Very good! Thanks for sharing this type of analysis. Just one remark: statistically, some of the win-rate differences (for instance, obsidian vs kelya in heavy hitters) are likely not significant and might be attributed to chance. To say they are statistically/significantly different would require a hypothesis testing, which is quite easy to do as there are online tools for it - I'm happy to help if you want. This does not detract from the value of your analysis though. Thanks again and keep on sharing these, please!