Visualizing Team Strength in the Pro Peloton

[MattTuck]

I did a bunch of visualizations last year around the spring classics. I wanted to do this one for a while and have been slowly teaching myself an SVG program to create it. Finally have something that I'm not embarrassed to share.

Hope you guys find it interesting.

It basically looks at how transfers from 2016 to 2017 impacted the strength (as measured by top 10 placings in WT events in 2015 and 2016) of various World Tour teams.

Unfortunately, when I try to upload the file to the paceline server, it compresses it, and is no longer legible. I tried hosting it through my google photos account, but that also degrades the resolution. The best solution I've found for hosting right now is that I've created a twitter account and have uploaded the image there in a tweet.

Link to Image (https://pbs.twimg.com/media/C2i8xWQUAAIMpF7.png:large)
And if that doesn't work, here's the Link to the Tweet (https://twitter.com/visualvelodrome/status/822114509947019265). You'll have to click on the image, then open in new tab. Should be full resolution at that point.

[weisan]

Very cool, Matt!

[Jad]

That is a great way to see movement and consequences; I hadn't know the current state of affairs, so this helps set context for the Classics and the season. It's cool to see the number of symbols some of the riders are bringing with them. Also, it's clean and really nice to look at--thanks for sharing this.

Are the teams arranged in an order vertically?

[sandyrs]

This is really cool. Is the underlying data (rider teams and wins) available in a consumable format somewhere or did you gather it manually?

[MattTuck]

That is a great way to see movement and consequences; I hadn't know the current state of affairs, so this helps set context for the Classics and the season. It's cool to see the number of symbols some of the riders are bringing with them. Also, it's clean and really nice to look at--thanks for sharing this.

Are the teams arranged in an order vertically?

Initially I was arranging teams to try to minimize the number of lines that would cross, but that got increasingly more difficult, and I didn't feel like consulting a mathematician for help. So, at some point, I just went with what I had and added teams to the bottom as I collected data. In retrospect, I think Team Sky could have been moved up a bit to make it look cleaner. But honestly, there are too many sources and sinks to make it look truly clean. Bora and Bahrain are the two obvious sinks, with Tinkoff and IAM being the obvious sources.

I'll probably do another one for grand tours. Top 10 GC placings over the last 3 years. So a total of 90 points.

[MattTuck]

This is really cool. Is the underlying data (rider teams and wins) available in a consumable format somewhere or did you gather it manually?

I'm sure an enterprising developer such as yourself could have written a program to parse it all. My process was more manual.

Go to PCS, select the top 10 for each race, ctrl-c, ctrl-v into Excel. Then use conditional formatting, one team at a time to get all the data. And then you have to double check where each rider was in 2015, 2016 and 2017 to make sure you're putting their results in the right place.

If I had known how much work it would be when I started, I probably would not have done it. The layout/graphics (not my forte) took much longer than expected. I thought I was getting into a 3 or 4 hour project. Took easily 4 times that, maybe more.... also included learning Inkscape, which is a SVG editing program, similar to Adobe Illustrator.

[thegunner]

super cool :D

the one nit i have (and this is extremely minor) is to preserve the overall structure of the team results - sky goes from 3 rows to 2 rows which changes the perspective.

i await for sabermetrics to enter cycling.

[Jad]

Initially I was arranging teams to try to minimize the number of lines that would cross, but that got increasingly more difficult, and I didn't feel like consulting a mathematician for help. So, at some point, I just went with what I had and added teams to the bottom as I collected data. In retrospect, I think Team Sky could have been moved up a bit to make it look cleaner. But honestly, there are too many sources and sinks to make it look truly clean. Bora and Bahrain are the two obvious sinks, with Tinkoff and IAM being the obvious sources.

I'll probably do another one for grand tours. Top 10 GC placings over the last 3 years. So a total of 90 points.

I think the order ends up looking great and that Sky balances the image. Initially, I was thinking the teams were ordered by amount of newly accumulated accomplishments. Great to study this and follow all the lines.

[MattTuck]

super cool :D

the one nit i have (and this is extremely minor) is to preserve the overall structure of the team results - sky goes from 3 rows to 2 rows which changes the perspective.

i await for sabermetrics to enter cycling.


It may be minor, but it is valid. Will need to rectify that on my next attempt. You should have seen the first draft. You'd have had major nits with it. FlashUNC can attest to how bad it was, compared to this.

[thegunner]

It may be minor, but it is valid. Will need to rectify that on my next attempt. You should have seen the first draft. You'd have had major nits with it. FlashUNC can attest to how bad it was, compared to this.

i love the effort, so thanks! i think other sports have been put through the wringer in terms of how to quantify team strength (baseball has VORP, WAR, UZR, as well as a bunch of individual metrics to define player performance like FIP - hockey has Corsi possession etc.), I love data analysis moving into cycling. There's unpredictability for sure who wins a race, but i feel like if there were machine learning applied to a race like... flanders, you could reliably place a rider within +/- 5 spots of their finish 90% of the time (at the top 30 at least).

that's why these graphics are fun for me :)