Drance Numbers is the silly research wing of PITB. While Messrs. Wagner and Mooney blog nationally and solve mysteries, Drance Numbers will look into the minutiae of quantifiable NHL data and bore you with it every Friday. Today, Drance chats with Brian King, the inventor of PDO.
A few weeks back, when I wrote about PDO in this space, I mistakenly credited Vic Ferrari, widely regarded as the Bill James of hockey, as the creator of PDO. I’ve previously read that PDO took its name from the internet alias of the first person to suggest the number’s existence and relevance, but whether this slipped my mind, or I’d just assumed that Vic Ferrari was the first to flesh it out, I can’t say for certain. Either way what I wrote was incorrect.
A few days after, I received an e-mail from Brian King, the true inventor of PDO. He complimented me on my take, but corrected my erroneous statement. I figured I’d ask him if he was interested in speaking with me about the metric, and how he created it. What better way to run a correction? What follows is the conversation I had with Brian King, aka PDO.
TD: Brian, all too often Vic Ferrari rather than yourself is credited with having created PDO. In fact, I’m guilty of that myself – why do you think that is?
PDO: It’s probably because I came up with it on his site, I guess. I was in his comments section and just kind of threw it out there because he was talking about on-ice save percentage and he talks that about it quite a lot. And he talks about on-ice shooting percentage quite a lot too. And he’d talk about them separately, and he’d talk about guys having really good luck or really bad luck. It just seemed to me you could have a guy who had really good luck on one end and really bad luck at the other that it could cancel them out. So I think in the summer of 2008, and the funny thing was when I added it together I didn’t pay a lot of attention to the guys who were really close, but I looked at all the Oilers who were over 101 and under 99. The funny thing was almost every guy over 101 got an extention and every guy under 99 got traded or wasn’t re-signed. So I looked and I thought okay, well, that could just be a coincidence, or it could be that guys just couldn’t buy a bounce at either end and that’s the way it goes.
TD: so you were the first to suggest the existence of PDO in the comments section of Irreverent Oilers Fans, were you ever tuned into the idea that it would regress to the mean over a large enough sample?
PDO: Well, I knew that on-ice save percentage did, and with shooting percentage it does regress a bit but not as much – because, if you’re Dany Heatley, you’re likely to have a higher on-ice shooting percentage. So the reason it regresses is mostly because of the save percentage aspect, but it still separates if a guy is getting a lot of bounces. I mean, what’s Phil Kessel shooting right now 30%? He’s off the charts, he won’t keep it up, but small samples sizes make things fun.
TD: On the subject of small sample sizes, PDO has become something of a hot number even among more mainstream hockey writers in the early going this season, partly because it allows us to qualify our opinions and figure out who is likely to regress… Are you pleased to see your number gaining this sort of traction and catching on?
PDO: Yeah, it kind of surprised me actually. The first time I saw it was James Mirtle on Twitter, and he was talking about Kessel and Lupul, who had the two highest PDO numbers in the league. I think part of the reason this has happened is because Toronto, biggest hockey market there is, has had this huge start. And the Canucks too have kind of been on the other end of it, just not being able to buy a bounce. So, to me, a large reason this has happened is there’s two big market teams on opposite ends of the spectrum. Hockey stats in general have been picking up, but with PDO, it’s nice to see, and I think there is value in it.
TD: Why do you think PDO has proven to be more accessible to people?
PDO: I think it’s because it is simple, I mean, it’s not rocket surgery right? It’s a really simple thing to do, it’s just two numbers added together and that’s it. Intuitively it should be about 100, or 1000, depending on how you’re doing it (and it doesn’t really matter). So I think it’s an intuitive stat and an understanding of it comes naturally. Basically anyone who has ever played hockey can relate to the idea that “sometimes you just don’t get the bounces,” and over time you like to think they work out and that’s basically what the stat says.
TD: I’m curious about your take on recent converts to applied analytics in hockey, do you still keep abreast on recent writing on the subject?
PDO: Yeah I try to follow along. I read mc79 a lot and Desjardins, I read him a lot. Anything else is something I’ll stumble upon on twitter. It’s always interesting to see the newer takes – you can only read the same thing so many times. Sometimes you get somebody new on it, and they have a brand new idea.
TD: What are your thoughts generally on the direction of independent hockey coverage, do you think it’s improved?
PDO: I think it’s vastly improved, there are so many more sites than there were even three years ago. And a lot of the MSM stuff, they’ve gotten a lot better recently with guys like Mirtle are paying attention. But five years ago I got bored of reading the same story with different names attached. “Well, they didn’t try hard enough,” and whatever other cliches you can think of. So yeah I think it’s way better, because now you get a lot more in-depth analysis.
TD: Are you still actively thinking about new metrics? Do you have anything interesting you’re chewing over right now?
PDO: I’ll look into things from time to time, but with most of the stuff I read, I’m just keeping up.
TD: Any particular area of contention – say shot quality – that you’re paying attention to?
PDO: Shot quality is important and I’m intrigued by it. I mean, how much of a difference does it make? Are you better off shooting from “better spots” or are you just better off shooting more often? I find that question very interesting, and the other thing is chances given the score. That’s always drawn my interest, generally speaking a team with the lead will start to get out-chanced. I’ve always been interested in that because it’s obviously a coaching a decision… I’m just not sure why they do it, coaches are too conservative in all sports.
TD: Yeah it’s like the prevent defense in football…
PDO: Exactly, it just doesn’t add up for me. The Oilers were playing New York a couple of weeks ago on Hockey Night in Canada, and New York gets a powerplay in a one-goal game with three minutes to go. Torts pulls his goalie with a minute thirty to go, and the play-by-play guys start talking about how “aggressive” it was. To me, it’s like, why aren’t you pulling him with three minutes to go when you have the powerplay and the offensive zone draw? Your chances are just so much better of scoring the earlier you do it.
TD: Yeah, that’s one of those counterintuitive ones. Like not laying down a sacrifice bunt in baseball.
PDO: Yeah, if you’re going to pull your goalie half way through the powerplay, why not do it at the start of the powerplay when you have your best players on? Just take your best chance at tying it up, because your best chance is right there.Tags: drance numbers, pdo