In part 1 of this 2-part series on Wednesday, I wrote in some detail about Carolina Hurricanes Hockey Analyst Eric Tulsky, his history and role with the team and why I think the Hurricanes could be positioned to be a leader in this important area.
Today I will pretend that I am smart enough to figure out what Eric Tulsky might be working on this summer. Being just slightly smarter than I look, I knowingly do this behind the safety that the team will never, ever share what exactly Eric Tulsky is working on and make me look stupid. 🙂
Three projects for the summer of 2017 for Carolina Hurricanes Hockey Analyst Eric Tulsky
1) Assigning value to draft picks relative to prospects after various levels and degrees of success
Ron Francis’ mantra from day one taking over as general manager of the Carolina Hurricanes three summers ago was to build an organization capable of winning from within and doing so with consistently. For three seasons, he steadfastly worked on doing exactly that.
At the NHL trade deadline in 2015, Francis dealt Andrej Sekera, Tim Gleason and Jiri Tlusty and netted first, third and fourth round picks plus Roland McKeown who was recently selected in the second round. (Total: +4 picks plus prospects)
At the trade deadline in 2016, Francis dealt Eric Staal, John-Michael Liles and Kris Versteeg and netted two second round, a third and a fifth round pick plus Aleksi Saarela who was recently selected in the third round and Valentin Zykov who was recently selected in the second round. (Total: +6 picks plus prospects)
And at the trade deadline in 2017, Francis dealt Ron Hainsey and Viktor Stalberg and netted second and third round picks. (Total: +2 picks and prospects)
Francis did part with a few three of the extra picks in the deals for Teuvo Teravainen and Bryan Bickell and more recently Scott Darling, but he has still netted extra prospects.
Right now the game is transitioning from 100 percent “stockpile futures for later” to a more balanced approach that includes more “spend some (not all) futures to win now.” The Hurricanes system right now has a good mix of prospects and futures. Francis has three extra high (two second-rounders and a third-rounder) for the upcoming 2017 draft. He also has a few near-ready/AHL prospects in players like Haydn Fleury, Lucas Wallmark, to some degree Haydn Fleury and maybe even Aleksi Saarela. And he has a collection of promising players either still playing in juniors for 2017-18 or just moving up to the AHL. The list for this category includes Julien Gauthier, Jake Bean, Janne Kuokkanen and Nicolas Roy in the higher tier, and also a collection of rising dark horses in Spencer Smallman, Steven Lorentz, Warren Foegele and others.
I think there is a fascinating and complex math project to be done to assign relative values to these potential trade assets at different levels of development. Plenty of work has been done to value draft picks relatively speaking within a draft and even assigning a dollar value to them. But I think the next level that is more interesting is assigning value to a draft pick relative to a Canadian junior/NCAA freshman/sophomore level prospect (for different draft rounds)and also an AHL-level prospect who is closer to the NHL.
The questions are many. What percent of each round draft picks ultimately make the NHL level (easy)? Do the numbers change significantly if you measure for true difference-makers who play in the top half of the lineup as compared to depth players who can be more easily added via free agency or with any kind of AHL depth (harder)? Other than scouting are there statistical measures that can help identify which players exiting juniors and starting in the AHL have a higher probability to make the NHL? How much should value be adjusted for players as they progress to next levels and are closer to being NHL-ready? A draft pick past the first round is often 3-4 years away from the NHL and only a potential pitfall away from never actually making it; whereas a player like Haydn Fleury is something between zero and one year away with a pretty high probability? How do you correctly value proximity to the NHL relative to players or draft picks who are much farther away and therefore with much greater uncertainty.
I think the basic version of the project is calculating and considering simple probabilities of players at different stages simply making (and maybe playing 100 games or some other cut level) in the NHL could be useful. Trying to put some measure of probability to players based on successful seasons after being drafted could also help. Ideal would be creating a metric that assigns even rough values to prospects at different levels of development would make it easier to value trade assets requested in different deals.
Application: Even though it will be imperfect, I think kind of analysis could be of significant use to Ron Francis as he looks to pedal some combination of futures to build bolster his 2017-18 roster.
Which is more valuable? A third-round draft pick and a player in his second year of juniors and progressing well (i.e. Janne Kuokkanen) or a first-round draft pick? Which is more valuable? A near NHL-ready AHLer like Lucas Wallmark or three draft picks? My intuition tells me that the farther a player is from being NHL-ready, the more overvalued he is relative to a near-ready player. Even mid/late first-round draft picks often do not materialize into good NHL players (Zach Boychuk, Philippe Paradis, Ryan Murphy). I think there should be a high premium placed on players like Haydn Fleury and Lucas Wallmark who are near NHL-ready and with top half of the roster potential.
2) Help finding a winning formula for overtime hockey
Certain aspects of the game have greater potential benefit for statistical analysis versus coaching, watching film and other more traditional means of hockey analysis, but there is almost always a mix. In my opinion, improving on the Hurricanes weak overtime record in 2016-17 leans more toward traditional coaching, film work, etc. that falls more in Bill Peters’ role.
That said, I do think there are some statistical projects that could provide some insight too.
Defensemen vs. forwards by puck possession situation: What percent of goals allowed are with two defensemen versus one? Same question for goals scored? Does this change within 30 seconds of a face-off where possession is roughly a 50/50 split? Could it make sense to always start an overtime shift that begins with a face-off with two defensemen on the ice to hedge the fact that there is a 50/50 chance that the team will need to play defense? Do the numbers change if you are swapping players on the fly with or without the puck? Etc.
Tendencies of winning teams: Again, some of this is surely video work, but might there be something in the data that separates the winning teams in overtime from the losing teams. It could be the positional split mentioned above. It could be that they generally hold the puck longer or shorter before taking shots. It could be using combinations of players who regularly play together at even strength. Or it could be…you get the point. That is for Eric Tulsky to mine through the data and see if he can find anything statistically that seems to increase the probability of winning in overtime.
Application: Seek out strategies, combinations, players or anything else that seems to tilt overtime odds favorably.
3) Assessing options to add a defenseman
After the addition of Scott Darling (assuming he is ultimately signed) which I broke down on Monday, the biggest remaining need is a top 6 scoring forward. While there is inevitably a statistical angle to that just like any evaluation/transaction, because of the limited options and the known quantity of the players to be considered, I think that falls more in the court of the scouting staff, Ron Francis and Bill Peters.
But if I am right that Ron Francis could still be in the market for short-term help on defense, that could very much be a project for Eric Tulsky.
My analysis has Francis looking for a #4/#5 defenseman with a short-term (one year would be fine) contractual commitment, a reasonable price and the ability to either step into the top 4 or be a leader on a third pairing that likely includes a rookie. At a basic level, top 4 defensemen are a rare commodity, and the good ones are either not available or are expensive. That leaves Francis trying to sort through players maybe coming off of a tough season, nearing the end of their career or otherwise something less than a perfect option (hence the reasonable price).
Francis’ hunting project could benefit from sorting through statistical details to find players whose 2016-17 season was maybe better than simple statistics or the eye test would indicate. No one suggests just signing players based on statistical analysis, but the idea is to identify players who maybe are worth a second look in terms of scouting to see if perhaps they are better than first impression would indicate or maybe had circumstantial effects (bad puck luck, weak partner(s), etc.) that suggest they can play at a higher level in a new situation in 2017-18.
Application: I am not privy to what advanced stats the Hurricanes might use to evaluate defensemen other than the obvious ones, but running a screen to identify players whose statistical level of play in 2016-17 was better than the general impression they made could be a great starting point for trying to find higher-end depth defenseman who does not cost a fortune.
When I sketched out a rough outline for this article a couple weeks back, it included building as much of a statistical profile as possible for the summer’s goalie options, but apparently Francis and Tulsky got started on that early and voted Scott Darling. 🙂 There are still two scenarios where the goalie thing comes back into play. First would obviously be if Darling does not sign. The other is if Darling signs before the expansion draft and Ward is exposed and acquired by Las Vegas. In that scenario, I think it is possible that Francis would still buy out Lack, open up another slot and add another goalie this summer.
What say you Caniacs?
Do you have any other situations where statistical/analytical work could improve the team?
Do you think analytics could help with the Canes’ 2016-17 overtime woes? Or do you think this falls completely, or nearly completely, on Peters and the systems and strategies?