How Much? The question that decides your returns — and it isn’t what you buy
In 2016, two researchers — Victor Haghani (a founder of the ill-fated hedge fund LTCM, now running Elm Wealth) and Richard Dewey — handed 61 people a rigged game and watched them fail at it.
The setup was almost insultingly generous. Each participant got $25 and thirty minutes. They could bet any amount on the flip of a coin that, they were told clearly and in bold, came up heads 60% of the time. A guaranteed, permanent edge. Winnings were capped at $250 — ten times the stake — a cap you could stroll to with steady, sensible bets.
These weren’t amateurs. They were economics and finance students and young professionals at real finance firms. And here is what happened:
28% of them went bust — walked away with nothing, from a game rigged in their favor
About a third ended with less than the $25 they started with
Only 21% reached the maximum, versus the ~95% who would have, with nothing more than a steady 10–20% bet each flip
18 of the 61 bet their entire bankroll on a single flip — and some bet it all on tails, against the very edge they’d been handed
They had the edge. They lost anyway. Because knowing the coin is biased tells you nothing about how much to bet — and that second question, it turns out, matters more than the first.
This is the most under-taught skill in all of investing. Ninety-five percent of financial content is about what to buy — the pick, the thesis, the ticker. Almost none of it is about how much, even though position sizing is what actually determines whether a portfolio of good ideas compounds into wealth or blows up on the way there. You can be right about your stocks and still go broke. You can be an average stock-picker and compound beautifully. The difference is sizing.
Here’s the uncomfortable part we’ll prove with arithmetic: a strategy with a real, positive edge — one that should make money — will still march to zero if you bet too much on each play. Overbetting a winning game is one of the most reliable ways to lose.
In this edition (premium):
The geometry of compounding — why a 50% loss needs a 100% gain to undo, and how volatility silently eats returns (with the tables)
The Kelly Criterion, made usable — the formula the pros actually size with, in plain arithmetic, worked on a real position
The single most important rule in risk: why overbetting is fatal while underbetting is merely slow — and the asymmetry that follows from it
A practical sizing system for when you can’t compute a precise edge — conviction tiers, position caps, and correlation guardrails
A full worked example — sizing one position from raw edge to final allocation — plus the 5 ways sizing quietly kills portfolios
Part 1 — Why sizing beats selection: the geometry of compounding
Most investors think in terms of average returns. Compounding doesn’t work on averages — it works on