The biggest glass-canon (so far) goes to Babybay who dies a whopping 90% of team fights! Big names like Tviq and Taimou follow closely after. Meanwhile SoOn is un-killable

The biggest glass-canon (so far) goes to Babybay who dies a whopping 90% of team fights! Big name...

That’s a really cool chart. It certainly seemed like Babybay/DreamKazper were dying a lot!

It seems though that this isn’t % of teamfights where they died, but instead it is just # of deaths per teamfight. Slightly inaccurate.

Example: On Dorado attack, Jake (Houston Outlaws Pharah) died, got resurrected, died, got resurrected again, died again. Theoretically all 3 of those deaths would’ve been added to his total for the teamfight.

I think an explanation for why so many great players are on this list might be that the level of their skill paints a target on their back.

It would be interesting to see something like a comparison of number of team fights won vs lost, and whether or not each DPS died during the fight. Theoretically a DPS could die every single team fight and still dominate as long as they did some work and picked enough people first to pull a win.

Things to consider when reading this:

Which heroes these player played

What was the team's composition

Which opponent did they face

Were they on the winning or the losing team (chicken and egg, but something to consider)

The best data visualizations don't need a lot of text, but it is also very important to add context to the data.

It's also a matter of who they faced. For example, Taimou and Effect were going against Seoul, Tviq was facing Spitfire, DreamKazper against NYXL. In addition, while they all played DPS, hero choice matter. You'll find players who played Tracer will die less compared to other roles.

Right, and also they would have died less if the enemy mercy wasn't ressing a lot as well.

With the example you provided if you died 3 times in the same team fight then yes you should have 3 deaths for the 1 team fight. So a 300% death rate per fight would be correct. Granted that this would be an outlier on the data and a larger sample size would be need for a truly accurate number (law of large numbers)

What I think would be better would be to exclude the deaths that were from say a tracer who flanks and gets a kill on say a zenyatta before the teams can even engage each other. So you would have 1 death for 0 team fights. That would cause the data to be incorrect

Correction: Dreamkazper ist first of course, not Babybay... something something Moonlight

Reported for feeding.

Cannon, not canon.

Soon die last or don't die. That's easy.

46% isn't unkillable.