A couple thoughts on analytics, and how best to integrate it into hockey operations.

With the Oilers cutting ties with analytics consultant Tyler Dellow, the club has an opportunity to re-set its current approach to analytics going forward. Now it was my hope that Tyler’s role would’ve expanded from a consultant position, perhaps into a director type, who would then lead a group of analysts to delve into various topics. It was my hope that the Oilers would’ve applied a more complete analytics strategy that could support and influence all aspects of hockey operations including on-ice tactics, player personnel decisions, drafting and salary cap matters.

And let’s not kid ourselves: analytics is going to have a role in professional sports, but its exact definition and scope is going to vary depending on the sport, league, team and goals. Analytics is becoming more and more ingrained in all industries whether it is healthcare, oil and gas, government, and the thought of the value of analytics fading in hockey is completely bogus.

Analytics should never be treated as some magic bullet or switch that an operations manager can just flip and make things happen. Analytics is a continuous process, one in which a business need or goal leads to questions. From there, an organization looks to its data to see what can be answered, and what other data can be collected and refined to potentially answer that question. From there, analysis of the data is done, which leads to discussion with the operations side and, quite often, leads to more data collection/analysis/discussion.

Before all of this is even thought about, teams like the Oilers have to treat analytics as a new concept that has to be assessed carefully before it’s legitimately integrated. And like most organizations, the Oilers need to look at three areas: people, process, technology (or tools).

People:

• Do you have the right skills in your organization that can support a team of people whose focus it will be to analyze data?
• Do you have the right type of managers that will keep analytics top of mind when reviewing the operations they oversee?
• Are people willing to learn the skills needed to either work with or within an analytics team?

Processes:

• How will the organization facilitate the work of the analytics team?
• Regular presentations to management? Crossover meeting between the analytics team and, say, the coaching staff?
• Do these collaborations need to be formalized or do you let the analytics team set up ad hoc meetings and working groups?

Technology/tools:

• Do you have the necessary tools for your analytics team, or is there a chance you’ll have to invest in some additional applications?
• It would be safe to assume that your analysts would dictate what tools are used, and the organization has to be prepared to support them.

One thing that I hope the Oilers consider doing is finding a way to tap into the knowledge of the fan community, especially those that spend hours analyzing data and publishing their work online. And this really isn’t a long shot for the team. Keep in mind, the Oilers have put together a volunteer advisory group in the past to support hockey operations and they’ve also hosted a Hackathon competition where they posted a question, released a pile of data and rewarded the best solution (nice work Parkatti!). What I think the Oilers can do here is enhance these two concepts and turn them into actual, formalized programs that can be sustained and provide value to the club.

One thing I have trouble wrapping my head around is when NHL teams hire a consultant or two to support their analytics process. Above is a chart from Gartner, which does a really nice job breaking out analytics into four types: Descriptive (what happened?), Diagnostic (why did it happen?), Predictive (what will happen?) and Prescriptive (how can we make it happen?). What we see without even caring about value and difficulty of each type are complex tasks that each require more than just a data analyst or consultant.

Looking at this through a hockey lens, the first type would be simple reporting, as in how many goals happened for and against. With Diagnostic, you would start looking into shot shares/location/player deployment/line match-ups, etc, basically looking at the things that you think do a good job at predicting goals. These first two types of analytics have become pretty standard things in the hockey world, and are published daily by fans online. But if you’re running a team, your hockey operations department could be looking at more than just goals for instance and the things that lead to goals. Maybe you want AHL player data, or a better way to track the passes that lead to shots. And if that’s the case, you need a way to gather and refine that data, which could require manual tracking and someone with programming experience. If your club wants the findings shared in a certain way, you may need someone who specializes in reporting or even dashboard reporting. And it’s also here that you may need someone who can break down video and compile their findings quickly for the coaches or management to use.

And when you start getting into the prescriptive analytics, you’ll absolutely need someone on the analytics team that has coaching experience or someone that can marry the data to the actual on-ice plays to make sense of it and provide recommendations. Reviewing these types of analytics and the potential value it can bring to a team, it would be imperative that a team like the Oilers put together a complete analytics team. This should include a director type, along with analysts well versed in reporting (dashboards), programming and on-ice coaching. Teams like Toronto, Florida and New Jersey have this structure, which should become the norm among NHL clubs soon.

It’ll be interesting to see what the Oilers do in regards to their approach to analytics. In my experience, analytics is one of the many tools that business leaders rely on to make informed decisions and is part of a holistic approach to finding success. The purpose of analytics, especially in a business setting, is to provide evidence, drive discussion and support the corporate goals. And it can only be leveraged to its full capacity when there’s complete support, at the strategic level and the operations level.

Curious to hear the thoughts of others on this one. Every industry is different, so I’d be interested in hearing how others have implemented/experienced analytics.

Also joined Lowetide on TSN 1260 to talk about this further. Audio is here.