Algos will now continue to be given additional burst matches until they have had 15 burst matches or have lost 3 burst matches. We expect this to slightly reduce the total number of ranked games being played, but dramatically increase how quickly a med-high level algo can receive good feedback.
This is the last update we planned out during the matchmaking overhaul. This was the first epic I spearheaded so its pretty exciting to push out the last change. As always, let us know about any issues, comments, or concerns.
There were also a few minor bugfixes and backend changes, but nothing major to note.
I’m loving this. Feedback rate is spectacular. Within 2 hours my newly uploaded algos are reaching around 2150 to 2200 ELO and are starting to face some versions of algos made by Top 10 opponents. In 2 hours!! This used to take about a week! Very nice.
That it stops at 3 is an awesome important addition. I still dont use the local client, so debugging and testing doesnt really need that many matches…
Ironically the winrate of my algos seems to not be affected too much by the ELO it has. (Epiphany_1.3 also lost to quite a few 1500, 1600 algos…
Its crazy that it was ever this bad looking back, glad we got it to a much better state
I figured three would be a good number, since 3 losses should be able to give some idea of what needs to be improved across a decent enough variety of games
I love the new system. I just uploaded an old algo of mine that floated around the 30-50th rank for a while an hour ago, and its already around what a 30th ranking algo would be, and still climbing. Really useful for getting practice matches in