Even-Odds is a sports modeling project focused on one simple idea:
Make better decisions by thinking in probabilities — not hot takes.
Most sports content is built around certainty (“lock,” “can’t miss,” “free money”). Real outcomes don’t work like that. Even great edges are noisy, and uncertainty is the whole game. This site is my attempt to treat sports betting more like a data problem, or financial market: estimate outcomes, quantify uncertainty, compare to market prices, and keep track of what happens over time.
What you’ll find here
- Model-based forecasts (spreads, totals, and other angles as I add them)
- Explanations of how the model thinks, written for normal humans
- “Lab Notes” — my ongoing notes, experiments, and lessons learned building the model
- Tools & dashboards (eventually)
If you’re here for guaranteed winners, you’ll be disappointed.
If you’re here because you like strategy, numbers, and figuring out what’s actually priced into a line — welcome.
What “Even-Odds” means
Sports outcomes are messy. The goal isn’t to be right every day — it’s to find spots where the odds imply one reality and the data suggests another.
Think of it like this:
- Books don’t need to be “wrong” for a bet to be good
- You don’t need to be “certain” to have value
- You just need the price to be better than the true probability often enough over time
That’s the core mindset behind everything here.
About the Even-Odds Modeling Approach
This is a probabilistic simulation model designed to generate realistic game outcomes — not just a single “final score guess.”
Instead of predicting one number, the model tries to generate a range of plausible outcomes, then uses that distribution to estimate probabilities like:
- “How often does Team A cover -4.5?”
- “How often does the total land above 228.5?”
- “What does ‘fair value’ look like vs the current market price?”
The approach in plain English
1) Model the game as a sequence of events
At a high level, games are built from repeated possessions/plays. If you can model what tends to happen on each one (and how long it takes), you can simulate full games.
2) Capture team tendencies, matchup effects, and context
Teams aren’t static. Opponents matter. Pace changes. Rotations and schedule spots can matter. The model is built to incorporate those factors without pretending it knows everything with certainty.
3) Treat uncertainty as a feature, not a bug
Rather than hiding uncertainty, the model quantifies it. The output isn’t “this will happen,” it’s “here’s how often this happens under the model.”
4) Evaluate like a grown-up (calibration + tracking)
A model isn’t “good” because it hits a few winners. It’s good if its probabilities match reality over time. That means focusing on things like:
- calibration (are 60% plays winning ~60%?)
- distribution accuracy (are we getting the shape of outcomes right?)
- tracking decisions vs market pricing (not cherry-picked results)
If you want the deeper technical version where we dive into the lower level details, experiments, and findings – that’s what Lab Notes is for.
A quick note on transparency
I’m building this in public because I enjoy it — and because it forces discipline. All results are public, timestamped, and and tracked by a 3rd party site on betstamp.
That said, no model is perfect, and sports markets are competitive. I’m aiming for a balance between:
- being transparent about the approach and learning process
- not turning every post into a 40-page technical paper
Over time, I plan to publish more standardized tracking and methodology notes so it’s clear what’s being measured and how.
Responsible betting
This site is for education and entertainment. It is not financial advice, and nothing here is a guarantee of profit.
Sports betting involves risk. If you choose to bet, do it responsibly:
never bet money you can’t afford to lose
use bankroll management
expect variance (even with an edge)
About me
I’m a data/analytics builder who’s spent a lot of time thinking about how to model complex systems — and I’ve always loved the strategy layer of sports betting.
Even-Odds is where I document that process, share what I learn, and build tools that make the modeling more useful over time.
If you want to follow along, check out Lab Notes and the latest posts.
Questions / ideas / feedback?
Reach out here