I’ve got a script for crawling the results of the CB matches and will put some highlights in this thread.
For those curious, the endpoint is https://terminal.c1games.com/api/game/competition/27/matches
Given that the submitted algo’s elo may not be the highest for a given player, I’m giving players elo credit for their top on the global leaderboard.
Round 1 saw a lot of “starter-algo” losses and no “upsets” worth pointing out. Also none of the global top20 players played in Round 1.
Round 2 saw all top20 players win their match. Interesting each matchup had an elo difference greater than 1000. Is this just chance? Or did C1 do some real seeding?
No top20 losses in Round 3, but the #1 seed lost in Round 4!
Round 4 Top20 loss:
Masteroliw(Highest elo: 1806)'s algo ‘pizza_time_algo_V0.2.12’ beat kkroep(Highest elo:2000)'s algo ‘EMP^ERROR_v1.0’ in 33 turns - https://terminal.c1games.com/watch/627743 (pizza_time_algo_V0.2.12 was created yesterday and currently has an elo of 1617 which is why it was seeded at 128 for this competition)
Notable upset - the user who won the UIUC competition was beat:
Round 3 Upset - Prohibidado(Highest elo: 1427)'s algo ‘v1.2’ beat mjtomei(Highest elo:1799)'s algo ‘makethempay3’ in 24 turns - https://terminal.c1games.com/watch/627648
Cinderella stories - highest global elo under 1600 that are advancing to round 5 (beyond zachwvk from above):
ROBERT SOJAK(Highest elo: 1569)'s algo ‘Specimen-D1’ beat ZigZagNinja(Highest elo:1841)'s algo ‘ggezv1’ in 14 turns - https://terminal.c1games.com/watch/627724
notingmyself(Highest elo: 1576)'s algo ‘No Name Reworked’ beat richard11235(Highest elo:1820)'s algo ‘CashOrCredit2’ in 74 turns - https://terminal.c1games.com/watch/627729
Krysuu(Highest elo: 1598)'s algo ‘pyra-v0.3’ beat fezhub(Highest elo:1804)'s algo ‘tunnel3’ in 9 turns - https://terminal.c1games.com/watch/627726
I believe the discrepancy in your chart comes from each algo’s elo being updated as it plays in the global competition. So the seeding was as snap shot in time and the elo’s we see now are “live” and not even a representation of the elo at the time the match was played.
@n-sanders To show that the initialSeed is not correlated to the user’s elo on the global leader board, I included that into the plot. Notice: There were 2 users, which had no entry on the leader board (maybe a name change and I could not use user id’s because they are not available in the leader board). For those, I just used the following calculation: (next + previous) / 2 (they are -24.5 and 39 elo).
From this we can make out that the data is correct because there is no point significantly below the line (which would be impossible because the leader board always shows your best algo and thus your orange point (leaderboard algo) could only lie above the submitted one).
Because of that, some are a touch below their blue point.
It also means that quite a few did not select their highest ranked algo.
Cinderella update - one user with a top global elo less than 1600 advances:
ROBERT SOJAK(Highest elo: 1569)'s algo ‘Specimen-D1’ beat wz443(Highest elo:1667)'s algo ‘wait’ in 10 turns - https://terminal.c1games.com/watch/627756
13 of the top 20 users (based on highest global elo) advance to round 6.
I wonder if the users missing from the global deleted all their algos? Since deleting it doesn’t remove it from C1’s servers, that could explain why they are still in the competition but not on the leaderboard?
I’m one of those people! My algo was added the morning of the competition, so it hasn’t had time to climb very far in the global competition.
GG to all, especially to AdrianMargel his sawtoothV2 algo. Watching the replay was very painful to me because I knew that even if I win games on my side of the bracket I’m going to lose to these value focused algos which domiated left side. Anyway I’m happy that i did end up in 2nd place.