if these stats will somehow change in a way that will prevent them from being biased, it'll be the most useful contribution of dotabuff since forever.
what would be even more awesome is showing item/skillbuild winrates and buildrates for different brackets so these stats aren't completely irrelevant, right now the guides are the only way of roughly estimating what ppl do and what works
given how almost every 5k+ game is parsed with truesight i don't see the problem with doing that, even if it was a plus feature - i'd probably even buy plus if that was implemented
Early game items like wand and aquila have terrible winrates cuz it only judges item winrate if the item is in the inventory at the end of the game, not if it was purchased. Since anyone who buys aquila will usually replace it buy the time the game ends if theyre winning, and wont have the money to replace it if theyre losing, it will have a terribad winrate.
Please sign in to post comments.
I think we all know that the correlation between talent tree and win rate might be misleading at the moment because, for example, players tend to opt for "Reduce respawn time" or "Increase gpm" when being in disadvantage. Hence, how about taking into account the net worth difference at the moment the talent is upgraded? Perhaps we can have more meaningful stats on win rate between 2 talents when the net worth difference is 0-5k, 5k-10k, 10k-15k, etc...