@Speck: based on the very small dataset that CSL has, the Czech Republic template from the 1v1 league (
https://www.warlight.net/MultiPlayer?TemplateID=667635) has the underdog (as per Elo rating at the start of the game) winning 23% more often than expected. Without looking a bit further into the games (as to whether players going into the game were underrated or overrated going in- CSL's first 10 cycles fluctuated rather wildly), that might be the best bet for a strategic player who suspects they're worse than Platinum and wants to even the odds (or odd the odds? odd the evens? idk) to make Platinum less likely to succeed. On the other hand, Rise of Rome 0% Luck (
https://www.warlight.net/MultiPlayer?TemplateID=655095) has the overdog winning 23% more times than expected so someone who believes themselves to be better all-around players than Platinum might want to opt for that. Guiroma (
https://www.warlight.net/MultiPlayer?TemplateID=655101) interestingly has neither the overdog nor the underdog winning more or less than expected, so maybe if Platinum wants a single template for the entire challenge, that could be a go-to. I mean, Guiroma's also a really good template regardless, and definitely fun af to play on as well.
The old Strat 1v1 template (ME, 0% WR-
https://www.warlight.net/MultiPlayer?TemplateID=655094) had the overdog winning 18% more than expected, while closest thing to the new one (ME, 0% SR -
https://www.warlight.net/MultiPlayer?TemplateID=655104) had the underdog winning 11% more often than expected.
Of course, the caveat: these data come from absurdly small sample sizes (10-13 games) and I wouldn't personally consider them anything near statistically significant- even on the CSL sheet, none of them have p-values below 0.05 (i.e., the probability of obtaining these seemingly biased results purely due to chance given that the template itself isn't biased ranges from 7% to 30%- so no conclusions should be drawn from these data unless you just have really low standards for statistical significance). These are just some starting points that may be used to construct further hypotheses, ofc, but I'm offering them because I don't know of any serious efforts to collect this sort of data.
Edited 11/25/2015 20:50:25