We Built a Sports Betting Model! (With Help from Elihu Feustel + Python + Pure Stubbornness)

On this week’s episode of Gambling with Good JuJu, we’re proud (and a little stunned) to announce something big: we’ve officially built our first sports betting model. Inspired by Elihu Feustel’s brilliant book Beyond the Odds, and powered by a healthy mix of code, trial-and-error, and straight-up perseverance, our answer key is finally operational—at least in theory.

If you’ve ever wondered what it actually takes to go from reading a sharp betting book to building a functioning predictive model for college football, you’re going to love this episode.

🧠 How It All Started: From Elihu’s Book to an Idea That Wouldn’t Die

Last September, we interviewed professional bettor and author Elihu Feustel, whose book Beyond the Odds lays out a framework for creating your own betting model using historical data and a concept called the “Answer Key.” That conversation lit a fire under us—but we didn’t realize at the time just how much work it would take to bring it to life.

We had no experience building a model, but we had passion—and we had Uncle Justin, our show's superfan and resident Python wizard. Once we all read the book, we decided to tackle this thing as a team. And then the real work began.

🏗️ Building the Foundation: Scraping Data, Writing Code, and Learning on the Fly

The first challenge? Historical data. We needed over a decade of college football box scores and betting lines—combined into one clean, usable file. After striking out on finding an existing solution, we decided to build our own data pipeline. Juice took this on personally, spending months learning to pull, process, and store the data from multiple APIs. Think ETL pipelines, only with more yelling.

Meanwhile, Justin got to work translating Elihu’s “Answer Key” logic into Python. That meant:

  • Creating a universal game data format across multiple sources

  • Implementing a “distance factor” to rank how similar historical games are to a given betting line

  • Generating samples of similar games and adjusting for balance between spread and total

  • Calculating what the first-half spread and total should be based on historical outcomes

Once we got the pieces in place, we pivoted from college basketball to college football, and suddenly… we had a functioning model.

🎙️ Bringing Elihu Back On the Show: “Did We Do This Right?”

With the prototype built, we brought Elihu back on the pod to walk through our process and ask the big questions:

  • Is our distance formula (currently weighted 1.85 to total) correct?

  • How important is it to have consistent sharp closing lines (e.g., Pinnacle, Circa, Bookmaker)?

  • Should old games from 2015 carry as much weight as recent ones from 2024?

  • Does home/road status change the way games should be grouped in a sample?

  • How do we determine a go/no-go threshold? How far off from the market line is enough?

Elihu patiently answered everything—and made us realize that while we’ve come a long way, we still have plenty of refinement to do. But the core idea? It’s alive and well.

🏈 The Future of the Answer Key: From Model to Betting Edge

We’re now entering the testing phase—running the model against real college football lines for the upcoming season. We’re not yet pulling live market lines, but that’s next. We also still need to backtest first-half lines… and for that, we need better historical data. (If you know a good source, hit us up!)

Our goal? To eventually originate lines with confidence, identify edges, and share what we learn with our audience.

This episode is a deep dive into what it takes to build your own predictive model in sports betting—and a transparent look at our journey from fans to DIY modelers.

🔗 Resources Mentioned in the Episode

🎧 Listen Now

Whether you're an aspiring model builder, a data nerd, or just someone who enjoys a good gambling story, this episode is a must-listen.

👉 Click here to listen to the full episode
👉 Watch it on YouTube

🧢 Support the Show

Want to rep the show? We’ve got brand new Parlay Pete shirts and Good JuJu hats available now in our merch shop. Grab both before Juice decides to take them down without warning.

🛒 Shop now at goodjujubets.net

🎲 Final Thought

This episode isn’t just about sports betting—it’s about learning, building, and chasing edge through trial, teamwork, and tech. We may not be experts yet, but we’re getting closer. One line, one script, and one bet at a time.

Good JuJu to your bets, and we’ll talk again soon.

Next
Next

Buffalo Ascension, Blackjack Lessons & Casino Shenanigans – Triple Trip Report Gambling Podcast Madness