I am currently in my second year at UCL studying Mathematics and Statistical Science. Statistics, and especially probability, has always fascinated me. In my eyes, understanding probability almost feels like gaining the ability to see slightly into the future. The moment it really connected for me was during a lecture where the professor explained how probabilistic models can be used to predict outcomes far more accurately than intuition alone. That was when I realised how powerful mathematical expectation actually is.

Around the same time, I had been doing arbitrage betting as a small side project. After speaking with a few friends who study software engineering, we decided to combine our interests and start building something for fun. At the beginning, the software could only scrape very basic arbitrage opportunities. As we kept working on it, we added more ideas and more structure, and eventually the first proper version of the bot was created. It could now find both arbitrage opportunities and true value selections.

The early version was far from perfect. It made mistakes and missed profitable spots. But over the next two months we kept improving it, testing everything, adjusting the logic and slowly increasing its accuracy. Eventually it reached a point where it became reliably profitable.

This was when DBArbitrage was created. In the first month more than 80 members joined and made profit using the system. By the end of the season we had grown to more than 140 members. For us, that success came with a cost as some of our own betting accounts were restricted because the bookmakers clearly did not appreciate our winning record and arbitrage approach.

The summer brought uncertainty. Many under market providers stopped offering the lines we relied on, which meant our entire approach had to change. We decided to rebuild the bot completely, from scratch. The goal was to capture every arbitrage opportunity possible while also improving the model for positive value bets. At the same time we developed our own formula for calculating true value on each event, which made identifying profitable opportunities even more effective.

Now, after months of rebuilding and refining every part of the system, the new version is finally complete. We are ready to relaunch, and this time the software is far more accurate, efficient and powerful than anything we have built before.