Welcome to the PGstats open beta!

We'd like to use cookies to help analyze and create a better experience for you.

You can find out more in our Privacy Policy and Terms of Service.

Just how unpredictable is online Ultimate?

By Jack "Jackie Peanuts" Moore | 05/19/21

The final online qualifier takes place this weekend, comprising the Northeast United States and Canada, and it projects to be a doozy, featuring over 2100 competitors and what appears to be the strongest player pool we've seen yet. Given the wild upsets that have occurred in the other events to approach this scale -- NA Southwest and Southeast as well as Europe, it's reasonable to assume we're in for another bloodbath this coming weekend.

But just how unpredictable have the online qualifiers been compared to in-person Smash events of similar scale? We now have a decently sized sample—10 online qualifiers—to analyze and compare to the established patterns in offline Smash Ultimate (14 S-Tiers dating back to 2019) as well as Melee (looking at 19 major events dating back to 2015's The Big House 5).
To analyze this consistency (or lack thereof), I looked at how many rounds each Top 64 seed finished from their seed projection in either direction. For the sake of this analysis, Seed Performance Ratings of +1 (beat seed by one round in losers) or -1 (underperformed seed by one round in losers) will count the same, as will +2/-2, +3/-3, etc., a metric I call Seed Error (EDITOR'S NOTE: we are now using "Seed Deviation" for this statistic) (the more mathematically inclined will notice this is just the absolute value of Seed Performance Rating, which you can read more about here). Unsurprisingly, Ultimate WiFi is significantly less consistent than either its offline form or Melee, but just how bad is it?
click to expand
The average Melee tournament features an average Seed Deviation of 1.64; that is, Top 64 seeds at Melee S-Tiers finish on average within 1.64 rounds of their seed. At Ultimate S-Tiers, this number jumps to 1.88, a 15% increase, which may be partially explained by the higher average entrant counts at Ultimate majors.
The average Seed Deviation for the 10 online qualifiers jumps another 30% to 2.44, an average seed error just above that of the single most chaotic Ultimate S-Tier (Umebura Japan Major, with a 2.38 average seed error) and just below that of the most chaotic Melee major (Smash Con 2018, which featured two sub-50 seeds in and make runs to 9th, with a 2.54 average seed deviation). The obvious culprits are the four biggest events, NA Southwest (3.8), NA Southeast and Europe (both 3.4) and South America (3.1). But even some of the smaller events, like Oceania (2.0 Seed Error despite just 155 entrants), Japan (2.2 Seed Error, 210 entrants) and Central America (2.3 Seed Error, 229 entrants) prove that it's not just an event size issue, as these tournaments are producing as much chaos as offline events with four to five times the number of entrants.
The return of offline majors is on the horizon, and top players will be understandably relieved when that day comes, because for now? No high seed is safe online.
For more Smash stats and analysis, follow @PGstats on Twitter!