Opposition Research - C1Ryan do not read

Well, after all his talk about adaptive algos, @C1Ryan’s current top algo on the global leaderboard is a direct clone of the static SawtoothV2:

	"id": 757080,
	"turns": 33,
	"round": null,
	"winning_algo": {
		"id": 32220,
		"name": "UMICH-Special",
		"compilationStatus": "Completed",
		"lastGameAt": "2018-11-02T14:53:03.132Z",
		"elo": 1770,
		"language": null,
		"createdAt": "2018-11-01T15:54:46.715Z",
		"crashCount": 0,
		"user": "C1Ryan"

It would seem he’s trying to figure out how to beat the current state of the meta too!

Sarcasm aside, since only a user’s top algo gets shown on the leaderboard there’s a decent chance the performance of UMICH-Special will cover up whatever he’s cooking up for the competition. But if someone happens to notice a match against C1Ryan in their play history, we could discover some of his other algo ids :wink:



I uploaded this to test play-by-hand vs sawtooth to save me 20 minutes of cloning it myself lol. I promise not to use it for the C1Ryan competition. I might take one of your algos instead >:)

(Just kidding ofc, i’ll actually be playing to win)

1 Like

Just take all of our best algos and dump them into some sort of reinforcement-learning-statistical-regression-genetic-selection monstrosity and upload the resulting Frankenstein’s Algo :rofl:


Haha, this is literally something I have started working on :laughing:.

1 Like

Hello n-sanders, how do you get this information? Is there a way to let your algo collect this?

1 Like

There’s a public API for getting leaderboard information:

And I crawled that with a script looking for C1Ryan’s name which got me the id of his top algo.

I used that to get the match history from this public endpoint:

Same here and I got quite horrified when I calculated the size of the state-action space haha: Lots of work to do :rofl:

Well, C1Ryan has employed a new tactic and put up his algo named “starter-algo-ZIPME 2” :smiley:

It isn’t what I would consider “adaptive”: https://terminal.c1games.com/watch/845190 (he’s player 1).

Edit: the linked match is the only one he’s won on the global competition so far (it’s record is 1-8).
This is his highest elo algo at 1407 at the time of this posting.

I’m starting to think he’s just messing with us here …

1 Like

C1Ryan rewards the guy who beats his competing algo with 100,-. If he just deploys the starter algo that will be the most underwelming waste of prize money I’ve ever seen. Because of that I think he genuinly has a strat. However, I don’t know how much testing he got out of it not playing much on the ladder. Doing that requires a lot of free time to think the game mechanics through. Admitedly this is his job, so maybe he pulls of a devastating algo, who knows.

1 Like

He could technically be playing against all our algos without putting his on the ladder, since he has access to the servers. He at least has all the special bosses.

He mentioned the maze algo he put up was just so he could test by hand in the playground, so I imagine he’s been doing a lot of local testing offline.

I’m looking forward to seeing what he’s cooked up!

He could also just be fixing the starter algo… I made a thread yesterday about a bug, and he responded in another about another issue.


I uploaded Starter_algo to production as a sanity check to make sure it was still working after making some changes (I fixed the issue @Ryan_Draves is referring too, and a few other minor ones).

I also will be pulling algos and playing against them, but I am pretty familiar with most of the strategies in the game from casting tournaments and watching all the matches that end up on the forum, and I think I have a really strong ‘core premise’ for my strategy, so we will see how it does going in relatively blind.

I’m not sure that was his (maybe?) but that’s my bot in blue that I just uploaded.