Every investor pays tuition. Mine came in three instalments, and I'd pay all of it again for what I got back — the losses were real, but what I learned from them has saved me a great deal more than they cost.
Here's what I got wrong.
Chasing the story
Nikola. A company with an extraordinary narrative about the future of trucking and, at the time I bought it, very little in the way of an actual business underneath it. I lost over 75% of what I put in.
I wasn't ignorant of the situation. I'd have told you at the time that the revenue wasn't there yet. The problem is that I'd talked myself into treating that as the opportunity rather than the risk — get in before the business exists and you capture all the upside on the way to it existing. That reasoning feels sophisticated while you're doing it. It is not.
What I actually owned was a story, and stories reprice fast when the story changes. There's no floor under a company that isn't producing anything, because there's nothing to value it on — no earnings, no cash flow, no book of business. When sentiment turned, there was nothing to catch the fall. A stock that has fallen 75% can fall 75% again; "it can't go much lower" is not a valuation.
What it taught me: a business has to actually be a business. That's the entire Quality pillar in the model I later built — returns on capital, real cash conversion, margins that exist. Not "will this be great someday," but "is this producing anything right now."
The value trap, and the falling knife
AT&T. This one was more expensive in a quieter way, because I got it wrong twice.
I bought it because it looked cheap. Low multiple, enormous dividend, a household name — every surface signal said undervalued. Then they cut the dividend. And rather than treat that as the business telling me something, I averaged down. I tried to catch the falling knife, and it went exactly the way that phrase implies.
The thing I hadn't separated is the difference between cheap and undervalued. A low multiple means the market has priced the business low. That is sometimes an opportunity and sometimes an accurate read on a company whose earnings are going to keep shrinking. Nothing about the multiple itself tells you which one you're holding — you have to look at the business, and I was looking at the price tag.
A large dividend on a struggling business isn't income, either. It's a payout the company can't indefinitely afford, and the market usually works that out before you do. The yield looks most attractive right before it's cut, because the yield is high precisely because the price already fell.
Averaging down made it worse in a specific way. Adding to a loser feels like conviction and is often just refusing to re-underwrite the position. The honest question after a dividend cut isn't "how do I lower my cost basis" — it's "knowing what I now know, would I open this position today at this price?" I never asked it. If I had, the answer was no.
What it taught me: valuation only means something in the context of the business behind it. It's why the model grades Value against sector peers rather than in the abstract, and why a low multiple alone never produces a high score.
Buying income 25 years too early
This is the one I'd most want a younger investor to hear, because nothing about it felt like a mistake at the time.
I went through a long stretch of buying for yield — Coca-Cola, Pfizer, that category of large, stable, generous payer. Nothing was wrong with the companies. They're perfectly good businesses. The mistake was entirely mine: I was a young investor building a portfolio designed for someone about to retire.
And this one doesn't show up as a loss. That's what makes it dangerous. My statements looked fine. The dividends arrived on schedule. What I'd actually done was trade away years of compounding for a modest, predictable cheque I did not need and wasn't spending. The cost was invisible because it was an opportunity cost — the growth I didn't get, in an account with decades of runway to let it work.
Losing money at least announces itself. Quietly under-earning for years doesn't.
What it taught me: growth is a requirement, not a bonus. It's why the model treats it as a lead signal rather than a tiebreaker, and why a stable business that isn't getting bigger can't score well here no matter how safe it looks.
The pattern I only saw later
Three mistakes, and each taught me exactly one thing. Nikola taught me quality. AT&T taught me that cheap and undervalued are different words. The dividend years taught me growth.
Those are the three pillars this site's score is built on.
I'd absorbed all three as ideas long before, from years of reading and listening to people who explain investing well. But there's a difference between knowing a principle and having paid for it. The screen I eventually built is, honestly, a system for stopping me from repeating these three specific errors: it won't rate a company with no real business, it won't call something cheap without asking what the business is doing, and it won't reward a company that isn't growing.
That's why the score blends all three rather than optimising for any one of them. Every one of my mistakes came from looking at a single thing in isolation — the story, the multiple, the yield — and ignoring what the other two would have told me.
Would I take it back?
No, and I mean that literally rather than as consolation. The knowledge those losses bought has saved me multiples of what they cost, and it has kept me out of positions that would have cost far more. I'd gladly pay it again.
What I'd change is the size. The lesson costs exactly the same whether you're wrong with 2% of your account or 20% — the tuition is fixed, and you get to choose how much you pay for it.
Nothing here is advice, and none of these companies is being recommended for or against today. They're what I owned while learning. The rules the model runs on are set out in full on the about page.