[HN Gopher] SpawELO - small free matchmaking system for LAN parties
       ___________________________________________________________________
        
       SpawELO - small free matchmaking system for LAN parties
        
       Author : Spawek
       Score  : 85 points
       Date   : 2024-11-02 10:44 UTC (12 hours ago)
        
 (HTM) web link (blog.spawek.com)
 (TXT) w3m dump (blog.spawek.com)
        
       | digitalPhonix wrote:
       | Has anyone tried something that isn't based on ELO/trueskill for
       | team-based games?
       | 
       | It feels to me like summing/averaging the ELO of a team to do
       | matchmaking is:
       | 
       | a) just a hack to fit a square peg (team based matchmaking) into
       | a round hole (individual matchmaking)
       | 
       | b) losing a lot of information about how intra-team dynamics
       | would affect the result (as in player A and B play well together,
       | better than their individual ELOs summed; but A and C play poorly
       | together, worse than their individual ELOs summed)
        
         | pacbard wrote:
         | You could assign ELO at the group level. It will increase the
         | number of possible groups by C(N, g) where N is the number of
         | players and g is the group size. It could work if the groups
         | are stable enough.
         | 
         | If you have individual ELOs, you could pre-seed the ranking
         | using averages and then let the algorithm take over.
        
         | KaoruAoiShiho wrote:
         | Sorry for pedantry but Elo is a guy's name, not an acronym.
        
         | Scene_Cast2 wrote:
         | Agreed. ELO is a crude but simple statistical model, and its
         | primary benefit is that it's easy to understand and reason
         | about.
         | 
         | I'd love for there to be a multidimensional player skill vector
         | (aka "embedding") that encapsulates more information as well as
         | a more nonlinear model to go on top of it.
         | 
         | For example, in many games, a team usually needs a support
         | player, but a single number isn't enough to capture that for
         | the purposes of matchmaking.
        
           | Spawek wrote:
           | Author here: Elo is commonly used and most gamers understand
           | it. It's nice to be able to show each player their Elo.
           | 
           | I love the idea to use per-player embedding and to add a
           | nonlinear model on top. I think it may improve the model
           | dramatically, but I'm afraid that the model may be
           | overfitting strongly as we have only 35 games in the
           | historical for now. Anyway - I think I'll try this out.
        
             | Scene_Cast2 wrote:
             | Yeah, 35 games is hard to work with. My rule of thumb is to
             | not go above 1 parameter per 10 data points in order to
             | prevent overfitting and memorization, so I'm not sure how
             | well it will work in your case.
        
         | dragontamer wrote:
         | Yeah. Sports science is about evaluating individuals in a team
         | and trying to come up with statistics that make team picking
         | more reliable.
         | 
         | Moneyball is a good movie that dramatizes the dawning of this
         | science.
         | 
         | Fantasy Football is a game version that's become popular for
         | online gamblers.
         | 
         | -------
         | 
         | I've seen a paper or two floating around on the subject. But
         | I've forgotten the names of them by now and I dunno any good
         | references unfortunately.
         | 
         | Hope you find something in your search!
        
         | InvaderFizz wrote:
         | Reminds me of a group of friends that played call of duty
         | together in the CoD4/WaW days. Individually we were mid-tier
         | players on average. As a group, we completely dominated. We all
         | knew each other's strengths, weaknesses, and general tactics.
         | There was often no need to communicate specifics because we
         | knew where each other would be and what we were doing.
         | 
         | We were always giddy when we got matched with a clan,
         | especially if they were a mouthy crew.
        
         | leni536 wrote:
         | I would guess that in team sports/esports betting some people
         | use more advanced models, but they obviously keep it secret.
        
         | DigiEggz wrote:
         | Excellent point about team dynamics. When trying to dynamically
         | guess a group's Elo, it could be interesting to try and create
         | a game-specific radar chart for each player and add those
         | values instead.
        
         | tonyhart7 wrote:
         | its tried and tested by multiple online games used in
         | dota,cs2,lol,val,cod etc
         | 
         | it is not perfect but close enough that it didn't matter tbh, I
         | mean multiple company do it for good reason
        
           | Jach wrote:
           | There are a bunch of GDC talks about matchmaking systems over
           | the years too for those curious about things that have been
           | done. For a successful game a lot of things have to be taken
           | into account, not just skill.
        
       | joaquincabezas wrote:
       | what about trying with Shapley values?
        
       ___________________________________________________________________
       (page generated 2024-11-02 23:00 UTC)