Position Sizing Comes Before Picking

Trevor L Taylor is a Miami private investor whose approach grew out of systematic and quantitative investment models, and this entry takes up the question those models put first: how much capital any single idea should receive. Private investors tend to talk about companies, buildings and funds. They rarely talk about sizing. Yet the share committed to one position usually decides more about long run results than the quality of the pick. A strong idea held at two percent of a pool barely moves the total, while an ordinary idea held at forty percent can define it in either direction.

Why picking gets the attention

Choosing feels like skill, and sizing feels like bookkeeping. That difference in feel is the reason sizing gets neglected, and the reason it should be decided before an investor meets a persuasive sponsor or founder. Once a story has been told well, an investor’s sense of the right amount is no longer neutral.

The data on individual stocks shows why humility about picking is warranted. Hendrik Bessembinder found that four of every seven US common stocks since 1926 earned less over their lifetimes than one month Treasury bills¹, and that the best performing four percent of listed companies account for the entire net gain of the market. Returns are skewed. A handful of positions do the heavy lifting, most do little, and nobody knows in advance which is which. Sizing is how an investor stays in the game long enough for the few winners to matter.

The arithmetic of a deep hole

Losses and gains are not symmetric. A position that falls by half must double to get back to even. The larger that position was, the deeper the hole in the whole pool, and the longer every other holding must work to fill it.

Mathematicians formalized the link between sizing and growth decades ago. In a paper on information theory, J. L. Kelly showed that a bettor with an edge has a maximum exponential rate at which capital can grow², reached by staking a particular fraction of capital on each wager. The result, now known as the Kelly criterion, implies that staking far more than that fraction on a good idea is itself a mistake.

People find this hard in practice. In an experiment run by Victor Haghani and Richard Dewey, 61 quantitatively trained young men and women bet on a coin weighted to land heads 60 percent of the time³. Most bet poorly, and a large share lost everything despite holding the edge. The Economist’s Buttonwood column summed up the study as evidence that even a simple probability exercise exposes poor betting choices⁴ even in a game far simpler than real markets.

Six rules that do the work

Rules written in advance protect an investor from enthusiasm. The six below translate cleanly to private markets, where positions are lumpy and hard to exit.

1. Set a ceiling per idea before looking at ideas. Choose the maximum share of investable capital any single private position may take at cost, and write it down while calm.
2. Size to the loss the pool can absorb. Ask what happens to the whole if this position goes to zero. If the answer changes how the investor lives or sleeps, the position is too large.
3. Start smaller when capital is locked up. A position that cannot be sold for seven years should be smaller than an otherwise identical one that could be sold next month, because locked capital cannot be redeployed.
4. Count correlated bets as one bet. Three apartment buildings in the same submarket with the same lender behave like one large position. Group them before testing them against the ceiling.
5. Leave room for the second check. Capital calls, reserve shortfalls and follow on rounds arrive later. Size the first commitment so the likely second one still fits.
6. Never resize upward because of a story. Add only when evidence shows the original thesis working, measured against criteria set at entry.

Rule four has a famous illustration in public markets. Amir Khandani and Andrew Lo traced the losses that hit quantitative long and short equity funds in early August 2007 to a coordinated deleveraging of similarly constructed portfolios⁵. Positions that looked separate moved together once everyone sold at once.

When concentration meets borrowed money

Concentration becomes most dangerous when it is financed. The Wall Street Journal described how banks earned fees financing the family office Archegos Capital Management, then learned the extent of its debt fueled bets on a handful of major stocks⁶ only as losses mounted. The mechanics are simple: a few large positions, leverage on top, and no room to wait.

The same pattern appeared a generation earlier at Long Term Capital Management. When that fund was hours from collapse, 14 banks and brokerage houses put in $3.6 billion and took 90 percent ownership⁷. The New York Times Magazine reported that the fund lost more than $4 billion in a six week panic⁸, despite staff who had built some of the best known models in finance. Clever analysis did not protect positions that were too large for the capital behind them.

A worked example

Imagine an investor with a pool set aside for private positions. The written ceiling is ten percent of that pool per idea at cost, and anything locked up for more than five years starts at half the ceiling.

A real estate development deal arrives with an experienced sponsor and a projected seven year hold. Under the rules, the maximum starting size is five percent because the lockup triggers the illiquidity haircut. The investor already owns a stake in a nearby project financed by the same lender, so the two are grouped and tested together against the ten percent ceiling. That combined test leaves room for only three percent in the new deal.

The second check matters more when distributions slow. Bain reported that distributions as a share of net asset value fell to their lowest rate in over a decade⁹, which leaves investors waiting on old commitments while new calls keep coming. In this example, the three percent stake leaves room under the ceiling if the sponsor asks for more.

Without rules, the same investor might have committed eight or ten percent after a strong presentation. With rules, the commitment is smaller and the pool survives a late, over budget or failed project. If the project goes well, three percent still contributes, and the investor has learned something about the sponsor before deciding whether to commit more next time.

The habit to keep is short. Write the ceiling down before the next deal arrives, and let that number, rather than the story, decide the size.