Bet the Process - Rufus is Back

Podcast Release Date:
Podcast Link:
hagrinBPR Review Date: 2024-04-06 03:09:44
Rating: 6.75

Notes:
5:30 - Rufus goes over some modeling concepts when it comes to selecting what data you're going to use. Brings up throwing out the obvious / no real decision data. Brings up another good research question I also don't know the answer to which is what exactly is the optimal situation for maximizing points to close out an NCAAM game. Probably a good research topic to publish and either find yourself some edge or get some published work out there. Jeff and Rufus debate if something like this is even "knowable" and it's a good debate (I see both sides). Jeff brings up a good point about how with the DP guys "look for situations where models are generalizing where you have some insight" (paraphrasing). Think that's good advice.

16:00 - How do you deal with incomplete information / datasets? "No dataset is complete and this is a good thing - processing incomplete information is where the edge is." (Sidenote - I always wondered if esports would be the Holy Grail for this, but it hasn't been as of yet). Goes over some of the different types of incomplete information. Rufus gives a bunch of very good real world examples.

27:45 - Rufus brings up this idea of "how cool would it be if you could get the wearable information from athletes?" While pie in the sky stuff, for those blissfully unaware, wearable data has been instrumental in locating military and intelligence personnel (search for "The Strava Heat Map and the End of Secrets" by Wired as an example) so these devices leak all types of data. Again, not actionable, but interesting train of thought.

37:30 - "I care more about the relative differences in weather that players face. It comes less from forecasting the mean and more from forecasting the distribution." I don't bet golf, I care about weather in NFL and soccer, I thought this was the best point of the pod. Makes me think my own approach to weather is primitive. Golf sharps might say this is fairly obvious, but I don't have the domain knowledge there and I found this point pretty insightful.

38:10 - Jordan section starts. Recap - surprising, Duke's approach offensively to not attack Burns on D (100% agreed). 43:00 - Alabama vs UConn variance discussion (this has been a common topic on almost every betting pod). More Bama shot profile talk. Good point about prior event vs drop coverage team w/large man in middle in Purdue game "where they shot nearly 50 3s". (Was 19-46 41.3%). Brings up the concept that Oats has implemented which is "shrink threes" (i'll attach an image, good knowledge by Jordan). 47:45 - Purdue tough time with pin downs & secondary actions, NC State very good at exploiting drop coverage. Thinks Edey & Burns will likely cross matchup. Abandoned the press which is a very good deviation from the norm for Keatts.

There's going to be a subsection of misanthropes that hate everything BTP, but I thought not only was this pretty good, but I thought there was a noticeable uptick in effort and I thought there was a pretty winning formula here - Rufus talks about abstract modeling issues, provides real world examples and Jeff pushes back when appropriate to extract more. Then, bring on a sport specific expert who is well prepared, go PTI style short segments on issues and keep it moving. Just have to figure out how to stop talking over each other.