfect) Stats
Huston Street isn’t an expert on analytics. Nor does he claim to be. But he’s by no means a neophyte. The veteran closer is more knowledgeable about advanced stats than the average player. He also has some strong opinions.
A little over a year ago, Street shared his negative view of FIP in one of my Sunday Notes columns. This past spring, I approached him at the Angels’ spring-training facility, in Tempe, to get his thoughts on shift strategy. I ended up getting a lot more than that. A simple question segued into what might be best described as a stream-of-consciousness look at the state of analytics, in classic Street style.
———
Street on defensive shifts: “I’m a big believer in match-ups when it comes to shifts. When they’re done really well, they’re based on the individual match-up and not on a general approach to shifting. Half the league throws 96 and the other half throws 93. Then there’s me, who throws 90.
“A lot of pitchers, especially veteran pitchers, can tell you what happens on their stuff — where balls get hit. Good coaches and front offices can identify that. They can see how a guy’s stuff plays over time. I’d be more interested in spray charts off the pitcher, and shifting off of those, rather than the hitter’s charts. By and large, pitchers give up hits to the same spots. A pitcher like me… if guys want to be successful against me, they have to hit balls a certain way.
“I think we’re going to be extremely successful with that this year. We have some brilliant baseball minds here. We have guys like Mike Scioscia, Ron Roenicke and Bud Black, not to mention the staff Billy Eppler has put together. I think we’re going to see the Angels have a pretty solid, positive delta in that area.”
On FIP, WHIP, and improving imperfect stats: “I’ve seen a lot of really bad stats. Some of them oversimplify. I’ve yet to see a stat that actually predicts future success. WAR is a self-fulfilling prophecy. FIP is the worst stat ever. I don’t think it’s fair. A player starts off at a baseline, which is based on a quotient, which is based on everybody else’s performance. His is only 1/1,000th of that percentage. If he’s the absolute best player in the league, that makes him seem less amazing. If he’s the worst player, it makes him seem less so.
“I don’t believe in any pitching stat that doesn’t take into account singles, doubles and triples. That’s mainly why I don’t like FIP. It totally negates singles, doubles and triples.
“I think WHIP is a phenomenal stat — it’s obviously simple — but it could be expanded on. The higher your WHIP, the more your strikeouts need to go up. You can create a ratio there. If you have a higher WHIP, but a much higher strikeout rate, you’re probably going to survive. If you have a lower strikeout rate, you probably need a lower WHIP to survive. Then you could even add the number of doubles a guy gives up. That’s what scores runs a lot of the time — the doubles you give up when you’ve got a guy on first. You want to correlate these things.
“I’m sure that’s what FIP thinks it does. The problem with FIP is the arbitrary values assigned to home runs, 13, walks and strikeouts, two and three. Baseball is a sport that is divided and computed to the 1,000th. If you bat .299, that last little digit is one more hit out of 1,000, then you would have hit .300. It becomes hard for me to understand how a stat that is so widely accepted would have three whole numbers. Not 13.647, or 12.983, but just 13. There’s no possible way that it’s 13.”
On appreciating analytics: “I love stats. Don’t get me wrong. They can be very helpful. That’s how the stock market works. Those are some of the smartest people in the world and they all use extreme analytics. I mean, that’s how Facebook works. Numbers, analytics, predictive behavior… I just think that baseball is kind of in its infancy stages with that. But again, we’re talking about a massively helpful tool, when interpreted the right way.
“If somebody reads an ancient text and interprets it poorly, you get a bad definition. If somebody takes a whole collection of numbers and scrambles them up together, you can get some misinterpretations. That said, I think the Angels are well positioned to read the numbers the right way. But again, I’ve yet to see a stat that truly predicts future success.
“That’s what I’d like to see. I can think of some ways to do it, but it’s still a very imperfect science. It’s an inexact science. ‘Will this person be able to sustain what he’s doing?’ is difficult to do. That’s something I’d need to educate myself on better. I’m sure some of those analytics already exist, and I’m sure the front office already has them, but I don’t know them. That’s not something they share with players.”
David Laurila grew up in Michigan's Upper Peninsula and now writes about baseball from his home in Cambridge, Mass. He authored the Prospectus Q&A series at Baseball Prospectus from December 2006-May 2011 before being claimed off waivers by FanGraphs. He can be followed on Twitter @DavidLaurilaQA.
That was the smartest “You nerds don’t know what your talking about” rant from a ballplayer I have every read.
Mostly because he’s saying you’re not nerdy enough.
“FIP is the worst stat ever”
“I think WHIP is a phenomenal stat”
It sounds like what he wants is something like wOBA-against, which would be fine except it doesn’t isolate him from the fielders behind him.
I think the issue is that WHIP does and says exactly what is says it does and says, no more no less. It tells you how many baserunners (minus ROE and HBP) a pitcher allows an inning, which is certainly useful information and really hard to misconstrue. While FIP might be a better indicator of overall pitcher quality, it fails to do what it’s purported purpose is, which is to isolate and properly weight the things a pitcher controls. I think that’s what his problem is with it if I’m reading it right.
He doesn’t sound like he understands the stats. Maybe they’ve been poorly explained to him.
Eg., his comments about the FIP scaling factors — those are just to make it somewhat comparable to a familiar scale (ERA). He also seems to strawman it a bit; FIP’s purpose is to measure what are primarily in the pitcher’s control, versus events that other players impact. Although the creators and FG staff are very clear that it’s not predictive; all it measures, all it purports to measure, is Fielding Independent Pitching.
Rather than criticizing a guy who is paid to throw a baseball, we should wonder who explained these statistics poorly to this guy. If this is how advanced stats are being explained to players, no WONDER they are anti-stats.
This is also important for explaining sabermetrics to the average person. There are lessons our community can learn here.
Considering Street is basically a contact management reliever rather than an overpowering one, what he is saying makes complete sense. From 2012-now, he’s 11th in ERA, 12th in WHIP, and 114th in FIP out of 226 qualified relievers.
Yes, he likes the stats that make him look good and doesn’t like the stats that make him look bad. It’s completely understandable!
The fact that he has a horse in the race doesn’t change the fact that FIP doesn’t measure contact management on balls in play.
Would someone please offee anexplanation of what he means when he says “WAR is a self-fulfilling prophecy”?
Is he saying that if a guy is good he’ll have a high WAR? Cuz not all WAR are created equal…but then isn’t he really saying “WAR seems to be a very accurate way to value a player”?
I believe what he means by this is by starting out with a list of stats you think are good and productive, and then using those stats to formulate how good or productive a player is, you aren’t open to other possibilities.
For instance, if someone says they think pitches seen/PA make a player more valuable, and then formulates their own WAR to include that, they’ll be arguing in circles.
“See, his WAR is high and sees a lot of pitches, so naturally seeing a lot of pitches is good thing.”
Not that I agree with him or anything, but I think that’s what he means by that. WAR is obviously a little more complicated than that, but I think that’s his general point.
Thanks for that. That example of P/PA was perfect.
nice take Huston. #respect