I tried everything. Here's what actually worked.

I lost money for about two years before anything clicked.

Not catastrophically — I wasn't remortgaging the house or anything — but steadily, grindingly, in the way that suggests you might actually be the problem rather than bad luck. Which, it turns out, I was.

Here's what I tried.

TradingView strategies

I found a script called something like "Supercharged RSI God Mode v3" and backtested it on a single stock chart. It showed a 340% return. I deployed it with real money. It lost 18% in six weeks. The strategy was overfit to one chart and had never seen an actual market condition in its life.

You can make any strategy look profitable if you tweak the parameters enough. That's not backtesting. That's fraud you're committing against yourself.

Stock picking

I read all the usual stuff. Peter Lynch. One Up on Wall Street. I started finding "hidden gems" — small-cap companies with products I believed in. The problem is that a product you believe in and a stock that goes up are two entirely different things. Companies with genuinely brilliant products can have terrible balance sheets, predatory management, or simply be in sectors that nobody is buying right now.

I was right about the companies. Wrong about the stocks. Wrong about the timing. Wrong about basically everything except the product.

Earnings plays

Buy before earnings, sell after the pop. This is such a well-known strategy that any edge has been arbitraged away by institutions running the exact same trade at ten thousand times the scale. What I actually experienced was buying into the elevated pre-earnings implied volatility, watching the stock do exactly what I predicted, and still losing money because the options had priced in so much expectation that a merely-good result wasn't good enough.

The YouTube guys

You know the ones. The thumbnails. The Lamborghinis. The "I turned £1,000 into £300,000 trading options" captions with extremely misleading charts. I'm not going to name them because they've got lawyers and I've got a blog, but I will say: if someone is making real money trading, they're probably not spending eight hours a day filming themselves in a rented supercar.

I wasted months on Discord servers where everyone was calling the same trades and congratulating each other and nobody talked about the three-quarters of the time when everything went wrong.

The moment things changed

I found a paper. I think I was down a Wikipedia rabbit hole about quantitative finance at about 11pm on a Tuesday, which is not a sentence that describes someone who's doing great.

The paper was about equity pairs trading. The idea was simple: some stocks move together. Not because they're in the same index, but because they're fundamentally linked — same customers, same suppliers, same macro forces. When one gets ahead of the other, the gap tends to close.

Coke and Pepsi. Visa and Mastercard. Apple and Microsoft. The gap between them is the trade.

Not directional. Not "I think tech goes up." Just: these two stocks have drifted apart by more than they usually do, and they'll probably come back together.

I spent three months building a scanner. It watches 1,180 pairs every single day. It calculates how stretched the gap is — a measure called the z-score, which I'll explain in another post — and flags the ones that have moved more than two standard deviations from their historical mean.

The trades aren't exciting. They're not "I called the bottom on Nvidia." They're quiet, hedged, market-neutral positions that close at a small profit when the pair reverts. Most of them do. Not all of them. But enough.

The part that humbled me

I'd built this whole thing and was quietly pleased with myself when I read another paper and discovered that hedge funds have been doing exactly this since the 1980s. Literally. Morgan Stanley in 1987. Nunzio Tartaglia's quantitative trading group.

I'd independently reinvented a forty-year-old institutional strategy. The main difference was that I was doing it with £15,000 and a laptop, and they were doing it with billions.

Turns out, that difference actually matters — in my favour. But that's a different post.

Now I share every signal the scanner finds. Because there's no reason not to, and because I spent two years searching for something that worked, and I'd have appreciated it if someone had just told me.