Expected value is the average outcome across many parallel trials. But you don't live in parallel trials — you live in one sequence, and in a sequence, hitting zero ends the game permanently. This gap is ruin risk, and it means a bet with positive expected value can be catastrophic if repeated with too much of your capital at stake.
The classic illustration: a coin flip that returns +50% on heads and −40% on tails has a positive expected value on paper. But bet your entire stack repeatedly and you go broke almost surely, because the multiplicative sequence of gains and losses compounds down, not up. This is why leverage destroys sophisticated people: it raises expected returns while introducing a path to zero. And zero is absorbing — you can't recover from it with a later good year. The practical rule: never risk what you can't afford to lose, no matter how good the odds look, because surviving is a precondition for compounding.
Long-Term Capital Management was arguably the most credentialed fund ever assembled: its partners included Myron Scholes and Robert Merton, who won the Nobel Prize in Economics in 1997 for the options-pricing model that underpins modern derivatives. Their strategies were genuinely sound on expected value, exploiting small, reliable pricing discrepancies. To make small edges into large returns, they used enormous leverage — at points over 25-to-1, with derivative exposure in the notional trillions. In 1998, Russia defaulted on its debt and correlations that their models treated as independent all moved together at once. The fund lost roughly $4.6 billion in four months and required a Federal Reserve-orchestrated bailout by fourteen banks to prevent a systemic cascade. The math had been right on average. The leverage meant they only needed to be wrong once.