Alright, back to the main course :
SMURFS REPORT W33-35
SILVER LEAGUE
Using a randomized procedure (similar than for the gold league) :
How to read : cf. first post (part II)
Method used : cf. dedicated post.
Quickie :
Over 105 users at silver MMR :
- 20,95% of confirmed smurfs
(13.33% of freelosers + 7.62% of non freelosing).
- 79.05% to 56,19% of regular players.
Some thoughts about this :
- Contrary to what I thought, this data suggest there isn’t more smurfs in silver than in the other metal leagues. Random variation due to the minute statistical power put aside, for the moment evidence suggests that the gold players would be the one suffering the most of smurfing issues. Aside the currently modest sample size, two other factors should be taken in account though
â—¦ The fact that I did not set a minimal career games to reach silver league (as I believe one could perfectly be placed in silver with 50% winrate right away). So that makes one less tool to spot smurfs, and could translate into less non freelosing smurfs detected than there really are.
◦ The fact that for the moment, there seems to be more freelosers. - This put aside, the results are still in the same range than what was gathered in gold, and in my observational platinum report. This suggests our estimates aren’t that inaccurate after all.
- About the method used : finally, I didn’t reduce the amount of MMR randomizations compared to v1.1, as I feared it would decrease the screening power of the survey. I did very minor changes regarding discarded players, as I do not discard users anymore if their profile already gives me enough info to conclude ; and do also classify users as « undefined » when the data is to scarce for me to reasonably conclude. More info about current 1.1b procedure here.
- A lot of players who met one of the smurfing criterias in fact met several criterias simultaneously. For example, one that would have been gold with too few games career would also have a suspiciously high winrate, or on the contrary have ≥ 3 freeloses. This makes me think we’re in a good place regarding those criterias, and I feel satisfied with current version.
FUN FACTS
- If an user reaches a league with too few games in his career, that can allow to confirm him as a smurf (for example platinum MMR with 10 games in his whole career)… but it seems the reverse is also a hint. For example I’ve confirmed a freelosing smurf who had 12 000 career games while still being around silver3-bronze 1. Even extremely non gifted users would’ve gotten at least gold by that time. I’ve seen regular players staying in silver for more than three years though, so this should be only considered as a hint.
- It also seems that an irregular progression (bronze → gold → bronze → gold) is a good clue.
Now for the funny part :
Once upon a time, was a lad known as « Ok ».
Ok was a high league player, but stagnated. So, one extremely gloomy day, he decided to stop playing in diamond 1, and to freelose 1000 MMR+ in order to feel like the man again. And it worked wonders, he finally could get the 20-30 games winstreaks he had always dreamed of. By continuously slaying unsuspicious lower leagues, he felt he had regained his virility. He felt like a Flash among the men.
But good things aren’t meant to last. One Saturday in the evening, he was matched against another lower league player, that we will call Pony. Ok didn’t noticed, but the pony seemed shaken by his manly apparel, as if he had seen a ghost.
Full of confidence, Ok deployed a splendid exemplary of skill : with his 235 APM, he hit every macro timing magnificently. He even killed the opponent’s scout who was trying to harass his CC building SCV. Such a foolish attempt !
But upon preparing to lay waste on the pony’s mineral line, Ok noticed the wall was made with three supplies. Suddenly, he didn’t feel ok anymore. An alert immediately sounded : his base was being assaulted by a small squad of marines, and they weren’t friendly. Listening only to his 200+ APM fingers, he got back, and tried to temporize an SCV pull. But while doing so, the second Pony’s SCV had (barely) had the time to land a bunker. He had just lost the game to a vulgar pony. And Ok wasn’t ok with that at all : he did ragequit, and by that had his first real lose in a while.
What was the difference between the pony and the 20 previous players ? Skill ? Nah, otherwise the pony wouldn’t belong to an intermediate league. It was information, as contrary to the others, the pony had learned to recognize smurfs right away.
Moral of the story : would the devs let the ladder be ruined by smurfs, we regular players would still have a chance to deal with them by ourselves. Know your enemy, goes the saying. Even more so in a strategy game.
PS : first post updated. See you in about three weeks for Bronze league ! 