Why expected value is not enough to trade

LessonThis is a lesson restated in new words. Any figure in it is an example from the source, not this desk's record.

An edge, in the quantitative sense, is a positive expected value, the probability-weighted average outcome. You should not be able to narrate the last trade if the edge is real and the bet is sized to survive, just as a professional poker player cannot narrate the last twenty hands and still be playing thousands of them. Quants become bad traders when they treat a backtest as the match they are about to play, or when they put the whole portfolio on one positive-expectation idea and call the concentration “conviction.” Survival across independent bets is the rest of the job.

Why it matters

People ask “what is your edge?” because they want a reason you are still ahead after many iterations, not a reason yesterday was green. The casino is the only clean classroom. Roulette, craps, and blackjack without a card-counting fantasy have a fixed edge. The house’s expected value is positive, the player’s is the negative of that, and the two sum to zero. The arithmetic, if you keep playing, moves the player’s wealth toward zero. Nothing in that paragraph is a metaphor about grit.

Trading and poker do not hand you that fixed number. “Buy low and sell high” is as helpful as “score more goals.” Market making and systematic long-short books have institutional versions of an edge. The lecture is about a different skill: holding risk when the probability is not on a card.

The idea in plain words

A backtest says how a rule behaved in the file you already have. It does not name next month’s regime. Judging every rule by whether it beat the S&P 500 is an after-the-fact sport. Had the index fallen 30 percent, a smaller loss would look wise, and you did not know the fall when you chose the rule.

Questions expose the hole. Why is volatility so often in contango, with longer-dated implied volatility above the near term? Why is the reverse shape, backwardation, fleeting? Why should volatility revert? Another chart of the old sample will not answer.

Sizing sits next to the expectation. One favorable bet, sized as the whole account, can still hit zero. The lecture’s repair is several bets whose drivers are separate enough that one piece of bad news is not fatal. The scoreboard is compound growth over years. A 30 or 40 percent year in a popular name, then a 20 percent miss, can leave a geometric result near zero if you never rebalance. The rebalance gain people call Shannon’s demon appears only if you rebalance.

Poker keeps the ethics honest. A better hand can lose a pot, and a professional playing thousands of hands should not narrate each one. Cash in Treasuries is a risk. Cash in equities is another. There is no training wheel at the terminal.

Updating is the live skill. Lose a set and you either trust the plan as variance or change it because the live edge looks negative. A quarter of bad passes might be noise, or a reason to run the ball. Markets are only a cleaner scoreboard. If you owned more of a risk than its premium required and you were right, the residual shows up as positive alpha. If you were wrong, it shows up as negative alpha.

How to use it

Write the expectation, then write the loss that ends the strategy, then add a second bet that does not die from the same event. If you only have one bet, you do not have the portfolio the lecture is describing.

What usually goes wrong

Remembering every trade, which means the sample is too small or the risk is too large. Demanding that every book beat the index after the index’s path is known. Running one high-expectation idea at a size that can take you to zero. And refusing to change a plan mid-game because the backtest still looks clean on the screen. Last year’s champion is a clue, not a proof, which is why the lecture points at long sports dynasties as arguments rather than as destiny.

Questions people ask

What is an edge if it is not a win on the last trade?

A positive expected value over many bets. If sizing is sane, the last bet should be forgettable. A huge win or loss is a sign the size was not sane.

Why is beating the index the wrong hurdle?

You only know the index path afterward. A strategy can look better or worse than that path for reasons that were not available when you chose it.

What is the updating rule in a match?

If the plan’s live edge has turned negative, change the action. If you still believe the shortfall is variance, continue. The backtest cannot make that call for you.