The start
For as long as I can remember, I've been obsessed with figuring out how to win. It started with chess, then poker, and eventually sports betting. I've always loved the challenge of finding an edge, especially when it feels like everyone else has already figured something out.
I've also been a massive sports fan my entire life, so getting into sports betting was pretty natural. During college at Georgia Tech, I started out on platforms like PrizePicks and Underdog, since traditional sportsbooks weren't legal in Georgia at the time. I got really into finding +EV opportunities, comparing lines across books, and figuring out how to take advantage of correlations that weren't being priced correctly.
For a while, that was pretty much all I did. But as I started running into limits and the opportunities became harder to find, I got more interested in the markets themselves. Platforms like NoVig, Kalshi, and Polymarket opened up a completely different side of things for me. Instead of just looking for a bad line, I wanted to understand how that line was being made in the first place.
That's what pushed me toward building my own models.
Building models
My first real project was modeling MLB pitcher strikeouts. I had no idea how much work I was getting myself into. There were countless late nights spent digging through data, rewriting code, questioning my assumptions, and trying to figure out why something that looked great on paper completely fell apart in testing.
I'd think I had something figured out, only to find another problem the next day. But I kept at it, and eventually, after more sleepless nights than I'd like to admit, I felt like I'd finally made a breakthrough. For the first time, I had built something from scratch that I felt could actually compete with the market.
That feeling was addictive.
Since then, I've expanded into other markets, particularly NFL player props, where I've spent even more time building simulations, testing different approaches, and trying to understand what drives player performance. The more I work on these models, the more I realize how much there still is to figure out.
Why Projexions
And that's really why I started Projexions.
I wanted somewhere to share what I'm working on, document the research behind my models, and keep a public record of how they actually perform. Not just the winning picks or the models that work, but the mistakes, failed experiments, and everything I learn along the way.
I don't know if I'll ever reach a point where I feel like I've completely figured any of this out. But I think that's what I enjoy most about it. There's always something new to test, something I overlooked, or another edge worth chasing.
— Mani