Conviction Is Not a Plan

Most families at the $5–50M threshold did not arrive there through a balanced portfolio. They arrived through a concentrated position — a business sale, an early stake in a company that grew, a property bought at the right moment, a commodity accumulated with conviction over a decade and held through everything since. The very concentration that built the wealth is now the largest single risk in the portfolio, and the first real test of discipline as a principal is the question of what to do with it.

That question almost never gets answered well. Not because the people facing it lack intelligence or information — they typically have more of both than the average investor — but because the decision sits in the part of the mind where conviction and attachment look identical from the inside, and the longer the position has rewarded its holder, the harder they become to tell apart.

What follows is the discipline I now run on every concentrated position the family office holds, distilled from a decade of navigating this question on my own capital. The illustration is drawn from commodities because that is where I have spent the most time, but the discipline applies to any concentrated holding: founder stakes, employer stock, property, a single large equity bet. The asset varies. The method does not.

Why Conviction Is the Problem, Not the Solution

A decade of conviction earns the bull case the first word in your head every morning. That is not a failing — it is what made the position possible in the first place. The discipline is not to abandon conviction. It is to deny it the right to speak first when you are deciding whether to hold.

Most principals, when asked why they still hold a position, answer with a version of the original investment thesis. I bought it because of A, B and C. A, B and C are still true. Therefore I hold. This is the standard self-report, and it is the wrong question every time.

The right question is which of A, B and C are still open. Theses have life cycles. They get bought, they pay out — partly or fully — and the reward they promised gets collected. A thesis that has paid out is no longer a reason to hold; it is a memory of a reason. If you bought on three theses and two have substantially played out, you are no longer holding the position for the reasons you bought it. You are holding it on whichever single thesis remains open, plus the gravitational pull of already owning it. Any portion of the position that the remaining open thesis does not justify is being held purely because you already own it. That portion is attachment, not analysis, and your cost basis tells you nothing about whether it should stay.

This is the move conviction resists most fiercely. The narrative of I was right fuses with the narrative of I should continue, and the second does not actually follow from the first.

The Bear Case First

Before any conviction-tested position gets re-evaluated, the bear case has to be heard. Not as a token gesture — as the opening argument, given the first word and the longest paragraph.

I do this in writing, on every concentrated position I hold, on a rolling basis. Writing forces me to make the case in a form a sceptical peer could read; speaking it to myself does not. And I do it first because the case for prudence has to be allowed to land before the case for conviction gets to rebut it. If the bull case speaks first, the bear case becomes a list of objections to overcome rather than a position to weigh. The order is the discipline.

This is the section where AI has been genuinely useful in my own practice. A well-constructed analytical prompt — make the strongest case for being smaller, before any rebuttal — produces an output uncoloured by my own decade of attachment. It is not better than my reasoning. It is unattached to it, which is a different and sometimes more valuable thing.

Pillar Status: Fired, Open, Breaking

The core analytical move on a concentrated position is to classify each pillar of the original thesis into one of three states.

A pillar is fired when the reward it promised has been substantially collected. A relative-value gap that has closed. An expected re-rating that has happened. A catalyst that has played out. Fired does not mean the pillar was wrong — it usually means it was right. It means it is no longer a forward reason to hold, because what it promised has been delivered.

A pillar is open when the thesis it represented is still in play and still warrants exposure. The structural story has not yet resolved. The catalyst has not yet arrived. The cycle has not run its course.

A pillar is breaking when evidence has emerged that the thesis is weakening or wrong. Note that price falling is never the evidence. Prices fall inside healthy long-term trends; that is the whole point of holding through volatility. A pillar breaks on a structural development — supply responds, demand softens, the policy environment shifts, a competing technology cures the bottleneck.

The exercise is to take a position you currently hold, list the three or four reasons you originally bought it, and honestly label each one. In my experience, the answer is almost always that one or two have fired, one remains open, and zero are breaking. That distribution is what makes the question hard. If everything were still open you would simply hold. If everything were fired or breaking you would exit. The interesting case — and the case you will face on every long-held concentrated position — is the one where you have collected most of the original reward, one structural reason remains, and the question is how much of the position the single remaining reason actually justifies.

In my experience, the honest answer is usually less than the position actually held.

The Anti-Anchoring Check

Once the pillars have been classified, the next move is to strip out cost basis entirely and ask the only forward question that matters: at today’s price, with the reward from the fired pillars already collected, would you buy this size fresh?

Most principals find this question difficult, which is itself the signal. The difficulty is the cost basis fighting back. The question feels artificial because you did not buy fresh — you accumulated over years, at lower prices, with a different thesis weighting. All of that is irrelevant. The portfolio you carry today is the only portfolio you have, and the only honest question is whether you would build it from cash if you had to.

If the answer is no — if you would not buy this size today — the gap between what you would buy and what you currently hold is precisely the part of the position you are keeping only because you already own it. That gap is the trim signal. It is also a quantitative answer to a question most principals can only answer qualitatively, which is part of why the method is useful: it converts a vague feeling of probably too big into a specific number.

I run this check on every concentrated position every quarter. The number that comes out is sometimes zero: the position is correctly sized and nothing changes. More often there is a gap, and the gap becomes my trim plan.

The Floor Question That Almost Nobody Names

The single most important and most underrated question on a concentrated position has nothing to do with price, ratio, momentum, or any of the analytical moves above. It is simply: what is the level of wealth below which my family’s security is genuinely threatened?

Most principals at this tier have never written this number down. They have intentions about it, held in one person’s head, and a sense that they would know if things were getting dangerous. That is not a floor; it is a guess about a floor. And without the number, you cannot honestly answer the question the position is actually putting to you — which is not will this go down but can I afford for this to go down.

Run the scenario the position fears most. A 50% drawdown across the position and its correlated holdings, conducted with the structural thesis still intact. Mark the portfolio after. Then compare that number to the floor.

When I ran this on my own situation, the result reframed the entire decision. The drawdown that had been keeping me frozen did not come close to the floor. The capital at risk in the concentrated position was genuine risk capital, and a deep drawdown — while painful — would not threaten anything I had defined as untouchable. That single finding changed the meaning of the trim decision: I was no longer trimming to protect a floor that was not in play. I was trimming because no single thesis, however correct, deserves to be that large a share of the liquid book. Affordable conviction still warranted right-sizing, but for reasons of single-asset risk rather than fear of ruin.

This is the move most operators skip, and it costs them clarity in both directions. Some trim too aggressively, in retreat from a drawdown that would not actually have threatened them. Others hold too long, on a conviction that would not survive the drawdown they have not bothered to model. The floor question is what tells you which mistake you are at risk of making.

The Single-Bet Collapse

Five concentrated holdings is not five bets. This is the trap that hides the true concentration of a portfolio from the principal carrying it.

If you hold, say, gold, silver, uranium, copper and oil, the closeness that makes you a good commodity analyst is the same closeness that lets you experience each metal as its own well-researched conviction rather than as one position expressed five ways. In the calm market, the distinction has some merit — the commodities have different drivers, different cycles, different sensitivities. In a genuine risk-off event, the distinction collapses. A growth shock, a USD spike, a China hard landing, a disinflation — and the cross-correlations converge toward one, precisely in the moment you would most want diversification to hold.

The honest accounting is to ask: how many factors are you exposed to, not how many tickers. Most family-office portfolios that look diversified on a holdings list resolve into two or three genuine bets when traced back to factor exposure, and those bets often share a deeper macro spine — a real-asset, inflation, AI-infrastructure, de-dollarisation story, for instance, that binds them together far more tightly than the ticker count suggests.

This matters because it changes the meaning of adding diversification. A principal who responds to commodity concentration by adding a different commodity has not diversified; they have relabelled the same factor exposure. Genuine diversification means exposure to something that does not fall when the original position falls. What that is depends entirely on the family’s circumstances; the point is that a different ticker carrying the same factor is not it.

Where AI Helps, and Where It Has to Stop

Pillar classification, the anti-anchoring computation, the factor-exposure decomposition, the scenario modelling — this is exactly the kind of work AI is now genuinely good at. It can hold the framework steady when you cannot. It can ignore the cost basis you keep wanting to reintroduce. It can run the same check with the same rigour every quarter without becoming bored or sentimental. For a principal trying to impose discipline on themselves, that consistency is the entire value proposition.

But there is a boundary, and it is the same boundary that ran through the previous article in this series. The lens tells you the position is oversized against the remaining thesis. It does not, and cannot, tell you the pace at which to right-size it. That decision turns on things no model has: the floor you have not given it, the family’s actual risk capacity, the tax structure, the timing of liquidity events, the question of whether the right answer is to trim aggressively now or to trim gently into strength over a year. The analysis identifies the gap. The principal owns the response.

Confusing those two — letting the analysis make the decision — is the failure mode that turns a useful tool into a dangerous one. The principals who get the most from AI in this work are precisely the ones who treat it as a relentless, unsentimental analyst that does the work they are too close to do, and who keep every allocation decision firmly in human hands.

How I Trim: Into Strength, Never to Zero

When the gap has been identified and the floor checked, the execution discipline is the last piece. These are the three principles I pre-commit to in writing, before any volatility arrives. They are how I run my own book, not rules for anyone else’s.

I trim into strength, not weakness. On a position whose remaining thesis is intact, selling into a fall is the move I have learned to regret. So the triggers are defined in advance: a new high crossed, a ratio compressed to a set level, a milestone weight reached. If the run does not come, I was disciplined. If it does, the trims happen at progressively better prices. This rule cannot produce the regret of I sold too low.

I do not trim to zero while the structural thesis is open. A core size, small and well defined, sized to the stake I would still choose if starting from cash, survives the plan. A structural thesis can take years to resolve, and a pullback with the thesis intact, after exiting, is the most expensive way to be right.

I pre-commit the plan in writing. The version of you watching the position in real time, with the news in front of you and the chart moving, is not the version that should be making the next decision. The decision should already have been made — calmly, in writing, with the bear case heard first — and the screen-watching self’s only job is execution. That is what pre-commitment buys: a plan that does not need to be re-decided every Monday.

The Mandatory Disconfirmer

The discipline closes on a question. The remaining open pillar — the one structural thesis that justifies the core you continue to hold — what specifically would prove it wrong?

This is the question conviction resists most, because answering it honestly creates an exit condition for a position you love. But asking it is what separates analysis from belief. The named disconfirmers should be structural, specific and observable. A price decline is not a disconfirmer; markets fluctuate. A supply response confirmed by inventory data is a disconfirmer. A demand erosion confirmed by industry consumption figures is a disconfirmer. A policy reversal in the jurisdiction the thesis depends on is a disconfirmer.

Write them down. Set a quarterly check. If none have fired, the structural pillar is intact and the core is justified. If any have, the analysis just became simpler in a way you will not enjoy: the only remaining reason to hold has weakened, and the core itself is now up for review.

There is a second layer of disconfirmer worth running, not on the position but on yourself. Ask honestly: is my target weight for this position higher than a year ago, and if so did it rise on new evidence or because the price rose? Has any fired pillar been quietly re-narrated as still-open to justify holding? Am I holding the position, or am I holding being the kind of person who holds this position? These questions are uncomfortable, and that is precisely why they are useful. Serenity about a concentrated position is the symptom, not the safety; well-calibrated unease is what keeps the analysis honest.

A thesis you cannot disconfirm is not a thesis. It is a belief. Beliefs are perfectly fine to have. They are not what a family office should be allocating capital against.

The Test of the Threshold

The threshold of a family office is the first time a principal has to govern the wealth that built itself almost by accident. The concentrated position that did the building is the first real test, and the way it is handled sets the operating culture of everything that follows.

A principal who learns to run this discipline — pillars, anti-anchoring, floor, factor exposure, AI-with-boundary, trim plan, disconfirmer — has done more than navigate a single decision. They have built the muscle that every future allocation will need. The concentrated position is just the place that muscle gets built, because it is the place where the temptation to substitute conviction for method is strongest.

Conviction got you here. It is not what will keep you here. The discipline of holding through the cycle is what does that, and it is entirely learnable. The principals who build it now — while the position is still rewarding them, and the question still feels theoretical — will hold their wealth through the next twenty years far better than those who try to learn it under duress in the next drawdown.