I wrote recently about NVIDIA — that I bought it nine times, never sold a share, and let it become my largest position by a wide margin. The obvious question, which that piece didn't answer, is whether that's reckless.
It's a fair question and it deserves a real answer rather than a shrug. Mine is that I don't have a position-size limit, I've never trimmed anything for being too big, and I think both of those are defensible — but only because of some other things I do that are less visible.
How I actually buy
I decide the total amount I'm willing to put into a company before I buy any of it. Then I buy it in increments, usually in thirds, over weeks to months. A planned dollar-cost average into a position I've already sized.
That's the whole method, and the discipline is entirely in the first step. Deciding the total in advance means the buying decision and the sizing decision happen at different moments — sizing when I'm calm and looking at the business, buying when the price is in front of me. Collapsing them into one act is how a position ends up larger than you ever intended, one convincing afternoon at a time.
If the price stays attractive as I work through the thirds, I'll scale further with conviction. That's a deliberate departure from the plan rather than a drift away from it, and I want the difference to be clear: I add because the case got better, not because the position got exciting.
I don't have a maximum
At its peak, NVIDIA was about 25% of my portfolio. I didn't trim it.
That was a real exposure and I want to be plain about the arithmetic rather than wave it away: at a quarter of everything, that one company halving would have cost me 12.5% of the whole portfolio in a single move. That's the bet I was carrying. It happened to work. It has not, so far, cost me anything — but "so far" is doing real work in that sentence, and anyone reading this should weigh a live position at 25% differently from one that has already been vindicated.
What I'd say in its defense is that my thesis was still evolving and getting stronger, not weaker. The reason to trim a position is that the case for owning it has deteriorated. Mine was improving quarter after quarter. Selling into that because a percentage had crossed some line I'd invented would have been letting arithmetic overrule the business.
What brought it down wasn't selling
Here's the part I think gets missed in most writing about position sizing, which tends to assume trimming is the only lever available.
My expectation was that the percentage would fall on its own as I kept contributing and other positions grew. That is exactly what happened. A position's weight comes down through addition, not only through subtraction — new money going into other names, and other companies doing well, both dilute a large holding without a single share being sold.
That only works if you're actually still contributing and still buying other things. If a position has grown large because you stopped adding anywhere else, the dilution never arrives and you really are stuck choosing between trimming and living with the concentration. But for someone still building, time and contributions do quietly what a trim does bluntly, and without the tax bill or the risk of being wrong about the sell.
Casting many nets
The thing that makes the no-limit answer defensible isn't anything I do about the big position. It's the shape of everything underneath it.
I hold 36 individual stocks, and about 20% of the portfolio sits in index funds that just move with the market. So the average individual position is a little over 2% of the total. NVIDIA at its 25% peak wasn't a concentrated portfolio with one enormous name in it — it was a wide portfolio in which one name had run about eleven times further than the average.
There's a school of thought that treats owning this many companies as over-diversification, a guarantee of mediocrity. I don't buy it. I'd rather cast many nets and let them run. Spreading entries wide and then refusing to interfere is what lets the outliers actually become outliers — you cannot get an eleven-bagger's worth of weight from a position you keep cutting back to size.
So the sizing philosophy is really two rules pulling in opposite directions on purpose: diversify at entry, and concentrate by outcome. I don't choose my large positions. The companies earn their way there, and my only job is to not get in the way.
The index allocation matters more to this than it looks. It's twenty percent of the portfolio that is definitionally never going to be a stock-picking mistake, and it takes some of the risk out of everything else I've just described. A stock-picking site that didn't mention it would be presenting a more confident picture than the one I actually own.
What I do trim for
Never size. But I do sell.
I trim to raise capital when I want to start a new position and don't have cash for it, which is an allocation decision rather than a judgment on the company being sold. And I trim when a valuation has gotten frothy — when the price has run well past what the business supports, which is a judgment on the company.
Both of those are about something changing. Size isn't a change; it's a consequence.
Why the published portfolio doesn't work this way
The model portfolio on this site has rules I don't apply to myself, and the inconsistency is deliberate.
A published portfolio needs rules precisely so my judgment stays out of it. The point of showing it is to let the algorithm speak for itself — to show what the GARP philosophy actually does, unassisted. The moment I start making conviction calls inside it, it stops being evidence about the method and becomes evidence about me, and a reader has no way to audit the second thing.
There's also a plainer reason. I can assess my own risk tolerance, my time horizon, and how much I have coming in every month. I can't assess a reader's. Rules that don't require any of that information are the only honest thing to publish.
The change I asked for before the October rebalance
The one place I did intervene was the entry rule, and it was about mechanics rather than conviction.
At a rebalance the model was dropping large performers and putting the freed capital heavily into whichever new name came in. That never sat right with me. The screen picks thirty strong names at a moment in time — it isn't claiming the thirtieth is a better idea than the twenty-ninth, and it certainly isn't claiming a brand-new entrant deserves several times the weight of a name that has been compounding for a year.
So a new entry is now capped at the equal weight and the excess is spread across everything already held. That's not a conviction call. It's a correction to a rule that was making an implicit claim the model never intended to make.
If you're just starting
Someone with a small account asked me recently how much to put in a first position, and the honest answer is: don't overthink it. Just starting is the most important step. As you invest more you'll fine-tune your own process and sharpen your own judgment, and the sizing rules you eventually settle on will fit you far better than anything you could adopt from me today.
The sizing question feels enormous before you own anything and turns out to be one of the smaller decisions once you do.
The obligatory caveat, which I mean sincerely
I'm not a licensed financial advisor and none of this is a recommendation. Position sizing is genuinely personal — it depends on your horizon, your income, your tolerance for watching a number fall, and none of those are things I can see from here. What I've described works for someone still contributing regularly, holding a wide book, with a meaningful allocation sitting in index funds underneath it. Take any part of that away and the same approach is a different, riskier proposition. A concentration that hasn't hurt me yet is not evidence that concentration doesn't hurt.