About Oddslab
From Long-Slip Accas to Monte Carlo Simulations
Like most football fans, my relationship with the bookies started on a Saturday morning, standing at a counter with a pen, hunting for the weekend accumulator. I’d build slips as long as my arm, convinced this was the week I’d crack the code. Unsurprisingly—though it was a mystery to me at the time—those lottery tickets almost never landed.
But I always had a natural fascination with numbers and mental arithmetic. Probability wasn’t just a school subject; it was how my brain processed the world. That fascination eventually shaped my entire career. I started out working behind the counter at a high street bookmaker at twenty, before transitioning into the world of media, marketing, and predictive analytics. For the last decade and a half, I’ve worked behind the scenes with some of the biggest sportsbooks and betting brands in the UK, developing data strategies and marketing campaigns.
I got a front-row seat to how the machine works. And I noticed two massive things:
- The markets are inherently stacked in the bookies’ favour, especially when you feed their profit margins with massive, narrative-driven accumulators.
- Bookmakers don’t just price matches on pure probability. They have to account for public narrative, team hype, and exactly where the massive weight of casual money is moving.
Around the time the football data revolution exploded—bringing xG and a myriad advanced performance metrics into the mainstream—a lightbulb went on. If the bookies have to adjust their odds to manage their financial risk against public hype, then a cold, unemotional mathematical model could find the spots where the bookies actually have it wrong.
The Birth of the Engine
That realization sent me down a multi-year statistical rabbit hole. I immersed myself in the foundational work of sports analytics pioneers: the Dixon-Coles framework for adjusting team strength, Nate Silver’s probabilistic forecasting, and Joseph Buchdahl’s rigorous testing of Expected Value (EV) and closing line value.
The result of that obsessive iterating was the first version of the xWin Engine.
Irony is a funny thing, though. After building a sophisticated, mathematically sound predictive model designed to isolate single edges, what did I do with it? I used it to pick my weekend accumulators. A few landed, giving me some decent beer money, but that elusive lottery jackpot still didn’t arrive. The model was working—more than 50% of my EV-flagged single picks were hitting—but the math of the accumulator meant a random, high-variance leg would always come along to ruin the slip.
Treating Data Like a Fund
A few seasons ago, I finally stopped fighting the math and listened to the engine. I stripped away the emotional desire for the “quick win” and started treating the model like a disciplined investment fund.
I implemented strict, systematic staking rules, set rigid EV entry criteria, and committed to letting the edge play out over hundreds of matches rather than expecting miracles on a Saturday afternoon. The model kept evolving, the variance smoothed out, and the beer money became consistent.
But truth be told, as the seasons rolled on, I realized the real enjoyment wasn’t even the payout—it was the process. I loved the coding, the modeling, the late-night data cleaning, and the constant iteration of the simulation.
Why Oddslab Exists
I still run my personal fund, and it does exactly what it’s supposed to do. But keeping a highly accurate predictive engine locked away on a local laptop felt like a waste of good data.
The bookies have had it their own way for far too long. They feast on bad data, public hype, and emotional betting.
Oddslab is my way of opening up the lab doors. I’m stripping out the noise, the talking-head punditry, and the media bias, and replacing it with pure, simulated probability. Whether you’re looking for an isolated edge for a single play or just want to see how a 10,000-run Monte Carlo simulation projects the final league table, this data is now yours to use.
Let’s level the playing field.