Risk

Ruin Risk, Ergodicity, and Why the Average Outcome Is Never Yours

Separate the average across parallel businesses from the growth rate along your own path, so that a positive expected value stops being an argument for taking a bet and starts being one input into how much of it to take.

  • Expert
  • 15 min total
  • 14 chapters

What decision this helps you make: Whether an opportunity with a genuinely favourable average is survivable at the size proposed — and what the size would have to be before the arithmetic of your own path, rather than the arithmetic of the spreadsheet, turns positive.

What this topic is

Ruin risk is the probability that a sequence of decisions ends your ability to make the next one. Ergodicity is the property a process has when the average across many parallel copies equals the growth rate along one copy over time. Business returns compound rather than add, which makes them non-ergodic: the two averages diverge, and the one that describes your experience is the second. The practical consequence is that a bet can have a healthy positive expected value per round and still drive the typical path toward zero, and no amount of repetition fixes it — repetition makes it worse.

Why it matters

Almost every financial model in ordinary business use computes an average across outcomes and calls it the value. That number is the mean of a distribution of parallel worlds, and you live in one world, in sequence, with an absorbing floor underneath you. The gap between the two shows up as the difference between a plan that projected steady growth and a decade that delivered stagnation with no single identifiable mistake in it. It also explains the two things practitioners already do instinctively — refusing bets that could end the firm, and sizing positions rather than accepting or rejecting them — and gives both a defensible arithmetic basis.

Who should learn it

Owners weighing a bet-the-company expansion, finance leaders who keep approving positive-expected-value proposals, anyone using leverage, and any operator who has watched an average-based forecast quietly fail to arrive.

What you will understand

  • How to compute the growth rate of your own path, and why it sits below the average of the outcomes
  • The volatility drag formula, which prices the gap in one line you can do on a whiteboard
  • Why leverage raises the average and lowers the typical outcome at the same time
  • How to write a survival constraint that a positive expected value is not allowed to overrule

Prerequisites

Common misconception

"Over enough repetitions the odds even out, so a positive expected value is worth taking." Repetition is exactly what the argument gets wrong. The averaging story requires many draws that are independent and, quietly, that you are still solvent to take the later ones — and when each outcome multiplies your capital rather than adding to it, the sequence you actually walk compounds at the geometric mean of the outcomes, not the arithmetic mean. The geometric mean is always lower, and it falls with the square of the spread. So repeating a favourable-on-average, high-variance bet does not converge you toward the average. It converges the typical path toward zero, faster the more times you play.