I decided to make an updated gdoc for the complete war rankings, with data on round eliminated (since that's the primary determiner of rank). Here it is:
https://docs.google.com/spreadsheets/d/18KKPCNUTssvsQcg8rMtCiOC1EbowaXdeF7PdUbZFGDQ/edit?usp=sharingNow, there's one thing before we can finalize the rankings. If our metrics are Round eliminated then Winrate, then Death and Entropy are tied for 5th. Of course a tie for 5th isn't that big of a deal, but historically we've awarded distinct ranks. So we need to devise some tiebreaker.
Last war duke broke it based on either #Wins or Vault size. Both are good metrics. A team with more wins had to work harder to maintain the same winrate over many matches. Also vault size at the end is not entirely discountable: a team with 49 cards left was just a bit away from surviving another round (a stray transmute may have done it). But one at 0 cards would've needed more to survive.
Previously, we've had the same situation thrice:
Time/Gravity - War 7
Air/Death - War 8
Time/Death - War 10
In the War 10 case, both #wins and vault determine Time as winner in tiebreak. In War 8 both teams had the same W/L record, but by vault size Air wins significantly. In war 7 they also had same record and the vault sizes weren't stated clearly, but by the number of spoils after Round 8 Time was a clear winner. So the historical rankings stand by these two metrics.
But in our case, the two metrics favor different teams. So we'd have to choose one to be superior. #wins seems more objective, but do keep in mind that cards left in vault played a role in causing this debacle (Death revived because they were only a few cards below 50). So which do we choose?
At that point, I looked for a more objective tiebreaker. The best answer? RPI. Not all W/L records are created equal; some teams have faced worse opponents than others. We don't have easy stats for RPI in case of War 7, but we do have kakerstats for War 8 & 10. I added Winrates tab to War #8 (
https://docs.google.com/spreadsheets/d/1gbh58GOlV0X8Jt4LmdddVGL8TkY0rybhBNPA9pwvqP8/edit?usp=sharing) and checked War #10 (
https://docs.google.com/spreadsheets/d/1u7ltro309bCjm9D18V1nzJ-N2QGSca8L1K6kbGIbfVc/edit#gid=3630884), and it's surprising. The teams that won the tiebreak above had lower RPI. Indeed, in War #8 Air had the lowest Strength of Schedule, so arguably they had the easiest time of all teams. Doesn't that mean Death deserves to win the tie over them?
Among the simple metrics, #Wins and Vault Size are both good. But I think we should resolve this by something more objective, something which accounts for how hard it was for a team to acquire their winrate. So I propose that all possible ties be broken by RPI. This would switch those historic rankings. RPI shouldn't leave many ties since it's very precise, but then we should decide which of #Wins and Vault Size is better.
