AP Hockey Story of the Day: January 12 – On Shea Weber, Matching Skillsets and Possession Numbers

Sam Page has been doing a great series at SI.com on reconciling the views of the analytics community with conventional opinions on players based on the eye test. Most recently, he took a look at Shea Weber, a polarizing figure among analysts because of his mediocre possession numbers.

Read the piece, because Sam does a great job of breaking down film and coming up with potential explanations for these discrepancies, and the differences between the Weber/Suter and Weber/Josi pairings. His analysis, though, reminded me of a bigger issue surrounding the use of such stats that is important to remember: While these statistics are useful because they are shown to be repeatable and predictive, they are still situation-dependent. As an example, it is tempting to look at With or Without You (WOWY) statistics and use it to say point blank that one player is driving the play and the other reaps the benefits. But the fact is, certain players have skillsets that complement specific other players, and sometimes it’s just about finding the right combinations, the right systems, and the right situations for players.

So what is the important takeaway from all this (other than the fact that a top puck-moving defenseman would sure help the Preds)? Well as a general manager, it’s not just as simple as picking out the players with the best past possession numbers and signing them. In order to ensure that those numbers remain positive in your organization, you need to ensure that said player is a fit in the system, and that you have linemates who can complement their abilities. Every year we see players with supposedly good possession numbers fail to live up to the hype on new teams. Sure, there’s a luck component involved, but a lot of it is a failure to properly evaluate a player’s skillset. As with everything else in hockey, there’s nuance. It’s important to remember that.

AP Hockey Story of the Day: December 30 – Maximizing the Eye Test

The NBA’s Detroit Pistons are increasing their professional scouting abilities by deploying four full-time pro scouts to different parts of the country. They file thousands of reports on players, so that when a move needs to be made, or even just when scouting the opposition, the GM or Head Coach can simply pull up recent reports on players or teams. Some will say that too many scouts can be a bad thing in terms of communication, but I don’t think any team has reached the point where that would be a concern. What this is really doing is maximizing the eye test. There is a recognition that the metrics we have to measure things are far from perfect (even more so in hockey), but rather than combine them with occasional viewings or going off of reputation, you can employ scouts to watch virtually every game. It’s a system that makes a lot of sense and one that NHL teams with the means should immediately explore.

On the Legacies of Marc Trestman and Phil Emery

A large feature of analytics is a focus on process. It involves dissecting the thinking behind decisions more than the decisions themselves, since there is so much variance involved in outcomes. There are so few large decisions that ever come our way, and by evaluating process you can filter out that variance. One of the problems with fans assessing coaches and general managers is that we either judge based on outcomes – which is faulty because of the impact of that variance – or we attempt to get inside the heads of people whom we likely have never met, and whom all we have to go on is decisions themselves, media reports, and press conferences.

The Bears today fired Head Coach Marc Trestman after a very disappointing 5-11 season marred by locker room controversies after coming into the season with Super Bowl aspirations. General Manager Phil Emery was also let go, the man who had hired Trestman two seasons ago. The decisions were completely defensible. For a team with this much talent to perform so poorly was simply unacceptable, and there wasn’t anything to suggest a turnaround for next season in the last few pitiful weeks of the season.

Now I have to admit up front that my connection with Trestman goes back to his time in the CFL back when I would have called myself a CFL fan and a serious Alouettes fan. Trestman took a team that had gone through a number of failed head coaches, couldn’t produce on offense, and was floundering for success, and led it immediately to the Grey Cup, eventually winning two consecutively in his time in Montreal before falling back to earth. When Trestman was hired by the Bears, I was ecstatic because I had seen what he could do with an offense, because of how highly recommended he came from people like Jim Harbaugh, Rich Gannon, and Jay Cutler himself. Some people will now say that Lovie Smith should have kept his job, but after seeing him struggle with an admittedly poor Buccaneers roster this season I can safely say that Smith never would have brought the Bears back the Super Bowl. More on that shortly.

So Trestman came in last season with a mandate to revitalize an offense that hadn’t had success in, frankly, the entire history of the franchise, as well as to defeat the archrival Green Bay Packers. And I think it’s important to note that for 15 weeks, Trestman did just that. The 2013 Chicago Bears featured arguably the best offense in the franchise’s history, and scored the second most points in the NFL. Jay Cutler was very solid, Alshon Jeffery had a breakout season at wide receiver, Matt Forte was once again a monster, the offensive line held up like it hadn’t in past seasons (Trestman had also worked miracles with the offensive line in Montreal) and Josh McCown, unlikeliest of all, stepped in seamlessly on a warm fall day in Washington – a game I attended – and the weeks that followed to bring his team to within one win of a division title.

The flip-side to all this was of course the degradation of what had been a feared defensive unit, into an undisciplined and mentally weak group that teams could run all over, and that would forget to pick up the ball after a questionable fumble. It is important, though, to recognize that the defensive group that Trestman inherited wasn’t the same bunch that led the NFL in turnovers in 2012, or the even more fearsome group that took the team to the Super Bowl six years prior. This was a mashed up bunch of aged retreads like D.J. Williams, sorely lacking their departed leader Brian Urlacher, who were all a year older and over the course of the season suffered injuries like no other team. It is difficult in situations like this to separate the role of the head coach with that of the defensive coordinator – just look at what Rod Marinelli managed to accomplish in Dallas this year – but I would venture to say that even if a Lovie and Marinelli coached Bears defense wouldn’t have allowed the second-most points like Mel Tucker’s D did both last and this season, it would certainly have more closely resembled the Tampa Bay defense that this season allowed the eighth most points than any of those Bears defenses of old. And let’s face it, Lovie’s offense would have remained largely dysfunctional. The Bears went 8-8 last year, but with Lovie I would be surprised if they would have done even that well. He would have been gone, deservedly so, after last year anyway.

But back to Trestman. He came in and introduced an offense that dominated the NFL with a team that had never come close to that in the past. He defeated the Packers – albeit after injuring Aaron Rodgers – in his first attempt, and – maybe most impressively – took a group of players torn apart by the decision to part with Lovie, and brought them together as a family just as he had in Montreal. It’s funny how now, after seeing everything go so wrong, pundits are quick to point to Trestman’s unfailing loyalty to his players and coaches, his unwillingness to throw anybody under the bus, his flexibility and acts of kindness, his actions like changing up locker stalls to bring his full team closer together, all as reasons for his downfall, whereas only just over a year ago he was being applauded for those same characteristics and actions bringing together a locker room that may have crumbled under the leadership of anybody with slightly less charm.

Before addressing this season, I want to talk a bit about Phil Emery. There were the obvious positives, like selecting Alshon Jeffery in the second round of the draft, and the obvious missteps, like taking Shea McLellin in the first round. But every GM has some of both. I listened to just about every press conference Emery gave while in his position, and I don’t think I’ve ever been more impressed by an executive in any sport. Now as I mentioned above, it’s impossible from my position to get a complete read on Emery’s process, and it’s also impossible to know whether other general managers have similar ways of doing business, and Emery was simply better at communicating it to fans and the media, but I would venture to say that Emery’s process, at least at an ideological level, was top notch. Emery consulted a number of different analytical services, he scrutinized every decision he sought to make, he never overreacted, and he seemed to legitimately understand the shift in importance in the NFL towards offense and the passing game, and how best to maximize his team. Many will look at the decision to hire Trestman over Bruce Arians, the likely 2014 Coach of the Year in Arizona, and see only the result. But I liked the decision then and I still like it now. Arians would have been the conventional and the safe choice, and NFL offensive coordinator who had had some success as an interim head coach with the Colts. But Trestman had the potential to be great, he had shown as much in Canada and in his past as an NFL offensive coordinator and in college. Trestman brought ideas new to the league, and while he brought higher risk, if he could overcome his shortcomings, he could easily become a Super Bowl winning coach. It’s easy to look back and say that Arians would have been a better choice – although we don’t actually know that for sure – but how could Emery have known that? Trestman came highly recommended; I would imagine he interviewed spectacularly; and Emery saw the upside there that he didn’t with Arians. The process, or at least what we can ever know of it, was largely sound. In the end, Trestman couldn’t overcome his weaknesses and everything fell apart. But for a season we at least got a glimpse of what Trestman could accomplish. That, plus obviously a superior defense, was what Emery had envisioned, and it’s not unreasonable to think that in a parallel universe we could have seen that become a reality. I understand why the McCaskeys felt that a clean house was needed, but I would have kept Emery, or I’ll put it this way: I think Emery deserved to keep his job, but I think he also understands that by choosing Trestman, he took on a large amount of risk to go with the upside I discussed. Emery took a leap with the hope that a Lombardi Trophy would be the reward but with the understanding that losing his job could be the repercussion. I would hope that any GM would make that move, but also understand why the reasoning falls apart if GMs don’t actually lose their jobs when it doesn’t work out. I wish Emery all the best and I’m sure he will find success wherever he goes next.

But back to Trestman one more time, and to the 2014 season. It’s hard to defend him for his handling of the Aaron Kromer situation – although I’m still baffled at what that was really all about, since apparently coaches blab like that all the time – but again his loyalty to his coach was a characteristic of Trestman that worked well up until it didn’t. In other words, Trestman’s whole culture is built around trust. It’s what makes players want to take a bullet for him. But as soon as that trust is broken even once, the entire system falls apart. Trestman is also used to coaching within this system; it’s the skeleton of his success. But Trestman also hadn’t had to deal with true failure as a head coach before, and when the team stopped winning this season, it became harder and harder to maintain that loyalty and that trust.

And that is why I believe that the mystery behind this both infuriating and confounding season – and for me it’s even more the latter – lies in Trestman’s bread and butter: the offense. Why did a unit that posted the 8th most total yards in 2013 tumble down to 21st in 2014? Why couldn’t the receivers who always looked so open last year after two or three seconds never manage to do so this time around, and why couldn’t Cutler ever find them when they were? In my mind there are two possible explanations. Either Jay Cutler really did ruin this offense, whether through a poor work ethic or an inability to hit throws, or a regression from being above average last season, or the rest of the NFL adapted to Trestman’s schemes, and he couldn’t make the necessary changes to get back on top. While the week 16 game in which Jimmy Clausen ran the offense better than Cutler in any week this season suggests that it is the former, the fact that Trestman didn’t make this change earlier suggests that maybe it was the latter. Was Trestman simply overly loyal to Cutler, and it doomed his offense as a result? Or was Cutler simply helpless in an outdated scheme? It’s a question to which nobody has been able to provide a decent answer, and to me it is the answer to this question that should truly shape Trestman’s legacy as a Chicago Bear. Was his demise the result of an inability to adapt offensively, or a result of his misplaced loyalty in a quarterback he wanted so badly to see succeed, but who simply wouldn’t? Unless somebody undertakes a detailed project to examine film from both last and this year, we may never know the answer (although the futures of both Cutler and Trestman will likely provide some clues), and that’s too bad. Was Trestman a great tactical offensive coach done in by his own system of trust? Or was he overrated tactically, and did that torpedo his interaction with his players?

Whichever version of the truth you opt to believe, there are a few points about Trestman’s legacy that everybody should remember before lumping him in with some of the worst coaches this league has ever seen.

1. Trestman was able to succeed with this offense for a season like nobody has before.

2. It was Emery and not Trestman who hired Mel Tucker to run the defense, and there’s no question it was an unmitigated disaster. What would this team have looked like with better defensive (and special teams) coordinators?

3. Trestman is by all accounts one of the nicest guys in football. There’s no reason to dance on the grave of a guy who gave it his all every week, did some great things, was never stubborn or dismissive with the media or with fans, had a solid process, and in the simply failed under tough situations.

To take this story full circle, there’s a lesson in analytics to be learned here. If it wasn’t obvious before, buying into the practice doesn’t mean you will be successful. It gives you a better chance to be, but it doesn’t mean that a certain GM or a certain coach will balance different numbers, scouting, gut, and other factors correctly in coming to decisions. This is where we are limited in our ability to evaluate. I don’t know how good of a process either Emery or Trestman truly had; we simply can’t with current public knowledge. But it’s clear that both were progressive minds, and maybe Chicago – a town of people with little patience when it comes to failure and accustomed to a workman’s team led by its defense – wasn’t the right setting for the duo to learn together. Managers like to say that they can’t bow to public pressure, but there is a snowballing effect that comes from public displeasure that can lead to a dissolving of locker room culture. It’s something that the Houston Astros are currently figuring out in baseball, and that other organizations will have to reconcile when it comes to analytical adoption sooner rather than later. There is so much nuance involved in sport, and the line between success and failure is so slim. At the end of the day of course, though, success is still success and failure is still failure. And it’s very difficult to reverse course with the same group of people once it goes down the latter path.

So we should all wish the best to Phil Emery and Marc Trestman, hardworking, intelligent, and kind football people, who will undoubtedly have future success in some capacity in football. Just because firings are defensible, doesn’t mean that anybody should dance on the grave of those who lost their jobs. And doesn’t mean that the times at which those people succeeded, or the process used to achieve those successes should be forgotten.

So good luck to you Phil and Marc, thank you for building the greatest Bears offense I have ever seen, and best of luck.

The Top 10 Bruce Springsteen Songs

When I started this blog, I made sure to apologize for the fact that there would be some non-hockey/non-analytics/non-sports content, and I’ve realized lately that I haven’t followed through on…you know…what I apologized for. It’s a well-known fact that many sportswriters like Bruce Springsteen. I don’t know why this is other than to say he’s amazing, so I’m not sure why everybody doesn’t. I’ve had the opportunity to attend one full Springsteen concert (3h45 mins in length) and he’s simply the best there is live. But even beyond that, his imagery and lyrical abilities are both excellent, and having read his biography I can tell you that there are a lot of fascinating stories behind the music he’s written that make me appreciate him even more as an artist. With that, for those interested, here are my picks for the Top 10 Springsteen songs. These are objectively correct and beyond reproach, but feel free to comment/tweet your own faves so that I can tell you whether or not they are acceptable choices.

10. Hungry Heart

Hungry Heart was Springsteen’s first real “hit” but it almost wasn’t even his. Bruce had initially agreed to donate the song to the Ramones, because he didn’t like how “poppy” it sounded. His manager and bandmates desperately pleaded with him to keep it, and finally he surrendered. The song put Bruce on the map, and the first time he played it in public, the crowd already knew all the words so well that Bruce couldn’t even get out the first few words without being drowned out by the crowd. Since then, he’s always let the crowd sing the first verse on their own.

9. Thunder Road

Thunder Road has always been a song I’ve had a hard time appreciating because it doesn’t have the same arena appeal that classifies some of his other huge hits. But the imagery is so poignant, and it’s so unlike any other song that’s ever been written, that it just has to be on this list. Thunder Road itself is such a great metaphor, and it really extends through all of his work.

8. Badlands

This song is pure energy; I’m not sure there’s a song in the history of the world that tops it in that category. The opening riff is genius; you could hear it faintly anywhere and know immediately what it is. Just soak it up.

7. The Rising

The Rising became an anthem for many Americans following the 9/11 attacks, and the underlying melody and lyrics are well thought-out and relevant but still subtle and symbolic.

6. The Promise

Some of Bruce’s best work comes on the piano. His upbeat stuff gets more praise, but his ballads are in some cases even more meaningful and well-written. Bruce was first discovered by a young record label employee named Mike Appel who fought for him and quickly became his manager. The two in many ways grew up together, finding their way in the industry, but following the success of Born to Run, Bruce found out that Appel had betrayed him, giving Springsteen an unfavorable contract and taking much of his money and music rights. Following a long drawn-out court case, Bruce split with Appel, and wrote this thinly-veiled song about a promise to him that was broken, one which obviously hits him hard.

5. Backstreets

Possibly the most epic piano opening to a song ever written. ‘Nough said.

4. The Promised Land

This is a song about searching for freedom; it’s the kind of tune you’d yell out driving across the “rattlesnake speedway in the Utah desert”. It’s got the perfect mix of catchiness and emotion.

3. The River

The story at the beginning of the video below is a perfect snapshot into Springsteen’s childhood: his fights with his parents, his lust for freedom and rock & roll, and difficulties of the time. The River was a song Bruce wrote about his sister; it very literally tells the story of her life. Fun fact, though, is that his sister had no idea he had written it when he first performed it live at a concert she attended. She was stunned at first, but over time it became her favorite (American spelling is more appropriate for this post) song.

2. Racing in the Street

Another slow ballad, but filled with brilliant piano and guitar solos and evocative lyrics. The song is about the clash between the freedom-searching youth and the domesticated adult. Bruce struggles between childhood and adulthood, and his woman takes the biggest hit, as she worries about him every night.

1. Born to Run

Not much to say here, except that this almost never was. Following luke-warm sales of his first two albums, Springsteen’s record label was set to pull the plug on his funding. Fortunately, his manager worked out a deal where the company would provide the funding to record one single. If the company liked it, then it would fund the rest of the album. If not, he was done. Bruce’s final chance at success became one of the most renowned songs of all time.

On Changing Environments and Sample Size

Micah Blake McCurdy (a good follow on Twitter here) has been posting division projections for most of the season, and today he explained how his model works in more detail. While his work doesn’t account for injuries, goaltending, special teams, or shot quality, I still think it’s one of the better projection models I’ve seen and is certainly useful. After all, he’s proven it works pretty well using past seasons. The most interesting passage from the article (read here) though for me was this:

Screen Shot 2014-12-23 at 12.22.04 PM

You may not understand all of that, and that’s fine, so let me summarize what was interesting to me about it. The predictive power of this model increases when you base your projections on the statistics from more past games, but only to a certain point. Once you get past about 25 games (the number Micah decides to use), the predictive power begins to decrease. Now hockey analysts are always trying to get the largest sample size possible, but this is a good demonstration of one of the biggest conflicts in analytics for any sport (I’ll have a post on these at some point), one for which we as of yet don’t really have a solution. That conflict is this:

One needs a big sample size to make reliable conclusions regarding data, but bigger sample sizes leave one increasingly susceptible to changing environments which can cloud the data in its own right.

For example, we know that shot quality is prone to lots of randomness. How do we best determine a player’s true shooting talent. Let’s pretend it’s the summer of 2012 and we’re trying to determine just that for Max Pacioretty. Take a look at what we have to work with.

Screen Shot 2014-12-23 at 12.35.37 PM

One couldn’t reliably say after one full season of significantly above average shooting percentage (’11-’12) that Pacioretty has significantly above-average shooting talent, so we want to include as much data as possible to filter out some of the noise. If we take his career shooting percentage at this point in time, it would come out to a nice 10.0%. Now take a look at Pacioretty’s career shooting numbers as of today.

Screen Shot 2014-12-23 at 12.38.47 PM

In his following full season, Pacioretty shot 14.4%. His shooting percentage since that summer is 12.2%, which is pretty similar to the 11.8% that he put up just in the 2010-11 and 2011-12 seasons combined. His data from the first 86 games he played in the NHL in which he could barely score is virtually useless in predicting his true shooting talent. Now this is definitely an extreme example, but it shows how some players can have difficulty adapting to the NHL, or are subject to other factors that can have real impacts on results. Believe it or not, there is such thing as a player coming back from injury and being afraid of crashing the net like they once did. That could decrease a player’s shooting percentage.

So what does this all tell us? Well it tells us to be careful with data, something I preach in just about every one of my posts here. There’s nuance everywhere, far more in hockey than in stop-start sports like baseball (although this particular factor could exist there as well). It is important to recognize hockey as probabilistic, but also as a product of choice. A power forward could make the decision to never enter the crease again, and that would have a significant impact on shooting percentage that wouldn’t be subject to the regression a model would predict. It’s not something that would happen much, but it could once or twice, who knows what goes on inside a player’s head? Ultimately, we must remember that a high Corsi percentage doesn’t tell us a player is very good, it tells us that by and large a player with that Corsi is very good. This is a distinction that is critical, and one which separates in my mind smart hockey analysts from those who go too far in the other direction from purists.

AP Hockey Story of the Day: December 16 – Might four-man defensive units work?

The NBA’s Sacramento Kings, who recently fired their head coach despite a not-horrific start by their standards, are rumored to be looking to try a new system that, were it successful, could completely change the sport’s dynamic. You can read about it here, but essentially the Kings are looking to play a four-man zone-style defense, which would involve one player consistently focusing on offense – a cherry-picker, if you will. Now as the article points out, the reality of the situation may mean a 5-man defensive unit but one man designated as a breakout player who immediately sprints up court whenever the opposition takes a shot. It’s a very interesting idea and it should be noted that this isn’t the first time the Kings have done something innovative.

So how does this apply to hockey? And what are my thoughts on this? Well my first impression would be that a set-up like this would involve a four-on-four at the defensive end (since surely the opposition would adjust to stick a player on the cherry-picker) and a five-on-five at the offensive end. At least in hockey (and I would imagine in basketball as well), offense is easier to generate with more room at four-on-four, so it would seem like all the team is doing is making is easier for the opposition to score while the only advantage would be potentially getting some one-on-ones up-court on the breakout. We can’t know for sure without seeing it though, in either sport.

The big question this presents though is, why don’t NHL teams use farm clubs more to test set-ups like this and see what works? Whether it’s a four-man defense, or playing with 3 D, 2F, or playing with “midfielders”, or whatever else, we can’t know what might change the game until we try it. Somebody’s gotta be the first mover. That’s what Sacramento is trying be, as they look for any possible advantage after years of poor performance. After all, they can’t be much worse.

AP Hockey Story of the Day: December 9 – Birnbaum, Tango, and Shot Quality Revisited Again Again

Phil Birnbaum, who is one of the great non-hockey analytics writers out there, has taken a number of stabs at the shot quality question in hockey over the last couple of years, and today weighed in on the Tom Tango controversy.

The issue – whether to weight goals significantly higher than other shots in corsi analysis – is one I’ve stayed relatively quiet on, and that’s simply because like any good jury, I want to see all the evidence presented before coming to a decision. We’re still not at the point where I’m totally confident evaluating the worth of Tango’s statistic, or the merits of shot quality overall, but I think that’s partially because the answer depends on the question we’re trying to answer. Does shot quality matter? Absolutely. Does it render large sample shot differential metrics useless? Nope. Can it be used to improve on what we have? I think it can. As Birnbaum has often pointed out, we know that shot quality impacts shooting percentages because we see it in score effects. Is it possible a team could play a system in which they more resembled a team down a goal than a team in a tied state, thus impacting shot differentials and shooting percentages? It’s possible, although it’s important to note that there are psychological factors involved in score effects, as well as the other team playing a certain way. It’s not just one team that impacts it. It also seems quite plausible that teams make the conscious choice to forego shot attempts in order to try for better shots. I think the Ducks are a team that have done this the past couple of years, and the Leafs may be as well. Those changes aren’t enough to impact the idea league-wide that more shots = more goals, but on a team level it could. This is where the sniff test comes into play. We may not have the statistics to prove such decision-making exists, but that just means we have to try harder to find them.

So where does this leave us? Ultimately, dealing with goal data is still very difficult because of the variance involved in goalie play and shooting combining into something that tells us very little reliably. But it’s time that the discussion shifts from the dismissive “we’ve done a regression and we’ve shown that shot quality doesn’t impact the numbers much so it’s not worth pursuing” to “what can we do to limit the variance involved and get meaningful data out of teams’ ability to convert shots into goals”. I don’t think the Tango statistic adds all that much – although I’m waiting to see a version that accounts for score effects – but I also think that’s because it’s so basic and doesn’t really do anything to account for the variance involved in goal scoring.

What’s next? Without tracking data, I’m not really sure. But I definitely wouldn’t want to be somebody staking their reputation or career prospects on how good a team is or isn’t based on corsi in cases where there’s the possibility that system effects or even a changing environment based on shots and carry-ins becoming more of a policy target (see Goodhart’s law) are skewing those numbers and failing to give a truly accurate representation of a team’s even-strength play.

Birnbaum is somebody I will be following closely through all this. As I will everybody else who is refusing to accept “shot quality doesn’t make much difference” as proven fact. After all, absence of evidence does not equal evidence of absence.

AP Hockey Story of the Day: December 1 – “We can’t get much worse”

DC’s own Tony Kornheiser, of Pardon The Interruption fame, came out yesterday and said that he feels the best next step for the dysfunctional Washington Redskins is to embrace analytics, to try and do to football what Billy Beane did for baseball. His reasoning is one which I think should be used to convince front office personnel more often, even if it’s a risky approach: “How much worse could it get?”

If you’re a team, and this applies to any sport, that hasn’t made the playoffs in a number of years, that doesn’t exactly look poised to take the world by storm, isn’t it worth doing something unconventional to turn the tide (non-McDavid year category)?

If you’re the Florida Panthers, for example, or the Carolina Hurricanes, or the Edmonton Oilers. You haven’t made the playoffs for a while (at least in a full season), it’s not like a radical change of approach is going to hurt your brand at this point. What is holding you back from going all in (and I mean all in) on analytics?

I’ve always said that analytics is an organizational attitude more than it is a single system or a single metric. Going all in on analytics doesn’t mean firing your scouting department or building an alter to the corsi gods. It means scrutinizing every decision you make, seeking information through data wherever that can be achieved, and trying to gain an edge in a very competitive environment with so much parity.

So yes, the Redskins should go all in on analytics. So should every other team, but for a team that’s been woeful for so long, there’s really no risk or downside. Worst case, you’re still bad but you were probably going to be bad anyway. Best case, you win championships, and maybe even change the game.

AP Hockey Story of the Day: November 18 – On the Defensive Shell

Garik16 from Hockey Graphs, Lighthouse Hockey, and Islander Analytics wrote a good piece today on the defensive shell (a topic that’s been on my list to address for a while), following up on David Johnson’s initial look into the subject a couple of years ago. I highly encourage you to read both stories, but the general conclusion from today was that the shell doesn’t actually help a team because the opponent’s scoring rate – what you’re trying to minimize – actually increases. I had a couple of thoughts on the issue, because while the material is interesting, I don’t necessarily agree with the conclusions.

First of all, it’s important to note that score effects are the result of a combination of four very different forces (more on this in a future post).

1. Players naturally playing harder/more aggressively when trailing

2. Coaches coaching trailing players to push ahead and take risks

3. Players naturally being risk averse and not going their hardest when leading

4. Coaches coaching leading players to make the safe play (ie a contain or prevent defense)

Generally speaking, items 2 and 4 are those that the coach can impact. A coach can tell his players to keep pushing up by a goal late in the third period (they generally claim that they do, and players often echo this notion) but items 1 and 3 will still naturally institute score effects, and therefore the idea that getting hemmed in one’s zone leading late is simply a poor strategy that should be discarded – like the 1-3-1 forecheck or the overload power play – is misguided.

Of course, there IS an element that is coaching-driven, but the problem you run into here is that the impact of such strategizing will tend to get swallowed in heaps of variance, since you’re adjusting one of the four factors, but the other three stay the same (the other team won’t, for example, suddenly agree to not take chances since you’re still pushing ahead).

So when I see something like this graph, where the author is using shooting percentage Down 1 as a proxy for results against a defensive shell, I worry that the intended effect isn’t being isolated.

Screen Shot 2014-11-18 at 4.39.21 PM

We know that the shooting percentage on average Down 1 is greater than Tied, but if one removed items 1, 2 and 3, leaving only the strategized intentional defensive shell, would that still be the case? Maybe, but we can’t say that from just this data.

AC Thomas’ more precise graphs on shooting percentage as the game goes on don’t rid this analysis of confounding variables, either.

Screen Shot 2014-11-18 at 4.43.22 PM

Sure, the shooting percentage when trailing is shown to be higher than that with the score tied right up until the end of regulation. But there’s a pretty logical reason for that as well. Teams tied near the end of regulation are playing for overtime. They are far more likely to dump shots from the blue line than to pinch to create opportunities. And both teams are far more worried about not allowing a goal against than scoring a goal for. There is massive loss aversion in play here; shooting percentage is bound to drop. You can’t compare the two situations because looming overtime (or more specifically the loser point) is a confounding variable.

So yes, scoring rates are higher against teams protecting leads than against teams late in tie games, but that doesn’t mean that the defensive shell, as a strategical maneuver, is responsible for that. Impending overtime, as well as natural factors that are difficult for coaches to account for, could just as easily be responsible.

On Russell Martin and the Differences Between Evaluating Baseball and Hockey Contracts

Today, the Toronto Blue Jays signed Canadian Catcher Russell Martin to a huge 5 year, $82 million contract. I’m no expert on baseball analytics, but I know enough to be able to find concepts to apply to hockey where possible. I have, however, seen many people making fun of this as a massive overpayment. They’ve called it “McCann Money” with the implication being that Martin is by no means the player that Brian McCann is. Now I don’t know how big the gap between the two players is, and frankly I don’t have the time nor the will to find out, but there is one factor fans – especially those who primarily follow hockey and thus appear on my feed – may not be taking into account that I’d like to address.

Hockey has its similarities and differences to baseball, and one of the most important differences is the salary cap. Hockey’s hard cap is an important differentiating factor because it levels the playing field amongst teams, for one, but it also changes contract evaluations because the deal has to be evaluated (at least for cap teams) based on whether it brings the best possible return for the team in a $70 million world.

Baseball is different not just because the teams spend different amounts of money, but because the amounts teams spend are variable, and can be influenced by the very return they procure from that money. A successful team can lead to playoff revenue which means the team can afford a higher payroll. A successful and marketable player can lead to additional ticket and concession sales, which again leads to more available cash.

With Martin this is particularly important because he’s a Canadian boy. More than that, he’s a Canadian star. Russell Martin may not be Brian McCann, but by catcher standards, he’s a star. And it’s not like the Jays are crawling with stars. Martin will undoubtedly very rapidly take over the team lead in jersey sales, will be a poster boy, will attract Canadians to the ball park, will do advertisements. That all brings revenue to the team, which helps offset the cost of his contract in a way that, say, the Montreal Canadiens signing Daniel Briere never could. The Canadiens didn’t get to spend to a higher cap because a Briere brought in additional revenue, but the Jays can.

So this is a caution – without too much in-depth knowledge of the situation – to Toronto fans who may be used to the hockey way of thinking. There’s more to contracts in baseball, and ultimately, whether or not Martin is a $15 million/year player, this contract will probably prove fruitful for the franchise.