I have a clean dataset with the last 20+ years of ncaa tournament games (round, seeds, result, score) along with ~100 traditional and advanced team stats from multiple public sources as they were pre-tournament. I\u2019ve done a lot of feature engineering and can add those metrics in too (ex: 3-pt % to opponents 3-pt defense, raw and normalized by diff SOS type approaches).<\/p>\n
It\u2019s nothing crazy extensive (no player stats, injuries, trends) but it\u2019s cleaner and more comprehensive than anything I\u2019ve found available for free download \/ scraping.<\/p>\n
I put the scripts together a few years ago with non-trivial code effort and manual QC (name formatting etc). It wouldn\u2019t be particularly difficult to reproduce for a decent programmer. I\u2019m sure AI has made that type of process more accessible but it\u2019d still take some time for most.<\/p>\n
Having never sold a dataset is there any value here? I\u2019m not expecting much but the work is already done.<\/p>\n
I\u2019ve started the process of including regular season games (stats at game time) if that would help but probably won\u2019t finish without understanding value. Same for game lines \/ betting info but only if the dataset is useless without them. They\u2019re messier to pull. <\/p>\n<\/div>\n
submitted by \/u\/yourfinepettingduck <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
