Valve's Pro Dota predictions are broken, and I can prove it.

Help this channel survive ⇢   / @thewondercow   This was all kicked off by some expert analysis by ‪@ccnc3944‬ during ‪@BLASTDota‬ VII. You can view his full comments in context here: https://www.youtube.com/live/V6pDw7Cj... His analysis is great... but that doesn't mean Valve's predictions agree for the right reasons. Looking into this match led me down a wild path that ended with a detailed analysis of all the ways Valve's predictions are not playing 4-D chess. For the full chart of late-game hero win rates in pro games, visit this blog:   / 161350914   0:00 Intro 0:47 What's wrong with Valve's predictions? 1:34 Quick summary of Pro Dota for n00bies (you're welcome here!) 2:34 Quinn's analysis that kicked off this whole investigation 4:36 Is this game's prediction specifically weird like Quinn said? 6:29 Is Valve doing big-brain pre-game predictions based on Draft alone? 8:13 Is Valve estimating success based on late-game hero power? 8:13 Is Valve estimating success based on late-game hero power? 10:19 I borrowed a cute video of a smiling dog to ask you to like & subscribe 12:31 A glimpse of how predictions behave from better models I made myself 15:56 Determining if the model itself changes predictions between pub games & pro games 19:08 Determining if the model changes predictions between different types of pro game 22:35 Skip to here for a summary of the two prior sections 23:46 Explanation of Slicing Data 25:11 Proving Valve's model is forcing closer predictions than it should 27:17 List of confusing findings & glazing Quinn a bit 29:00 Thanks to my supporters! #gaminganalysis #dota2 #prodota2 #prodota #esports #statistics #gaming