My AI flagged Heijmans 18 months before the market saw it

On June 1 2022 I would have bought Heijmans. Not because I had thought about it, but because my trading system told me to. It had given me a list of fourteen stocks to buy that day, and Heijmans was at the top.

The price was 8.50 euro that day. Today it's 25 euro. A return of nearly three hundred percent in four years.

Sounds like a glory story. But there's a catch. My system didn't even exist back then. I'm running a simulation in hindsight, with the data I have now, to see what the system would have done if I had believed in it. Meanwhile I look forward and ask: was this clever of my AI, or was it just lucky?

A stock nobody wanted

Heijmans is a Dutch construction company. Roads, houses, building projects. Not sexy. In 2018 the company nearly went bankrupt, in 2022 it was still recovering. My system looked at three things: how cheap is the company versus peers, how strong is the price trend, and how healthy are the fundamentals.

On June 1 2022 the system saw this:

• Heijmans traded at an EV/EBITDA of 6. The average construction company sat at 11. Cheap.
• The profit margin was 5 percent. Weak, but rising.
• The price trend was flat. No momentum.

Total: strong buy signal based on valuation. No buy signal based on price or story. That's exactly what the system is built for. To find stocks the market has forgotten.

The market did forget Heijmans, thoroughly. The price went from 8.50 in June 2022 to 8.80 a year later. For fourteen months nothing special happened. The system held on. The fundamental score stayed high. The price stayed quiet.

And then, 18 months later

Early 2024 it started moving. Heijmans had just published their numbers: EBITDA margin from 5% to 9%. An order book that was 33% larger. ROCE, a measure of how efficiently a company uses its capital, had nearly doubled.

The price reacted. 11 euro in January, 15 euro in March, 19 euro in July. Today 25 euro.

My system had said in June 2022: buy this. The fundamental transformation that became visible in 2024 was therefore already underway, just not yet in the numbers available at that time. Heijmans was getting better, but it wasn't in the annual reports yet.

The question is: how did my system know?

Answer: it didn't know. It saw a stock that was technically cheap within the construction sector. That it was also fundamentally improving was not the reason to buy. That was luck.

Or was it?

The three portraits of Heijmans

Let me show you three different views of Heijmans as they look today.

My system says: Heijmans is still worth buying. Fundamental quality is high, the business is running better than ever. But within the construction sector it's no longer extremely cheap. Other names are now more interesting.

Stockopedia says: Heijmans is a "Super Stock." That's their highest classification, given to companies that score high on value, quality, and momentum at the same time. Only a few percent of all listed companies get this label.

Morningstar says: Heijmans trades at a 167% premium to our fair value estimate. Sell.

Three professional quantitative systems looking at the same data. Three different conclusions. Who's right?

The answer is: nobody knows. Not Stockopedia, not Morningstar, not me. What they show is that different methods see different things. And Heijmans is a stock where those methods disagree.

The price of patience

Suppose I had actually put 1,000 euro into Heijmans on June 1 2022 on the system's recommendation. I would have had nothing for fourteen months. No return, no confirmation that the choice was right. A friend who bought a meme stock in the same period would have laughed at me.

In month fifteen I probably would have already sold my shares. That's what most retail investors do. Fourteen months of nothing is enough to lose your faith, even when fundamentals stay strong.

That's why I'm building this system. Not because an algorithm is smarter than a human. But because an algorithm doesn't get nervous after fourteen months without reward. It looks at the numbers fresh every day, and as long as those numbers keep checking out, the position stays.

In Heijmans's case: months 1 to 14 was nothing. Months 15 to 47: nearly three hundred percent return. Whoever had patience won. Whoever didn't, won nothing. The difference was not in the system, it was in the holding on.

Was it edge or luck?

Honest question. I don't know.

Edge would mean: my system has a method to find stocks that statistically outperform the market, and that keeps working in the future. Luck would mean: in this specific period, for these specific stocks, things happened to work out.

The difference can only be established afterwards, and only over years. A statistician would say: you need at least 30 years of data to make any meaningful claim about whether a strategy has alpha or not. I have four.

What I do know: in the same period that the system flagged Heijmans, it also flagged Boskalis Westminster (BAMNB), another Dutch construction company. On June 1 2022 they both stood at the maximum Value score, +3.00. Both flat in price. Both cheap within their sector.

Heijmans has tripled since then. Boskalis has risen, but not as strongly. Two stocks, the same signal, two different outcomes. Was Heijmans a better call? Or did I just win half the lottery?

What I do now

I don't buy Heijmans in new rounds anymore. The system says: there are better opportunities now, other construction companies are now preferred. But the position the system would have hypothetically opened on a day in 2022 stays put. It's up 200 percent and nobody is telling me the music stops.

In a few years I'll know if this story would have ended differently. Maybe Heijmans was at its peak. Maybe there's another transformation underway. Maybe there will be a setback, a lost contract, a political decision that affects construction.

For now I hold on. Not because I know it will turn out fine, but because the system says so and the system has been right so far. That's not a reason for certainty. It's a reason to keep going.

In five years I'll know if this was edge or luck. Until then it's both at once.