100000 Champions Simulations

I wrote a program to simulate champions 100000 times and these were the results:
https://ibb.co/ftpn6bs

The most likely winners were Gambit (33%), followed by Acend, Envy, Liquid and SEN all at around 14%. See the link for much more detailed data.

Notes about the simulation:

• Vivo Keyd is likely underrated because none of their data was with Mwzera.
• X10 is likely underrated because none of their data was with Patiphan.
• Liquid might be underrated because some of their data is pre-Nivera.
• "pts" and "diff" can be ignored, I originally designed this for football.

Interesting observations:

• Gambit has a higher chance of winning their group than any other team does to even advance.
• The simulation only gave VS a 2% chance to win. Personally I think it should be at least 10%. I think the program views VS poorly because it thinks they should've beaten the local Korean opposition by bigger margins (13-2s instead of 13-7s). But imo they just weren't tryharding those games because they knew they would win easily anyway. I guess we'll see who's right in a few weeks.
• Crazy Raccoon is more likely to make it to semis than Fnatic is to win the event.
• Team Secret has a 22% chance to advance while KRU only has 5%, yet they both have about the same chance to make the semis (0.5%).

how tf can brazil region who never showed a shit in other regions underrated?

Does this matter tho

hm

thats actually some next level work..Congrats

Desperado_o [#5]

thats actually some next level work..Congrats

no cap

ryugafpsz [#2]

how tf can brazil region who never showed a shit in other regions underrated?

it's just saying that VK added a player since their last matches and the simulation doesn't account for it, it's up to you to decide if that makes their chance go up or not

It’s just speculation but that’s a sick work, good shit

huge work, interesting info for gamblers

how did you get those figures, mind explaining, really impressed, but like what was the difference in parameters and what were the specefic characteristics of each team

Good job Max Verstappen!

Good work but what did u use for this
Looked like it took a long time

ShadowisBAD [#12]

Good work but what did u use for this
Looked like it took a long time

brief overview of process:

1. downloaded all of VCT stage 3, LCQ and a select few other tournament from vlr (automatically ofc)
2. used this data to calculate a base rating and uncertainty for each team using this software
3. each simulation started by slightly randomizing the ratings based on each team's uncertainty
4. simulations would play out the entire matches round by round, using the ratings to calculate the % that each team would win the round
5. teams progressing from groups were placed randomly into playoffs so that group winners always played second place teams
6. repeated 100000 times

since_beta [#10]

how did you get those figures, mind explaining, really impressed, but like what was the difference in parameters and what were the specefic characteristics of each team

#13

So it's a week later now, and I now have 10 million simulations! I also fixed a bug that was significantly affecting Group D.

Results:
https://ibb.co/02Szr3T

crazy_igel_fan [#15]

So it's a week later now, and I now have 10 million simulations! I also fixed a bug that was significantly affecting Group D.

Results:
https://ibb.co/02Szr3T

Yo this is actually hello dope and innovative. Despite what the simulation think, who do you think will take it all

YvngKurry [#16]

Yo this is actually hello dope and innovative. Despite what the simulation think, who do you think will take it all

I’d say Liquid or Gambit, rooting for C9 tho :)

Facts. Great work my guy

Ngl this is hella cool, did you make it from scratch ands i it made in a way where you can easily add more teams and.mlre data I'd love to see this type of thing run at every tournament and see how accurate it can get

Dope