Modern portfolio theory has been the most influential intellectual contribution to investing in the past sixty years and, in many respects, the most quietly destructive. It taught a generation of allocators that the appropriate response to uncertainty is diversification across uncorrelated assets. It is a perfectly defensible position for capital that has no information advantage. It is the wrong position for capital that does.
What diversification actually does
Diversification reduces idiosyncratic risk. It does this by averaging across positions whose specific outcomes are unrelated. The mathematical case for it is solid. The practical case for it is more conditional than the theory suggests.
Diversification works when the diversifier has no view on which positions are better than others. Once a view exists, diversification becomes a tax on conviction. You hold positions you have less confidence in alongside positions you have more confidence in, and you average down your expected return in exchange for a reduction in variance you may not actually need.
Most institutional capital diversifies for two reasons that have nothing to do with optimising return: career risk for the allocator, and regulatory or contractual constraints. Both are entirely rational. Neither is relevant to capital that has neither constraint.
The argument for concentration.
For long-duration capital that does its own work, concentration is the natural outcome of conviction. If we have spent six months underwriting a position, we should hold it in size. If we have not spent six months underwriting a position, we should not hold it at all.
This is, in our experience, an uncomfortable position to hold. It feels — to clients, to advisors, to other professionals — like the absence of risk management. It is, in fact, the presence of a different kind of risk management: the management of risk through underwriting rather than through diversification.
"We do not diversify away risks. We underwrite them. The difference is the work."
What we underwrite when we cannot diversify
When we hold a £40m position in a single development scheme, or a £60m position in a single distressed loan portfolio, we cannot use diversification to handle the residual risk. We have to underwrite it directly. This means three things:
Forensic asset diligence
Every concentrated position is examined in a level of detail that diversified investors do not need to apply. We model the asset, the operator, the structure, and the macro sensitivities at the level of individual line items. Most of the work that distinguishes our portfolio from comparable allocators is the work that happens before we commit, not after.
Operating-partner alignment
We do not hold any concentrated position alone. Every meaningful exposure is alongside an operating partner whose own capital is in the deal at material weight, and whose track record on this specific asset class we have observed across at least one full cycle. The diligence on the operator is, in many cases, more time-consuming than the diligence on the asset.
Structural protection
Where the structure permits, we build in protection that diversification would have provided automatically: subordinated capital protections, contracted minimum returns, exit rights at specified milestones. These are not always available. Where they are, we use them. Where they are not, we either underwrite the risk directly or decline the position.
Position sizing under uncertainty.
Position sizing is one of the most under-discussed disciplines in investing. The standard frame — equal weights, theme weights, or volatility-based weights — is a poor fit for our work. Our positions vary from £20m to £150m, and the variance is driven less by macro sizing and more by what we believe the position can absorb without destroying our balance sheet if it goes wrong.
The Kelly criterion, in its various practical forms, is closer to how we actually think. The position should be sized to the probability of being right, the magnitude of being right, and the cost of being wrong. We do not run the model formally on every position. We run a version of it implicitly, every time, and we tend to size below what the model would suggest because the model assumes more precision in our probability estimates than we believe we possess.
The behavioural element.
The hardest part of holding concentrated positions is not the analytical work — it is the behavioural discipline required when the position is in drawdown. Every concentrated position we have ever held has had moments where it was uncomfortable to hold. Some of those moments lasted years. The conviction that allows us to hold through them is the same conviction that produced the position in the first place: a deep, evidence-based belief that the position is what we said it was, regardless of where the mark sits.
This is why we do our own diligence. The conviction that survives a difficult holding period is the conviction that is grounded in the work, not in the recommendation of someone else. We have observed many investors who were unable to hold positions they had not personally underwritten, even when the underwriting was, in retrospect, perfectly sound.
When concentration is wrong.
Concentration is not always correct, even for us. There are two situations in which we deliberately diversify:
When we do not have an information advantage. Public equities are a market in which we do not believe we have a sustainable information advantage. Where we hold listed exposure, we hold it broadly and passively. The work of trying to outperform a market we cannot inform is not, in our experience, rewarding.
When the cost of being wrong is asymmetric. Some risks — counterparty risk, jurisdictional risk, technology obsolescence risk — have such asymmetric downsides that diversifying them away is correct even where conviction exists on the underlying asset. We diversify counterparties. We diversify jurisdictions. We diversify operators within a single asset class.
Beyond those, we tend to be concentrated. Our portfolio currently holds twenty-seven active investments across four verticals. The largest position is around eight percent of the book. The smallest is around one percent. The distribution is intentional, and the work behind every position is the work that makes the concentration justified.
The discipline that makes it work
Concentration without discipline is recklessness. Diversification without discipline is mediocrity. The discipline that distinguishes useful concentration from reckless concentration is the willingness to do the underwriting work, to size below your apparent conviction, and to walk away when the diligence does not support the position — even after you have spent the time and want to invest. We have killed more deals at the final committee stage than we have approved. The deals we do approve are, by virtue of that filter, deals we are willing to hold concentrated.
We are not arguing for concentration as a universal principle. We are arguing for it as the natural consequence of doing the work — of underwriting individual opportunities deeply enough that the appropriate response to a good opportunity is to put real capital behind it. The alternative is to invest broadly and shallowly, which is, for capital that has the time and the temperament to do the work, an inferior strategy.