Art & Futures
Visually it's cool yes, but it's very hard to map tennis like this under some arbitrary ratings based system like it has used.The idea is very cool, but some odds are really weird. I understand that they took the match ups into account, but still, Kenin, Ostapenko are Garcia odds are too high.
Tennis Abstract actually recommends that you use a 50/50 mixture of the overall ratings and surface specific ratings to generate forecasts, which is why Tennis Abstract's own forecast differs from this person's ratings which misguidedly uses only the grass ratings.Visually it's cool yes, but it's very hard to map tennis like this under some arbitrary ratings based system like it has used.
You shouldn't confuse this with odds either, because this won't reflect the actual odds too, I mean I'm sure there's tons to pick out on this graph, but eg. according to this graph Kasatkina would be comfortable favourite to beat Azarenka in R2, there is no chance that is the case. It says it uses Tennis Abstract's ratings which are here: Tennis Abstract: WTA Elo Ratings, which is true, but it's using the grass rating, that might sound right, but I'd imagine you'd be better off using the overall rating. The dataset on grass is just so small as players are playing 1-2 events on grass a year that you will never get a true rating, and there isn't much recent data to judge how someone is currently playing either.
Tennis Abstract system looks way too sensitive to recent results anyway. It hasn't updated for Pliskova winning Eastbourne yet, but they're grass ELO has Kenin 3rd favourite, Riske 5th favourite, Tatjana Maria 7th favourite. I mean come on, no-one actually believes that.