Visualizing playing styles in women’s tennis

On Monday, I took a look at graphing analytically-inferred playing styles to see what the landscape of the men’s game looks like. I reran the exact same process on the women’s players to see how their graph looked. Take a look below (click for full size):

Let’s take a look at what each color in the graph means.

“Hard-courter” is a little misleading here. Everyone in this group has above-average surface factors on hard courts, even if it’s not their best surface. I was a little surprised to see Sharapova and Halep here, thinking they’d get grouped into more of an all-courter class, but it looks like the ability to do well on hard courts is a super-classifier of its own, kind of like elite clay courters in the men’s graph.

This is a reoccurring theme on the women’s side, where hard-court prowess appears to be very important. I don’t know enough yet to say how or why, but it keeps popping up in every analysis I do.

It’s pretty easy to spot the names you would expect here (Errani, Navarro, Hercog, etc.) There’s not a clear split between specialist and non-specialist, which is what you see on the men’s side with hard-courters. But in terms of importance, everything so far is pointing to clay and hard being mirror images of each other on both tours.

Everyone here has hard as their best surface, with average or below-average factors for other surfaces. Don’t look for any French Open dark horses over here.

Everyone here is slightly above average on all courts, but not good enough on hard courts to get into the coveted blue region. The green dots are much more clustered towards the red and yellow areas than the blue, suggesting you don’t really want to be here (sorry, Eugenie).

Petra Kvitova is an indoor hard specialist. You can’t really say that about any top 10 player on the men’s or women’s side, which is why it’s so interesting to see her pop up here. Semi-related: the WTA championships are always held on an indoor hard surface. if she does well enough on other surfaces, she’s in arguably the best position to capitalize on all those points that tournament offers.

Down the road, I’d like to merge these graphs with information more related to the eyeball test. The goal would be to see how well the network graphs match up with known playing styles. There’s plenty of cool machine-learning stuff that can be applied when classifying labels are known ahead of time; it just requires a little grunt work for coming up with the labels. It’s definitely an endeavor friendly to crowd-sourcing. If you’re interested in helping out, drop me a line.