There are have been two instances recently where I have suspiciously found myself not at the top of the rankings according to the Kangaroo API. Thankfully on the official leaderboard hosted on this very site I have added ranking integrity checks to prevent this false information from being distributed.
However if we look at ranking history we can see this questionable data that appears to confirm that there were some periods where another player actually did have a higher rank than me:
How could this be?
Is it possible that another player genuinely played better than me?
There must be another explanation…
If we delve into the winrates we see something very curious. Obviously I have the highest winrate by a significant martin, but if that’s the case how is it that some players can supposedly at times catch up or even surpass my rank?
Delving into the head to head winrates highlights the bias even more:
Let’s take the player tann as an example. Not only do I have positive winrate against this player, but I have a more positive winrate than them vs every other active player bar one.
Yet somehow on 30th June the ranking system deemed it appropriate to give this person a higher rank that me.
Looking at the change in ranking per game reveals the sabotage. Up until that date, my rank increased by an average of 13.2 per win, and dropped by 19.3 per loss. However for tann, the average increase per win was an astonishing 15.8 and the average decrease was only 14.1 per loss!
Based on these figures I need to maintain a dominant 59% winrate just to stay at my current ranking. Whereas tann needs only a meager 47%.
Clearly the developer of the Kangaroo platform has is deliberately deflating my ranking for currently unknown nefarious reasons. On the official leaderboard on this site I have added a dynamic weighting to the X axis scale in order to more accurately represent skill differentials. Hopefully in future the Kangaroo platform will be updated to remove the current bias against skilled players.