Sanvich Hoovy wrote: ↑Sat Dec 30, 2023 12:24 pm
Publish the list, the people deserve to know
Here is v1 of The List as recovered from my Google Drive. Unfortunately it does not have the original stat collection sheet that was used to build out these numbers, so I'll just have to go off of memory.
The Steam stats page for L4D2 was used to compile this information for everyone listed, it includes all kinds of stats, average infected killed per game, average first aid used per game vs shared with teammates broken down per first aid item, percentage of games won in VS, etc.
We used these stats and weighted them how we found appropriate, certain stats would be weighed more heavily if you have more game hours for example (some people had very little experience but out of the four games they've played they won 3, not great for calculating anything accurate). Health items used on yourself above average was a small negative as it implied you were using up resources more than other people because you were getting hit more and not defending yourself as effectively, whereas items used on teammates was seen as a positive as it had the opposite implication. Obviously there's outliers in some of the stats, Xello has played a shitton of L4D2 so we made sure people didn't get a benefit to certain scores over a certain threshold as 2500 games of doing okay shouldn't outmatch 300 games of being excellent.
Game completion rate also weighed against you negatively as obviously we wanted people who would stick around for the full round. The lower your completion rate, the less your positive stats affected your final score.
Lastly we polled the players involved to get their read on what everyone would rate each other based on how teamwork oriented they were. This had a notable effect on the score, but not one that could otherwise override the rest of the stats in a significant enough way to put you somewhere you shouldn't be. Think of it as figuring out whether your B was a B- or a B+ with the scale being between the highest rating and the lowest instead of just a flat 10 point system.
And here is THE list, the v2.0 that used ELO and broke everything, as it stood after its final use (at least 10 games, assuming Mastermind didn't necessarily participate in ALL 10 with his game count being the highest). We plugged the original list in as the starting point and let the system affect the scores based on team win/loss.
[img]blob:
https://imgur.com/b50204c3-1f8d-4b51-9ee7-95b2d7114aa1[/img]
