this has been a concern of mine despite him being sharp and having good leadership skills.
4th and 5 from the 37 to open the game. Jones looking like he could gash the defense for runs if needed. Offense on a roll from last week. Kicking a 54 yard field goal. That wasn't a good "feel" for the game. You have to know Dallas is going to march it up and down the field on your team. Your only shot is a confident offense with it's foot on the gas for 60 mins. That "feel" for the game BS is garbage. There was no feel. JJ played it too conservatively which is not a rarity for him and we watched another season collapse as a result.
Bad call. Bad analytics. Bad "feel." Even Mike McCarthy has this shit figured out. You see more and more teams converting 4th downs in their OWN territory and we can't even do it in the situations teams have understood for years.
The idea that the Giants "get" numbers when the numbers so often add up against them is ridiculous.
Sadly this case is becoming more and more clear that JJ wasn't brought in to change things but convince us that we were in good hands so Jints central could stay stuck in the past.
I do believe there are other factors that can't be accounted for but this type of stuff is of high value.
Congrats.
Going ot post something similar on your other thread.
Great job.
I do believe there are other factors that can't be accounted for but this type of stuff is of high value.
Congrats.
Going ot post something similar on your other thread.
Great job.
Honestly this isn't even the good part of my model. These are just standard underpinning of applying game theory (expected points per play) to play calling. And you are absolutely right, models doing that effectively would need to play out the conditional probabilities of each and every matchup. Right down to expected trajectories and OL matchups of like the possibility of blitz vs. DE vs. OT one on one.
A real solid expected points calculation would take real data as well as millions or hundreds of millions of simulated plays. For an interesting read on that look at the link below.
Actually the strength of my system is more on resource allocation and using incentive structures and budgeting combined with advanced projection technology to optimize for wins. Along these lines you could actually arbitrage bad seasons. If you'd like to email me I would be happy to send you my full proposal / explanation of how it works. You'd probably find it interesting. George@prospero.ai
Mu Zero by Google DeepMind - ( New Window )