Read on Twitter
building on this a little further... once we (have the humility to) accept that we aren't clairvoyant... it naturally follows that we need a probabilistic framework to evaluate decisions... 1/n https://twitter.com/JezData/status/1348383072002170883
and although 4th down decisions are some of the most easily identifiable and quantifiable, they still are only the microeconomics of an NFL franchise. building similar frameworks for macroeconomic decisions can have substantial impacts on a franchise for decades... 2/n
if we can setup a "framework" for those macroeconomic decisions we start to tilt the odds in our favor, draft picks are the obvious choice but it can even apply to stuff like... coaching hires? 3/n
imagine a probabilistic framework for all decisions, whether or not to play a receiver in the slot or on the outside. compounding small wins works in all aspects of life, the 49% blackjack player is broke, the 51% blackjack player is (in theory) a millionaire 4/n...
those frameworks are nice and all, then we need tools to execute on those frameworks, don't think of player tracking data and random forest models as standalones for "analytics" and start thinking of them as hammers and nails for the house that is our probabilistic framework 5/n
btw... this stuff is really hard. this is why are data engineers become just as (if not more) in demand than data scientists... https://twitter.com/_rachelbalkovec/status/1347491855341027329?s=20 6/n
so stop asking "are you sure" and start asking...
