Flash Flash Revolution - View Single Post - Poll: Which global skill rating system is best ?

**Zlyice** · 05-23-2021, 04:48 PM

There are two main reasons I'm in favor of a weighted average. One, as Flynn mentioned, is that a weighted average does a better job of giving resolution to a player's top level of play. The current system does give a pretty strong weight to a player's top score, but I see this more of an issue with the current weights as opposed to a weighted average in general. WirryWoo's calculation earlier in the thread seems pretty reasonable to me personally.

Secondly, an unweighted skill rating is only going to be as representative as the full scope of scores going into the calculation. If we're considering, for example, an unweighted average of 100 songs, this would require a player to play enough things for these 100 scores to be reasonably representative of their level of skill, which could take a considerable amount of time. There's a lot of potential for an unweighted average to disproportionately rank more active players ahead of players who play a bit less but are ultimately a little more skilled.

05-23-2021, 04:48 PM	#9
Zlyice Slightly unpronounceable Lead Difficulty Consultant Join Date: Dec 2009 Location: Massachusetts Age: 33 Posts: 264	Re: Poll: Which global skill rating system is best ? There are two main reasons I'm in favor of a weighted average. One, as Flynn mentioned, is that a weighted average does a better job of giving resolution to a player's top level of play. The current system does give a pretty strong weight to a player's top score, but I see this more of an issue with the current weights as opposed to a weighted average in general. WirryWoo's calculation earlier in the thread seems pretty reasonable to me personally. Secondly, an unweighted skill rating is only going to be as representative as the full scope of scores going into the calculation. If we're considering, for example, an unweighted average of 100 songs, this would require a player to play enough things for these 100 scores to be reasonably representative of their level of skill, which could take a considerable amount of time. There's a lot of potential for an unweighted average to disproportionately rank more active players ahead of players who play a bit less but are ultimately a little more skilled.