There's been a lot of buzz about the Stars' use of advanced analytics lately, and for good reason. It's odd to see veteran coaches and long-time front office guys like Lindy Ruff and Jim Nill react calmly and even knowledgeably when the conversation turns to more innovative tools for measuring a players contribution to the on-ice events. There is still plenty of debate (as we are all very well aware) about the usefulness of the more recent metrics, though; even Toronto's new assistant GM Kyle Dubas would be happy to tell you just how inferior an isolated Fenwick score is when you have the ability to watch and perfectly document and compare every scoring chance, turnover and successful breakout across 82 games or more, all while tracking every player involved. In fact, you don't need to ask Kyle Dubas, because Lindy Ruff has been talking about the Stars' methods for collecting that data for a while now. One such instance was seen Josh wrote this post from January. Here's a pertinent excerpt from Ruff's interview on the Ticket:
"A common stat that is good for us is really the chances inside that certain area that we keep," Ruff continued. "There's this Pucks! Scouting System that keeps- and I'll reference that because we have that and I'll look at it- They give out a lot more chances that for me are low percentage."
Here you have a man whose job depends on putting players in the best possible positions to succeed, and he is revealing that his staff chooses to track critical things like scoring chances using, "computer technology." I am not sure why this is remarkable, since most all of us depend on computers to analyze data in our workplaces all the time. But don't forget that last sentence in that quote. The coaches use this data, but they have noted that the program tends to count more events as valid scoring chances than they would themselves classify as such. I am glad they are not just taking those numbers for granted, because this is all fairly new technology, and we should be cautious. They are.
Of course, despite this revealing information, much of the hockey world was still shocked (or maybe just bored and desperate) to hear about Nill's penchant for employing technology to aid the team in assessing their players. I suspect this is more a product of the convergence of Dubas's hiring in Toronto, perceived as a removal of the previous management who didn't employ advanced analytics from an original-six franchise, and Dallas's rising prominence as a contending force in the ultra-competitive Western Conference. Also, it is summer and everyone is desperate to spark debate about anything. Speaking of which, check out the poll at the end of this article!
The latest entry in the Dallas Stars Moneyball Computer saga is Michael Florek's article today on the DMN, which gave us even more insight to what we can assume is the same Pucks! Scouting System that Ruff referred to in January:
Nill and his staff now use a computer program that does the work of at least 100 college students: it measures Corsi, giveaways/takeaways, scoring chances and a host of other stats. After each game, the coaching staff reviews the numbers then goes back to the film and tracks its own stats to look for any discrepancies.
It's good stuff, so read the whole piece. Again, Nill emphasizes that they are on the look out for discrepancies before using the data. This is the way of all new technology. If your factory's new computer system tells you to start assembling five million pallets full of oil for woodchuck fur, then you know that either the system has flaws that need to be worked out or you are working at a very weirdly-specific pet supply company. But probably, it's the system. And that's where it becomes much easier to trust Nill, Ruff and their teams when it comes to implementing things like advanced analytics. They are not the type of people to jump on board a shiny new metric just because it's complex; if Fenwick told us that Sidney Crosby was terrible and Niklas Hagman was a generational player, then probably the Stars wouldn't be utilizing any such statistics. But they are, by all accounts, doing what one always ought to do in any study: looking at the results while trying to get rid of the noise. As we all know, hockey presents probably the toughest challenge for doing so due to its ongoing action and simultaneous events that make cause-and-effect extremely hard to parse in many instances. It is an exciting time to be having this discussion, and it's encouraging to hear about how hard Dallas is working to gain a better understanding of what leads to sustained success.
One other thing that relates to this topic came to mind when I was reading Puck Daddy's take on the DMN story:
Corsi and fenwick are the most widely used metrics to measure puck possession, and rely on the stats reported by the NHL, which has four people tracking stats at every game. As technology advances, in theory so do the accuracy of those stats: The, for example, use a computer program that does that work for them. But the Stars’ coaches don't trust those numbers completely, and not just because Lindy Ruff is probably a single-finger typer.
That last line is the type of thing that takes away from the conversation, even in jest. I don't know Lindy Ruff, the man, but I am sure he wants to succeed at his job. As we've known since at least January, Ruff is more than willing to employ whatever tools will bring him this success, even if they don't fit the mold of what a 50+ year old coach "probably" prefers. We can disagree with the Stars' evaluation of their players--we will always disagree with them about something--but let's not start characterizing "the other side" when it comes to these topics. As we've all seen, there's a lot of heated debate about the idea that an automated analysis of hockey games might reveal things we humans didn't observe while watching them, but it's good to remember that these programs were written because hockey people wanted to know more about hockey players. What's not to like about that?
*Also* If we don't have at least 100 college students of our own signed up for Josh's summer tracking project by now, we're lagging. Come on, you people!
* * * * *
Here are your links, sponsored in no way by Chareth Cutestory:
If you'd like to get a look at Nill and Ruff as men outside the game itself, Josh Bogorad has a longer piece in which he is "not surprised" that Ruff and Nill are getting acclimated to advanced statistics. [Stars]
Dallas has re-signed the still-possibly-useful Scott Glennie to a one-year, two-way deal. Is he really only 23? [Stars]
St. Louis' best player of all time has to be Brett Hull, as St. Louis Game Time concludes. Trivia question: Whom has Hull consistently referred to as the best he ever played with? [SLGT]
Los Angeles avoided arbitration with their most eponymous player. [Royal Half]
Should Vancouver incorporate Johnny Canuck into their primary logo? [Pass It to Bulis]
New Jersey has extended Andy Greene for a 5-year/$25 million deal. That's a lotta years for his age. [In Lou We Trust]
But don't worry -- Lou Lamoriello is not worried about the Devils' shootout incompetence. In fact, he is explicitly clear that he is not trying to "address something." That should fix it. [Puck Daddy]
Former referee Paul Stewart gives five tips on how to command respect as an official, and one tip on how Steve Yzerman could have stopped being a tool. [Huffington Post]
If you want to see a poll that was at least more blatantly subject to the whims of its author than TSN's, here are the results from ESPN's Best Franchise Player voting. Why'd they have to show Modano in that jersey? Well, I guess it could have been Mooterworse. [ESPN]
Grantland (to which I am addicted) looked at the biggest trades from last season. Holy cow, Islanders. I did not remember your giving away sixteen draft picks for like four games of Thomas Vanek. [The Triangle]
Finally, check out this cool video from Fear the Fin, showing the progression of every NHL Franchise and logo since the beginning of the NHL. Any excuse to look at the Golden Seals logo is worth it. And with the Brass Bonanza, no less! [FTF]