Wherein I Address Krossover and Advanced Statistical Metrics

(UPDATE: Some of this stuff really make me look like an ass. I'm going to leave the original words to prove that I'm an ass (I'm not above slamming myself for being incredibly stupid), but strikethrough some of this stuff given the fact that I now noticed Dr. Piette's presentation online (and that he was more than gracious in referencing this site and Great Lax State/Tempo Free Lax. My apologies to Dr. Piette and Mr. Kulkarni for me being a dumbass; sometimes my aggressiveness gets the best of me. There is a lot of truth to bloggers being insufferable jerks. (The push of the piece remains -- that these metrics aren't necessarily new, and I'm shocked that people didn't notice them until the IMLCA Convention.))

When I saw that Krossover was presenting at the IMLCA Convention in Baltimore, I was pretty intrigued. I've been interested in seeing if someone would finally give college lacrosse the treatment that Synergy Sports Technology -- the undisputed leader in basketball video and statistical platforms -- has provided to college coaches and analysts over the years. Marrying video to statistical engines is something that lacrosse has missed for a long time, creating a blind spot in approach and dissection. Krossover is trying to penetrate a market that is wide open, but after seeing a few pieces that have come out of the presentation (from ILGear.com and Lacrosse Magazine), I'm both fascinated and frustratingly amused.

Advanced lacrosse analytics aren't a new thing. I've personally been working on them -- developing a framework and using it as the baseline for performance analysis -- for six years. I've been pushing them on this site since its genesis, freely open-sourcing a workbook with the tools over that time. Some members of the cult and I worked on putting together a more functional display of the data, launching Tempo Free Lax last season so that everyone could utilize the information in an easier way. There is nothing new as to the approach and metrics -- in fact, there's an entire section of this site dedicated to advanced analytics -- and [W]hen I see something like this . . .

Because no one's ever done this before for lacrosse, we're still early on in really trying to figure out what are those most important set of statistics that a coach or a player should be looking at on a nightly basis over the course of a season?

. . . all I can say is this: That's a half-truth built on self-promotion. (I recognize the irony of self-promoting myself right now. We live in a crazy world.) What Krossover is doing -- providing video resources, most pertinently shot charts -- is unique and I think very useful; I really hope the company succeeds in its pursuit to provide these video services. The foundation for their presentation -- advanced statistical concepts -- isn't breaking barriers, though. That's the nature of the half-truth: a simple Google search can get you to these resources in less than a second (and for free!). Now, this isn't to say that James Piette, a Ph.D. from Wharton, isn't intelligent; he is -- probably much brighter than I am -- and the way that he has distilled some of these concepts for easy consumption for coaches and others is something that deserves a tip of the cap and praise. He, like the rest of us, are working in a sea of uncertainty, trying to parse what's important against what isn't, working with correlations and dependencies that don't provide a ton of clarity (and I'm jealous that I didn't think of piggy-backing the "Four Factors" idea from Dean Oliver in Basketball on Paper. I hope that he starts to crack the code a little more (much as I hope that anyone, including myself, does the same). It's just that this issue remains: If people are blown away by these concepts, you're already behind the curve and I'm not sure what to tell you.

With that out of the way, here are some thoughts as to some items that were contained in the ILGear.com and Lacrosse Magazine pieces:

• From ILGear.com: "However, defensive metrics, especially face-off record, are only loosely tied to a team's wins, so less weight should be given to them. Even though many coaches stress the importance of clearing (or riding), Piette believes that a team's record during clears is still greatly undervalued, assuming they're being recorded correctly." I 100 percent agree -- and can prove to a degree of statistical significance that gives me comfort -- with Piette's comments about riding and clearing. I am less inclined to throw my full support behind the statement that face-offs don't matter. I think the research is unclear as to the total value of face-offs. (Admittedly, there are fewer face-off opportunities than riding/clearing opportunities in a game, but I think the follow somewhat addresses the overall issue.) Piette's underlying foundation for this -- if we think conceptually -- is that your volume of possessions are less important than what you actually do when you have (or are defending) the ball (maintaining possession by limiting turnovers and shooting well; the inverse is true for defenses (none of that should be earth-shattering)). I'm on board with that -- I have been for a long time -- but there is conflicting evidence when put into practice (which doesn't necessarily mean that dependence is lacking): Inefficient offenses (for a variety of reasons, be it maintaining the bean or shooting or whatever) need possession volume to score (the inverse is true for defenses). Take the 2012 iteration of Yale, for instance. That team was sixth overall in possession margin (buoyed by a face-off percentage that ranked third in the country and riding and clearing rates that we're all that hot (50th and 39th, respectively)) and were ranked only 20th in adjusted offensive efficiency. The Elis went to the NCAA Tournament, were Ivy League Tournament champions, and finished the season 11-5. Was it Yale's 24th-ranked raw shooting percentage, 19th lowest turnovers-per-offensive-possession mark, or the possession margin fueled by face-off wins that focused the effort? Basically, I'm in the camp that we need to be careful about definitives and that face-offs -- while the correlations and dependencies may tell a somewhat different story (and I've written and read that story) -- aren't something that should be ignored across the board.
• From ILGear.com: "The number of saves a team has in a game is a terrible statistic due to a variety of reasons (e.g., not all saves are created equal)." I agree and disagree with this. Straight goalie metrics can be problematic -- especially raw saves because there's no context -- but I think there are ways to address this, including (but not limited to) video analyses that Krossover may or may not be able to provide. (I'm hiding the ball a little bit here, but if you go through the archives you can see where I've been, where I am currently with this, and where I'm going. I think there is a workable system that can be utilized.)
• From Lacrosse Magazine: "Last season we had probably about 20 or 30 Division I programs, and our data guy was kind of spending some time looking at the D-I stats that we collected, and trying to look for trends and see if we can find anything interesting." I have said this a thousand times before: You can't do this unless you look at the entirety of Division I lacrosse. You can't sample like an auditor testing inventory and make sound assumptions as to what is actually reality. Division I lacrosse is 60-plus teams deep with different styles, strategies, and talent levels all across the board. This is a potentially major issue, although Piette's umbrella thoughts aren't significantly misaligned with what I've been working with over the last four seasons of hard data for the entirety of Division I lacrosse. This dovetails nicely to this . . .
• From Lacrosse Magazine: "The other big thing, with the colleges, they use it not just to self-scout but also a lot them use it to scout out their opponents. Before every game that a team plays, they'll give us two or three games of their opponent that they would've gotten through their conference tape exchange policy." Outside of the video breakdowns -- shot charts, etc. (the real value of Krossover, to me) -- is the statistical information being adjusted for quality of competition? If Georgetown dumps 12 goals with high efficiency on Wagner, that doesn't mean much to me from a scouting perspective when staring at an analytics report. I made this contention on Twitter and I'll make it again: Unless you're looking at analytical reports -- statistics, not the visual representations that Krossover can provide -- those reports are useless unless you are tracking all 60-plus teams and adjusting for competition played. The two to three game approach works for video scouting, but is useless for statistical analyses (unless you're focused on recent trends, which can be very dangerous).
• From Lacrosse Magazine: " From our perspective, without making any crazy predictions, we don't have all the data. We're dealing primarily with a lot of the data we create, just because our data is far more robust than anything out there, so we look at that. We don't get to necessarily look at every single team in the country. . . . I wish I could say we had data on Loyola from last year." Yeah, about that: (1) We look at every single team; and (2) No problem (of course, we have no video).

I really like the video aspects to what Krossover can bring to the Division I universe; I'm just shocked that people are looking at (or taking the position that) tempo-free measures are new. (Again, the "Four Factors" approach is nice and distillation that hasn't been written before, but it shouldn't be taken as the entire story or otherwise as a silver bullet).