Last two weeks’ stats about the proportion of smurfs encountered at my MMR :
SMURFS REPORT - W20 & 21-2020
How to read : cf. above post.
TL;DR :
Over 80 users played :
- 17.50% of confirmed smurfs.
- 18,75% of suspicious users.
- 82.50% to 66.75% of regular players.
About those « suspicious users », those are the players which had profile inconsistencies strongly suggesting this was not their main account, but for whom I couldn’t find freeloses in their recent history. The problem here is that in some cases, it was obvious they were indeed smurfs :
— one user playing at D3 MMR with only 25 total career games on his account
— two of them had no loses at all in their 1v1 history
— one just confessed, etc.
but I still couldn’t find back the freeloses in their history. So couldn’t consider them as confirmed freelosers, even though they were smurfs…
I decided not to stray away from my initial protocol, but this highlights that the freeloses searching method provides an estimate which is in fact lower than the real proportion of users smurfing (which is consistent with first post, as we saw freelosers are only a part of smurfs).
One could suggest to change the protocol to include those, but then at one hand it may not allow to merge the future data with the previous, and that would require to define other objective criteria.
So for the moment, I’ll include the averaging between suspicious users and confirmed freelosers, which we could call Averaged Smurfs. But I’m not satisfied with this, coming up with objective criteria for smurfs without recent freeloses would be a better option in the future.
So we’ve got :
- Confirmed smurfs : 17.50% (low estimate).
- Averaged Smurfs ≈ 26.87% (average estimate).