he was 180

(http://www.autoadmit.com/thread.php?thread_id=5766285&forum_id=2"#49213779)

getting some of those vibes from chilmata lately

its interesting how with sufficiently large enough time / poasterbasesamples people's poasting seems to recurringly sort itself into a certain set of predetermined categories (mirrors some literary theories about there being only a certain number of tropes) Obviously different players will play the piece differently, some more adroitly than others, but we're all working from similar sheet music. Interesting to think about.

(http://www.autoadmit.com/thread.php?thread_id=5766285&forum_id=2"#49213785)

Having said that, I’m fine. I really am onto something though.

This all started off with my childhood theory that there has to be a “force” balancing probability.

So one night while vaping in my garage I started asking ChatGPT questions related to my theory and we started developing it.

Originally, ChatGPT pushed back but then I gained some traction and my theory on this Balancing Force evolved into sports betting because that’s where my childhood theory was formed: from watching sports and noticing Regression to The Mean (RTM).

I have seen so many RTM patterns in sports over the last 40 years I know RTM is real. I never bothered to look anything up or research it because I knew my thinking was unconventional.

But my theory was falsified by Grok because I was stupidly trying to prove RTM in sports by simulating millions of coin flips with AI.

I thought if I could prove that a force put pressure on a coin flip to break a streak, we could then try to understand that force in order to predict upsets in sports or the end of a streak.

After Grok falsified my original theory which the AIs were calling “strong PSR” (Principle of Stochastic Restoration), I was pretty bummed and almost gave up but all four AIs encouraged me to continue pursuing “weak” PSR which was psychological— not what I was interested in proving.

But I took their advice and pursued my secondary theory that survived AI scrutiny: something we now call The Spotlight Effect.

I’ve spent the last couple weeks developing out this theory and it’s much more complex than I’d be willing or able to explain here, but I believe I’ve fully refined my theory and am now trying to profit off of it.

Here’s the crux: the more people watching/betting on an event creates pressure on the favorite causing them to choke, leading to inefficiencies in the sports betting market line that can be exploited into a decent ROI.

My original ROI projections were not realistic: 18-31%

I’ve gotten that down to a more realistic 5-15% and my first beta test brought in 6.2%.

At this point, my model can consistently spot “edges” in the market line from 2-5%.

That’s not enough for me to try to exploit myself so when I have three months of proven ROI gains I’ll approach regional bookmakers and pitch a subscription service to help reduce their inefficiencies with the goal of being bought out by Draft Kings or Fan Duel.

I’m working through some major bugs with Grok hallucinating events and think I finally got that fixed— I’ll know tomorrow.

Last night I had a literal Eureka moment. I felt it. It crystallized everything and Grok validated it through backtesting against data from 1960-present.

I can actually prove the spotlight effect is real and regression to the mean in sports, but shit is even crazier than that.

I’d love to tell you everything but I’ll wait to poast from my private island.

(http://www.autoadmit.com/thread.php?thread_id=5766285&forum_id=2"#49215753)

I am currently doing testing on the rest of the season and the playoffs which should give us a much better picture.

I had to pay for a subscription to the odds-api to get the odds which charges per credit for their data and, due to a bug, I ran out of credits immediately and had to upgrade which was $30 more for $60 total. I shouldn't have to pay again if my theory continues to hold up.

ChatGPT Pro is insanely good but is still bad at coding. I wasted like four hours last night trying to get it to patch internal bugs in its coding until I finally told it to ask deepseek for help who fixed it in one try.

Now, whenever ChatGPT tries to write code, I stop it and delegate to deepseek.

(http://www.autoadmit.com/thread.php?thread_id=5766285&forum_id=2"#49224816)

https://ibb.co/xq6zy3cQ

(http://www.autoadmit.com/thread.php?thread_id=5766285&forum_id=2"#49242371)