The honest answer to "when do you sell?" is: rarely. Most of what looks like a sell decision in my account is a trim, and the distinction matters more than it sounds.
There are two things that make me exit a position outright, and one much more common thing that makes me reduce one. They're different decisions with different logic, and running them together is how people end up selling good businesses for bad reasons.
The two things that make me sell
The thesis breaks. Something changes about the company that invalidates the reason I bought it. Not the stock price — the business. A company acquires its way into a market I never wanted exposure to, or loses the cost advantage that made the margins work, or the thing I thought was a durable moat turns out to have been a temporary lead. If the reason I own it is no longer true, the position goes, regardless of what the price has done since.
This is the cleanest sell signal there is, and also the easiest to rationalize away. It's uncomfortable to admit a thesis was wrong, and there's always a story available for why this quarter was an exception.
Repeated weakness in the fundamentals. Note the word repeated. I want to see at least two or three quarters of deterioration before I exit — margins compressing, cash conversion softening, growth decelerating in a way that isn't reversing. A trend, not a data point.
That threshold exists for a specific reason: it stops me getting scared out of positions easily. One bad quarter is noise. Every good business has them, and selling on a single disappointing print is precisely how you get shaken out of the compounders right before they recover. A genuine deterioration and a temporary rough patch look identical for the first quarter or two, so I'd rather be a quarter late than wrong in the expensive direction.
What I actually do most of the time
Neither of those comes up often. What comes up regularly is a position becoming, in my judgment, overvalued — the business is fine, the thesis holds, but the price has run ahead of what the fundamentals justify.
I don't sell those. I trim.
I take some off the top, lock in part of the gain, and let the rest run. The business hasn't changed, so there's no reason to leave. But the position has grown into a larger share of my portfolio than I'm comfortable carrying at that valuation, and reducing it puts the risk back where I chose to have it.
How I judge "overvalued"
This is where I want to be careful, because it's the least mechanical part of what I do.
I use the Value pillar as part of the decision. It's the same signal the model uses — valuation graded against sector peers rather than in the abstract — and it's genuinely useful for flagging when something has gotten ahead of itself. But the final call is my own judgment. There's no number at which I trim robotically.
The reason is that valuation tolerance shouldn't be uniform. A business with a long runway of high growth ahead of it deserves more slack on valuation than a mature one, because the growth will grow into the multiple. A company compounding at 30% and a company compounding at 6% are not equally expensive at the same P/E, and treating them as though they are is a good way to sell your best holdings early.
That's not a personal quirk — it's the same logic the PEG ratio formalizes, and it's why PEG is one of this model's lead signals rather than a raw multiple. My judgment on when a name has gotten expensive is essentially the same idea applied by hand, with the runway weighted by what I actually believe about the business.
What this looks like in practice
The names where I do this most are the large, heavily-covered ones — Microsoft, Amazon, Apple, Nvidia. I've trimmed and added to all of them multiple times, at moments of perceived overvaluation and perceived value respectively.
That's the part people miss when they hear "trim." It isn't a one-way exit ramp. It's managing a position in both directions around a business I want to own for a long time. Take some off when the price runs ahead of the fundamentals; put some back when it lags them. The core position stays.
These names suit that approach for practical reasons. They're less volatile than smaller companies, so the moves are usually valuation drifting rather than the business lurching. And they're monitored so heavily that information reaches the price fast — I'm not going to find an edge in knowing something about Apple that the market doesn't. What I can do is have a view on whether the current price is demanding more than the business is likely to deliver, and act on it in increments.
Why trim instead of exiting
Two reasons, and the first is just arithmetic.
A smaller position is a smaller share of the portfolio, so whatever happens to that stock matters less to my overall result. That's not a psychological trick — it's position sizing. If a name has doubled and now represents twice the share of my account it used to, I'm carrying more concentrated risk than I originally chose to, whether or not I decided to. Trimming puts it back.
The second is that selling winners entirely is expensive in a way that never shows up on a statement. The whole premise of buying quality businesses with long runways is that the good ones compound for years longer than seems reasonable. If I exit every time one gets expensive, I systematically remove exactly the positions that would have produced the returns. Taking some profit and holding the rest lets me stay in the story while sleeping at night.
One thing I'm deliberate about, though: the shares I keep after a trim are not house money. It's tempting to treat a position you've already taken profit on as somehow free, and that's a genuine trap — the remaining shares carry exactly the same risk per share they did before, and they're competing for space in my portfolio against everything else I could own instead. A trimmed position still has to earn its place on the same terms as a new one. Taking profit reduces how much of my account is riding on a single name. It doesn't lower the bar that name has to clear.
What the model does and doesn't tell you
Worth being explicit, since it's the tool this whole site is built on: the Bull Rankings score is a buy screen. It has no sell signal.
It ranks companies on quality, growth and valuation as they stand today. A falling score tells you something — usually that fundamentals softened or the price ran up — but that's information, not an instruction. The model doesn't know what you paid, how large the position is relative to everything else you own, or what your tax situation looks like. Those are the inputs that actually decide a sell, and none of them are in the model.
So I use the score the way I'd use a research note: as a prompt to go look, not a trigger to act.
The obligatory caveat, which I mean sincerely
This is what I do in my own account. It isn't advice, and it isn't a system to copy without thinking hard about whether it fits your situation. My time horizon, tax position, and tolerance for watching a position halve are mine and probably not yours.
What I would actually suggest taking from it: decide your rules while you're calm, write them down, and keep "the business changed" strictly separate from "the price moved." Those two feel identical in the moment and they are not remotely the same decision.
Disclosure: I hold positions in the companies named above. The rules this model runs on are set out in full on the about page.