When the Devils fired their coach, John Hynes, there was a bit of glimmer of hope that the Devils could return to some level of competitiveness. Alex talked on this site about the “energy” that a coaching change could instill in the team. After two thoroughly embarassing losses at the hands of Our Hated Rivals and the Sabres, the Devils looked much more like a hockey team against a good Vegas club. Then we looked okay against in a shootout loss to the Blackhawks. Then our offense showed up for a change in another close loss against the Predators. But, energy or not energy, that’s still 3 losses. And through in-depth research and careful scrutiny of the data over the 27 years of my life, I’ve come to the following groundbreaking conclusion: people don’t like to lose. As banal as this epiphany may seem, it has a tangible impact worth investigating.
And now: a quick commentary on why I’m talking about this. In recent weeks/months, I’ve written more about public opinion than typical since it has infringed upon the territory of actual empirical analysis more than usual. What I mean to say is that fan reaction, while interesting, rarely impacts the on-ice product. If you disagree with this, feel free to have at it in the comments as I think that would be an interesting subtopic, but I’m going to operate under the assumption that this is true. This Devils season is atypical in the sense that it is factually true that the players noticed the fans reactions, and arguably true it had an impact on the on-ice product. Therefore, attempting to understand the fan reaction is uniquely relevant to the story of this season.
This puts me, as an analytically-inclined writer, at a bit of a crossroad. On the one hand, articles like what I wrote last week venture too much into personal opinion/narrative to be irrefutably true. On the other hand, the highly evidentiary pieces that have firm ground in analytics ignore the obviously important component of fan displeasure/reaction. I started looking into if there was a way to quantify fan reactions and stumbled upon the work of Ryan Ghizzoni (@mackinawstats) in quantifying positivity of Maple Leafs fans and personnel. I asked him for his code and he was generous enough to provide it. I rejiggered it for the purposes of calculating the Devils twitter sentiment. All of the following was done using the excellent rtweet package to scrape twitter data. To determine how positive or negative the tweets were, I used the Bing Liu’s sentiment lexicon from the tidytext library.
Devils Twitter Sentiment
Unfortunately, Twitter only stores 10-day lags of general tweets on the API so I’m not able to go all the way back to the beginning of the season like Ryan did for all of Twitter, but I am able to look at the fan reaction to and rebound from the Hynes firing. As you may expect, “Devil” is a term that has a heavily negative connotation in non-hockey circumstances, so, unlike Ryan, I had to specify only references to the Devils twitter hashtag.
As you can see, shortly after Hynes being fired, there was a little bit of an uptick in overall good feelings surrounding the team with regards to their tweets. It has, unsurprisingly, immediately returned to it’s pre-Nasreddine level.
While you’re only able to go a few days back to track twitter’s entire public database, you can search the contents of specific accounts dating back much further, depending on how frequently they tweet. To attempt to proxy some the fan reaction to some of the Devils news, I thought it’d be best if I used twitter accounts dedicated exclusively to Devils news. I figured that our own twitter — @AATJerseyBlog — would be a prime candidate for such an experiment. To ensure that there isn’t some bias about how we, in particular, covered the Devils, I included two other Devils blogs in the study. Whereas the first time, I grouped everything by day because there were plenty of daily tweets to aggregate, this time I’m only using 3 accounts so I averaged the positivity percent of each week (thin lines) and then loess-smoothed the results (thick line).
The Hynes firing seemed to somewhat stem the tide of creeping pessimism that had plagued the Devils all year. These blogs had some unique characteristics in how they had handled the news as well. You can see that DevilsArmyBlog was getting pretty monotonically getting more and more depressed (totally understandable) until the Hynes firing bandaged the wound. Our own lovely blog had somewhat plateaued and was more recently thrilled with ... something.
The Big Point
Why look into this at all? Hockey is a business. And you have to think that there’s a limit to how much people can hate-watch games and hate-buy tickets. In other words, it’d be nice to keep your fans happy. If your customers are consistently unhappy, your business becomes less viable. And when the business isn’t viable, people get fired. Hynes’s firing may have offered some temporary fleeting relief, but it increasingly feels to many like this is a level of ineptitude that could swallow up Shero as well.
I’ll allow you all to discuss this a little more in the comments, but I’ll preface this by saying that it’s difficult to fault Shero for many of his paper moves. He’s won a lot of trades, avoided bad contracts, and put a team together that many of us were excited about heading into the season. That said — it’s fair to be concerned about the complete impotence of the Binghamton team, the repeated inability to find a viable top 1 defender after multiple attempts, and his refusal to address a glaringly obvious problem at the goaltender position for years.
At the time of this writing, the Devils are losing 2-0 against the Stars in the 2nd period, getting outshot 21-4, and not looking particularly likely to stop sucking anytime soon. So, that graph you saw at the beginning of this piece is likely to keep getting worse, and the situation for Shero is going to get more and more precarious. If you’d like to see some specific complaints surrounding the organization as a whole, stay tuned for Jeff’s incoming post, and check out the responses to his tweet in the meantime.
Now, I can’t imagine that someone with Shero’s acquisition record will get canned before getting a second shot at addressing the head coaching position. But if he swings and misses, it seems possible he won’t get even one more season of ineptitude under a new coach without his seat’s temperature needing 3 digits. Something in the organization — NHL, AHL, development, etc. — needs to show promise, or the motivation to keep the GM that’s presided over seemingly-perpetual failure will dwindle. And I’ll probably map every minute of it.
I’d like to hear what you guys think about this type of analysis. How dumb is looking at sentiments of tweets? If you don’t think it’s dumb, do you have suggestions on other types of things to look into or different ways to research this topic? Is there some other way you think it should be done?
And what about the consequences? Is Shero already on the hot seat for you? Does he get one coaching change? Two? None? Share your thoughts in the comments below.
Thanks against to Ryan for his code on this piece — give him a follow on twitter.