Learn to Build. Learn to Sell. My Journey to building Mr. World Pool.

Naval Ravikant wrote a famous thread in 2018 called "How to Get Rich (without getting lucky)." It sounds like clickbait but turns out to be the most concentrated career advice ever published. I've come back to it more times than I can count. This Easter, sitting at my desk after nearly a year of the most intense work of my life, I realised that I've been living it — without quite knowing it — for twenty years.

I spent the last years learning to build. Now comes the hard part.

But first some back story.

I was 17 when I first set foot on a racetrack. It was in Mijas, near Marbella — our first trip abroad without parents. We didn't understand the language. We didn't understand the sport. We bet on a horse called Mister Hans because it was my father's name. It lost, obviously. I remember walking out of there thinking I never wanted to be the stupidest person in the room again.

The following summer, I went to Øvrevoll with friends for the Norwegian Derby. This time we actually studied the form. Circled jockeys we liked. Made a small score. It felt enormous.

That was the spark. Within a year I was skipping school to watch American racing at 1 AM — Penn National, Charles Town, Delta Downs, tiny tracks streaming in low resolution via VPN. I devoured everything I could find: Beyer's speed figures, Ragozin's philosophy, the Thoro-Graph message board where the real practitioners argued about methodology for pages and pages. I must have handicapped tens of thousands of races that year alone.

At 19, I started making my own speed figures for Scandinavian racing. No database. No tools. Just spreadsheets, notepads, a calculator, and an obsession. The year between school and university I worked as a postman, and I spent that year well — solving rating methodology problems in my head between deliveries, then building ratings at night.

My brother got involved later, watching race replays and recording ground loss data through the turns, and he became an important part of the racing analytics work that grew out of those early experiments over the following years.

That's almost true. There were thousands of hours of genuinely boring work — manual data entry, reading off racecards, often more than twenty hours a week of pure repetition. But even the boring parts had a strange magic to them: the repetitive work paired perfectly with podcasts and audiobooks on every topic that caught my interest, and there have been many. I've spent entire seasons punching in numbers while absorbing Kahneman on cognitive bias, Taleb on randomness, Camus on the absurd, Knausgård on everything and nothing, Dostojevskij, documentaries, talk shows, science shows, youtube lectures, the list goes on. The racing education and the broader intellectual education ran in parallel, and they fed each other in ways I didn't fully appreciate until much later.

The path to where I am now was anything but straight.

Many years ago, I put together a small team — a data scientist, a data engineer, a mathematician — to try to do what Bill Benter had done: build a quantitative odds model for Hong Kong racing. We never really got started. The data infrastructure required was so far beyond what we could pull off that the project felt hopeless before it began.

Before that, I'd tried to build a Scandinavian analytics platform with my my brother and my brother-in-law — "Full Galopp" we called it, a predecessor to what I have built now. We got to an MVP, and actually, that MVP has been alive ever since — I've been its only user for years, quietly maintaining it, knowing that it contained the seed of something much bigger. Mr. World Pool is everything I dreamed that little MVP could become one day.

I started "Galoppanalyse" with my brother. We created genuine buzz and enthusiasm for a sport that is, frankly, struggling in Norway. The industry didn't engage. That was disappointing, but it taught me something.

Ed Catmull, the man who co-founded Pixar, wrote something that has stayed with me:

"The world is often unkind to new talent, new creations. The new needs friends."

The racing industry should think about this. Especially in the smaller markets where recruitment is poor and the sport is shrinking. You cannot complain about decline while ignoring the people who show up with new ideas and genuine passion.

But of course it only led me on to greener fields, as I naturally reached the conclusion that it would be better use of my time to get access to the really thriving markets, rather than trying to save a dying one.

While all of this was happening in the background, I built a completely different career.

I studied organisational psychology at NTNU, wrote my master's thesis on creativity and high-performing teams, and spent 6.5 years in an IT company in Lillehammer. I designed apprenticeship programmes, built retraining initiatives, served as product manager for a development team. I was good at it. The trajectory was comfortable.

But the horses never left me. They were always there — the first thing I checked in the morning, the last thing I read at night. The parallel life that nobody at work, or at home, or at family dinners, quite understood.

In 2019, I started an AI learning group at my company. Most people thought it was a curiosity. I thought it was something else entirely.

During COVID, I spent countless hours teaching myself the basics — data science, Python, SQL. Just enough to understand what was possible. Then came the no-code era, and I built websites, data tables, basic tools. Nothing spectacular, but each step expanded what I could imagine doing alone.

Then the AI models started arriving, one after another. DALL-E was the first one that made me pay real attention. Then GPT-3. Then GPT-4 changed everything about what felt possible. Gemini. Claude. I tried them all, tested every one, spent real time with each. And every year I asked myself the same question: is it good enough now? Can I finally build what I've been wanting to build for twenty years — the ratings, the database, the analytics, the platform — by myself?

Every year the answer was the same: not yet. Close, but not yet.

Until spring 2025, when the answer changed to: it's possible now. Difficult. Extremely messy. But possible. And a year later, it's significantly less messy — though I should be clear: I'm building something that no AI in the world has been trained on, because it simply didn't exist before. This isn't a matter of writing a prompt and getting a result. It's thousands of hours of genuine problem-solving, and it has been the most educational and rewarding experience of my life, regardless of where the road leads from here.

For the first time in twenty years, the gap between what I wanted to build and what I could actually build started to close. And I no longer needed anyone else to close it. With AI, it came down to me — my effort, my standards, my willingness to go all in. I could set the bar wherever I wanted without having to account for the fact that not everyone shares the same energy, the same drive, or the same life situation to go completely, almost absurdly, all in on something like this.

So that's what I'm trying to do.

In August 2025, I quit my job. I took everything I had — twenty years of racing knowledge, the rating system I'd been refining since I was a teenager, the failed attempts, the IT experience, the AI fluency — and I went all in.

I set Easter 2026 as my finish line and started building. Every day. Most days ten hours or more. Deep, focused, uninterrupted work with no time wasted on unnecessary meetings or coordination or communcation or approval gathering or chit-chat, just me and Claude and a clear direction, this is how you gain super-human productivity. It felt good.

The to-do list has grown every single week — out of control, you could almost say. I'm still not entirely done, even with Easter upon us. But at the same time, I've already crossed off things that were marked for 2027. I've done things I couldn't have dreamed of when I started this nearly a year ago.

I started with Scandinavia and the UAE — markets I knew well. From earlier years I'd experimented with Hong Kong and Germany, put down serious hours to see if they were viable, but it's only now that I could go the full distance. Every race, years back, complete datasets. Then I got deep into UK and Irish racing — betting's heartland — and that has come remarkably far. I discovered French racing along the way and fell in love with it; it's not quite production-ready yet, but it won't be long.

I also started on Australia and Japan. Two of the greatest racing nations on earth — I couldn't resist. In hindsight, I probably should have waited. These two countries alone make up well over half my entire database, and the sheer scale has been humbling. They're dream markets for the future, and the ambition is real, but it was a bit overenthusiastic of me to dive in this early. I'll need to pump the brakes and make sure the foundations are solid first. But the data is there, and when I'm ready, so is it. It's just a matter of getting the ratings right for all class levels.

I've been comfortable with computers my whole life. At ten I was downloading pirated games, making ISO-images and burning CDs, finding crack codes, doing all manner of things I probably shouldn't have been doing. And we didn't even have broadband back then! I would occupy the land line for days without anyone noticing. I played Championship Manager and Civilization III until I went square-eyed. I've always been able to make computers do what I needed them to do.

But actual coding — building applications from scratch, writing production software — that was always out of reach, until now.

With AI at the level we're talking about now, everything falls into place. The domain expertise I've accumulated over twenty years, the technical intuition I've always had, the product thinking from my IT career, the early interest in AI, the grit and persistence and theoretical knowledge about Innivation I learned from writing my master's thesis — suddenly all of it is useful at the same time, in the same project. It's extraordinary to be part of.

I'm still redefining. But that combination — domain expert, builder, and primary user — is genuinely unique. I don't think anyone else in the world does exactly this, at this depth, across this many jurisdictions. That's not arrogance. That's specificity.

The product is ready. It's called Mr. World Pool — mrworldpool.com. It's not launched yet, but it's close. And it's better than I dared hope. The product exceeds my own expectations, and the data foundation is far richer than I could have imagined when I started.

Now comes the part I'm not particularly good at. Selling. But I guess I will take the next year to learn this craft properly and then become unstoppable. With my newfound building skills I'm sure there will be other things I can create for this world too.

I don't know anyone in the racing industry. Not really. I've never worked inside it. I have no network, no connections, no warm introductions. I've spent nearly a year building in silence, or you could argue that it has been nearly twenty years building in silence, and now I need to step into the world — present, socialise, network, sell.

But I'll find my own way to go about this too, with authenticity and integrity and honesty, even if it takes a bit longer that way.

So that I can enjoy this part too.

And maybe gain some new friends in the process.

So here I am. Under my own name.

Asking the racing world to take a look.

See also:

How I measure racehorse performance across 11 countries and one million races.

I’m about to launch mrworldpool.com — a premium horse racing analytics platform I’ve built from scratch with plenty of help from AI along the way. This article explains the core of the product: the rating. What it measures, how it works, and the 20-year journey behind it.

Anedvardsen.noAksel Normann W. Edvardsen