Alliance Wars match up

Alternatively you go for an Elo-based system which is entirely results-based and is dependent only on how well you are performing against others who have performed historically as well as you.

The problem with this is that it has to start somewhere, and will generate a lot of noise until it settles. Brobb shared a link to this suggestion earlier in the thread, but in case it was missed, here it is again.

Then again, I think Top30 could work if done properly. It would need to be distribution based (i.e. so you don’t have an alliance that is top heavy in very powerful heroes but with no depth being matched against one that has more average heroes, but more across the board. You’ll find the average attacks against the strong defences just won’t fare well. So the match needs to factor in the distribution of power across the overall set of in-scope heroes and match that to another that has a similar ‘shape’ (addressing @Gryphonknight’s point above) . This would require a reasonable knowledge of statistic and quantitative analytics to do well, however… but could maybe be faked by simple use of weighted average (giving higher power heroes more emphasis in the final match scoring).

This distribution based matching I suspect would be harder to game as you would naturally be matched against others that are shaping their squad similarly to yours. The main flaw with the system, however, is the flaky measure of ‘power’ in which the algorithm will be assigning obviously stronger heroes in Raiding/AW context the same weighting as much less suitable heroes that happen to have the same ‘power’ score. Even so, I don’t think there are too many examples of these types of discrepancy, so may work out ok on average.

I don’t envy SG the challenge of trying to get this one right… :thinking:

2 Likes