Ranked 1v1 Map of the Month - with Avatar

2

I had an idea. In addition to the existing ranked ladders there would be a 1v1 ranked ladder but with only one map. At the end of the month the highest rated player gets a shiney Supreme Commander of <map> avatar. Then the ladder is reset and a new map is selected.

2

Actually an interesting usage of the currently unused king of map avatars tbh, problem would be if you use rating to determine winner it would just be the same guy or handful of guys winning the avatars so you would want to do it during a short like 1 day or weekend period and maybe make tau way higher. Or use the league points to determine winner instead of rating.

0

King of the map is player that has the most success winrate during past month, not the player with highest ladder rating or... br-uh, league scores

0

Wow what a coincidence! It’s a 100 way tie between everyone that only played the map once!

0

Just try again on the next map such a coincidence won't happen twice!

But on a serious note, if you go by rating only a handful of players have a realistic chance of getting the prize, if you go by number of games won, you handicap high rated players. Does anyone have a good idea for a system that takes both into account in a sensible way?

0

Have a separate queue for this, 1 map, trueskill rating for that queue only. Next month halve the trueskill for everyone so higher players don't play lower ones right away.

0

But on a serious note, if you go by rating only a handful of players have a realistic chance of getting the prize, if you go by number of games won, you handicap high rated players. Does anyone have a good idea for a system that takes both into account in a sensible way?

At the beginning of each new map cycle, seed players a new rating for the one map queue that is based on their other rating(s) (ie: global, ladder, etc) with a significantly increased sigma. Then calculate player score in the one map queue using a formula like:
player score = (rating / 100)^2 * win percent * number of wins
where the rating, win percent, and number of wins are based on the values in the one map queue from the current cycle (ie: they get calculated at the end of the month and use the values from that point)

waiting for a mapgen-only TMM queue and Godot

3

Whole problem is that it isn't rewarding or enjoyable to play low rated players if you're high rated and there isn't enough high rated players to make it feasible to have a broad selection of winners for these events.

So the most feasible way to address it is to make it more games with low rated vs high rated more fun. Best way to get that across is a bunch of penalties that can then make games worth more "points" or whatever on the leaderboard for the queue. Impossible to know how much certain penalties are worth in terms of rating so it would just need to be experimented with. Ie person can't use overcharge, person can't make radars, so on. If it's a singular map queue you can make penalties quite refined as you eliminate the variable of map diversity.

Then whether you go by rating or league points it should be fine. A month is WAY too long for something like this. It would need to stay at around a week's length or maybe even just a weekend. Lowering length of time increases volatility in event winners, nobody feels like farming for an avatar every single Saturday or whatever.

This would be in combination with a high tau value so ratings are essentially never settled.

0

@ftxcommando What do we need to try this format? It sounds interesting and I think would entice a lot of people

Sorry for my English. I use translator

0

The penalty part needs to be implemented on the lua side mainly and some general plan of maps/penalties. The rest is just adjusting some trueskill parameters which is trivial.

0

Most wins could play highest rated best of one at the end of each cycle. Would still favor the better player but no getting around that without a handicap.

0