We all know that the biggest issue for the adoption of analytics in the day to day business of a team is education. Education on how insightful stats can be.
Education is not a bad word, and certainly does not mean coaches and managers who are not “educated” are late (trying to find a word without a mean connotation). Think about any skills you have acquired in your life, whether it is a software, management technic, gardening trick, etc. Before you knew it, you didn’t…
Education can come in two forms. Forced adoption when you have to remain relevant for a task, or deliberate adoption when you think there will be added value in learning something new.
Obviously, the use of stats and analytics need to be deliberate.
That is why, education of non-stat people in hockey (or sports in general) need to follow simple principles.
1/ It needs to be relevant for the user. A coach needs information on elements of the game he is using in his job. Sit down with him/her to go through the list of stats that can be provided. Only keep what’s gonna be used to save time for everyone. Also, if a coach is feeling pressured to use stats he/she doesn’t want is only gonna feed his/her reluctance to adopt stats…
2/ It needs to be user-friendly. Always get yourself to the level of understanding of the person you are talking to. Name stats in order to be understood (no acronyms beyond the obvious ones like xG), explain new stats in 1-2 sentences the first few times.
And make sure visuals are attractive and understandable in a few seconds. It implies findings the right balance between the amount of data you can place on a graph or a map while still being able to differentiate them (through colors, shape or size). Make sure to be consistent from one graph to the other (don’t flip shotmaps upside down all the time, don’t switch columns and rows in tables). Properly name your tables and graph (ex: “Transition” is not clear enough, but “5v5 - Transitions” gives enough context).
3/ Accompany the user down the road. The best way to educate new people is to help them getting the most out of the data. The first time you come up with a report template, sit down with the user to show him/her through it and make sure everything is clear. Then, if you are in an analyst position, it is your job to provide your insights alongside the visual materials on a daily basis. You are the professional who can look at multiple pages of data and connect dots together. Ideally, one game could be summarize in 5 bullet points, with a 10 pages report to support the findings.
Let’s use here the Game Report template we use at nlicedata.com, which needs to be both usable for the subscribers from the public, and works as a base for the teams we are consulting for.
For years, we used Tableau to build reporting visuals. Tableau is a quite easy tool to handle and helps create fantastic user-friendly interactive reports. However, the public version sadly does not have the ability to work with large database tools and fitting our huge amount of data was becoming problematic. Long story short, we replicated the reports using R and Shiny. We will present both versions below, the R version on the left, and Tableau on the right so you can see what can be done with either tool.
1/ Game Summary.
The Game Report opens up with a overview of the whole game. Goals, shots, Expected Goals (xG), scoring chances, rebounds, etc. in all situations and who deserved to win based on the xG created. Then a high-level split 5v5 versus power-play as well as transition numbers, the shot map with colors by situation and the Game flow showing the evolution of xG throughout 60 minutes.
2/ 5v5 Summary
The next page only looks at what happened at 5v5. Which is usually 45 to 50 minutes of a game in Switzerland. We show the same main metrics but also dive into how teams created/allowed shots and xG, through rushes, forecheck or in-zone plays. If scoring chances came from High-danger passes, Walk-ins, Rebounds or Royal-road plays, and a view of retrievals and rebounds won by each teams.
3/ 5v5 Transitions
Let’s stick to 5v5 for now, and look at transitions, zone exits, zone entries and high-danger passes. We show the number of attempts, by style (carries versus passes versus dumps), the success %, the number of transitions leading to a shot and the xG created.
4/ 5v5 Play Building
Diving deeper into transitions and what it led to, we have a page called “Play building”, aiming at connecting dots together. Here we differentiate exits and entries that happened through the left, center or right side of the ice, with the results it provided in terms of shots and xG.
Then we show how many xG both teams created/allowed from Full transitions (exit to entry to shot) versus from entries only (from a neutral zone retrieval), or from sequences without an entry at all (meaning turning a broken exit from the defense into offense inside the offensive zone). Here, R had the advantage to show both teams at the same time.
5/ Power-play and Penalty-kill
Now we move to special units. Lots can be analysed here but we summarized these situations by giving out the main stats, and the source of offense which tells how the power-plays created their shots. Zone entry numbers are also important whether you view this offensively or defensively.
6/ Goalies
Goalies have their page of course. With the number of shots and xG faced, goals allowed, goals saved (xG against - goals against). We have our special xG numbers only based on Shots on goal for goalies and the percentage of Controlled saves, compared to the expectation calculated for that game based on the quality of the shots the goalie faced.
7/ Players data
Players data does not need to be particularly visual. There is a lot to present and it could rapidly become too much for the reader. Simple tables are often better. That way, a coach can go through it, print it and annotate it any ways he/she likes.
8/ Player Cards
Something coaches really love is the ability to look at one player’s performance in details. They also can sit down with that player and go through the data together.
Our player cards shows the usual main stats, from time on ice to shots and points but also present the tiny details a coach wants to see. Such as transition success, offensive creation and battles or puck retrievals.
If you count other player pages on zone entries, high-danger passes, defensive metrics (zone denials, high-danger passes cut, rebounds recovered, etc.), you get that dozen pages report that covers everything that happen on the ice. It is way enough to be detailed enough while staying digestible and relevant for the user.
And consider supporting us with a coffee !
You can also follow me on Twitter pages here and NL ice Data here