An australian hockey analytics odyssey – the ice garden

I track live, which means I’m tracking as the game is happening, and despite the joy of the occasional replay I am unable to go back and triple-check who took that shot if I’m not sure. Thus, during the game, I rely predominantly on only myself for tracking information. If I don’t see a player make a shot they didn’t make it, but working in such a high-pressure situation has definitely made me lean more on my play-by-play counterpart and my growing knowledge of players’ shooting patterns/team systems.

For example, knowing a team’s play structures, lines, and power play units can make it significantly easier to work out who just had a beautiful shot on goal even when I can’t see the number clearly.

It’s a learning curve, and with each game, I get a little bit more accurate with a few less cross marks on the page from shots I never quite managed to identify. Example of an early 2018 AWIHL tracking sheet

After the game it’s just a matter of collating the paper, grabbing the boxscore from the scorekeepers to make sure I didn’t make any errors about the goal scorers, and entering the data so that the computer can do its thing. For shot locations, I use the ShotPlot tool that Andrew Pucci so kindly made earlier in the year after I mentioned on Top Shelf that I thought it was something that could be really useful to ‘grassroots’ analytics league.

Going team by team, period by period, I plug in the shot locations, flipping the paper for the second period so that each team’s shots stay at one end of the rink. While transcribing the shot locations I also have a spreadsheet open, entering the team, period, player and outcome of each and every shot. Once every shot on goal is registered in ShotPlot it’s just a matter of downloading the data and syncing it up with the Excel sheet I already have open. Transcription process using ShotPlot

In order to synchronise the X and Y coordinates between the teams and find the ‘true’ XY, we run the data through R (while also producing xG at the same time but honestly that’s less important). To do this we use the Google Sheets package, which means R grabs the data from the master file on my Google drive, pulls it into R and runs it through the program.

However, because R lives to cause me pain, instead of just adding some new columns to my spreadsheet with Google’s update function, R instead needs to export the data to a CSV file. It then uploads that data as a new file because, for some awful reason, the Google Sheets update function takes a minimum of two minutes a row and times out after an hour AND I have over 1000 separate game events and counting. The play by play summary is one of 5 generated csv files

Of course there’s some other stuff happening there as well; the main master sheet does get split into approximately for different spreadsheets while exporting from R so that I can draw both player specific and goalie specific data into Tableau (tutorial coming soon) more easily. The bones remain the same in V2.0 as it was in V1.0 with the goal of best visualising shot location, rates, and quality of shots on an individual, team, and goaltender level.

Sometimes, and not to sound like a hockey player while saying this, you really just have to pick up your team (or league) put them on your back, and just go for it. I’m not saying it’s going to be easy. If you read the above and still think it is, please go read it all again. People have been doing this very same thing for decades, pushing sport forward inch by inch.

When I jokingly pitched this story on Twitter, the title I gave it was “A guide to dragging your barebones stats league into the analytics era.” And yes, ‘analytics era’ is definitely a stretch when we’re not even at the NWHL’s level yet, and definitely nowhere near that of professional North American men’s leagues. I’m one person: I cannot also attach times to each shot, monitor faceoffs and somehow also track shifts as much as I’d like to, but in the space of just a couple of months, there’s more data available about a women’s hockey league in Australia than there was before.

That isn’t to slight other league, because the argument ‘there’s just not enough people’ unfortunately still rings true. That said, it takes one person to start, a serious dose of determination, and a willingness to literally drag a league on a journey whether they’re ready or not. I’m glad I did it, even if I’m still sinking more hours a week than I care to admit where my mother can admonish me for it into this project.

I’m not going to sit here and say that doing all this by yourself is easy, but I’m also not going to act like I’m doing something so mind blowingly individual that no one else could possibly do this themselves. All you need to track is a note pad and some badly-drawn rinks, or if you’re feeling fancy, a folder full of slightly modified NCAA tracking sheets. Regardless, if you’re looking at getting into tracking feel free to reach out; I will admit that most of what I know is crammed into what you just read, however if you’re looking at setting up tracking sheets for something specific, either paper or spreadsheet-wise, I’m more than happy to help in any way that I can just come find me on twitter.