I think the players who are actualy at least 5 ranks better than their shown ranks are getting more at the site here.
i think its bad because normal players dont want to have lichess or suroundings here

If you feel one of your opponents is sandbagging, please report them (by clicking their username and the “Report” button). We (moderators) will go through their history and see if they are manipulating their rank, and either warn or ban the account. That’s currently the only way we can catch sandbaggers.


I have no idea if sandbagging is really a big problem on these servers. Besides i would be surprised if these players would come and play go here.

In Wikipedia sandbagging is defined as “deliberately playing below one’s actual ability in order to fool opponents into accepting higher stakes bets, or to lower one’s competitive rating in order to play in a future event with a higher handicap and consequently have a better chance to win”.

How does OGS detect sandbaggers?
Is there a clear line that divides an unlucky losing streak from deliberately lowering your rank?

There are loads of sandbaggers on OGS, who keep their rank low in order to clobber weaker players. I believe I handled more sandbagging cases than any other mod, devoting perhaps half my time to those cases, but I think the mods would rather not discuss methods of detection. Sandbaggers want to get easy wins so they can feel good about themselves (or for purely sadistic reasons in some cases).

1 Like

O sure!
I simply found that describing them as an army of zombies jumping from the chess world to the go world was a bit unrealistic.

Fair enough (I was just curious).

Somewhere I read that some sport’s associations (maybe it was the USGA, golf) work with automatic sandbagging detection. If a player’s rank changes too abrupt (either up or down) in a certain period of time, that player is automatically labelled as a sandbagger (and also automatically reverted to the old rank).

For go this could be translated as too many rank changes in a certain number of games. (Probably wise to exclude the first 10 matches in the unranked phase of a new player.)

Is this something for OGS?

1 Like

statistical research should be done first to know usual length of lose-streaks

1 Like

I don’t like the idea of automatically labelling players, because I can easily figure at least one case where rank has a lot of fluctuations which definitely are not intended as sandbagging.

Moreover: sandbagging is against site policies, so sandbaggers are warned or banned. I don’t like the idea of automatically warn or ban users.
If the algorithm is wrong, we’d automatically warn or ban the guiltless.

While with an automatic warning to the mods team, the only issue would be an overwhelming amount of warnings for the team itself, which would then be encouraged to fix the algorithm or trash it.

1 Like

My first game against a human on IGS in 2016 was against a sandbagger. We were both 17k, and he beat me by 231 points. That guy still plays there almost every day and crushes his opponents bu resigns 50% of his games while being ahead to maintain a 17k rank.


How is that?
Is there no one willing to report that?
Oh, sorry: i misread IGS! :grin:
Anyway: don’t they have such policies and mods on that site?

1 Like

Yes it’s possible to report a player to a mod. Nothing about sandbaggers is written explicitly in the FAQ, which doesn’t necessarily mean that sandbagging is allowed. Maybe no one cared to report that user.

there can be automatically sorted list of everyone by how likely someone is sandbagger
mods can just check those who are on the 1st place if they have free time


That was what I meant. No public naming and shaming.

Is this your assumption or can you back it up by facts?
Amount of warnings depends on the settings of the automatic detection.

This might work!

1 Like

What I meant is: you need a trigger.
A difficult trigger to setup.
If it’s too loose, it won’t click at all.
If it’s too strict, you’ll end up with a lot of false positives. Which is the case I was thinking of.

1 Like