Quote:
Originally Posted by belvedere91
Another thing on CH that I dont think the market accounts for is that players' profiles can change a lot over time especially as they get older. Phil for example, is really a different player with respect to the competition than he was in 2007 when he won the Players. Completely different actually. He used to be longer and a much better iron player and just way better all around. Giving him the same bump for course history is kind of non-sensical if you know that he's a different type of golfer now. Same could be said for many other guys going through ebbs and flows in their career. Guys learn how to putt, get longer, improve their short-games every year. There's so much more nuance that just doing regressions on the entire player population and applying their indiscriminately. If people disagree or have better ideas, please share! I could talk about this **** all day lol
i'm no expert but I assume recency is weighted heavily in most models
ideally my dates would be rolling but that's too complex for excel but i assign multiples based on how recent so this year is 10x, last year 5x, year before 4x etc so this was someone who was really good 6 years ago gets some credit but not enough to overcome his performance of recent years
i also started off only going back 5 years into historical data, do 7 years now cause i already have it so why not
just realized how low lowry and westwood are on that list lol, man i got some major leaks to plug - but overall things looking good thus far as i'm super exposed to matt jones and russell henley due to those projections so it's a nice return to pga dfs even if it's of the broken clock is correct twice a day variety
hoping to get er done via python 100% by end of year, waiting 3 minutes on excel every time i switch cells is a horrible experience nobody should ever be subjected to