BEAT THE COUCH · THE MONKEY AUDIT

Was It Skill or Luck? The Monkey Audit

Beat the couch and the game does not congratulate you. It calls in 1,000 monkeys. Each one gets your two years, your days in cash and your number of trades, and spends them on random days. If you did better than 90 of every 100 monkeys, your win is stamped TIMED THE MARKET. If not, it is stamped LUCKY, because there is no evidence it was anything else.

How the audit works

1. Cut out your days in cash. Your 504 trading days split into stretches in the market and stretches in cash. Sell in October and buy back in December, and you have one cash stretch of about seven weeks.
2. Shuffle them. A monkey takes the same cash stretches, in random order, and drops them on random days. Every stretch stays whole, so the monkey is out for exactly as long as you were, in as many pieces.
3. Replay the same prices. The monkey rides the same real two years of the S&P 500, dividends included, and earns the same Treasury-bill rate in cash.
4. Do it 1,000 times. A thousand monkeys, a thousand random schedules, a thousand results against the couch.
5. Count. Your percentile is the share of monkeys you beat (a tie counts half). 90 or better and the win counts as skill.

The monkeys keep everything about your game except the timing: the same exposure, the same number of moves, the same market. So the audit answers one question and only that one. Were your days in cash better chosen than random days?

Run it yourself: the 1987 crash

The window below is the one the game uses for 1987, from Jan 2, 1987 to Dec 28, 1988, with Black Monday (the S&P 500's worst day ever, -20.5%) in the middle. Pick a trader, then release the monkeys. Everything runs in your browser, the same code the game runs after every win.

Sells a week before Black Monday, buys back seven weeks later.

The four traders, audited

Figures from one run of 1,000 monkeys at build time. Each run in your browser draws new monkeys, so the percentile moves by a point or two.

Look at the gap between the first two. Both sold into the crash. The prophet got out before the fall and is stamped TIMED THE MARKET. The Friday seller got out after the market had already fallen for a week, and lands right at the line: run it a few times and the stamp flips between LUCKY and TIMED THE MARKET. That is what a borderline call looks like. The fidget is 6.0% ahead of the couch and still LUCKY: with eight trades and months in cash, plenty of monkeys did as well by accident. The panic seller lost to the couch, and the monkeys only confirm it: most random schedules with the same months in cash did better.

Why monkeys

The idea is old. In A Random Walk Down Wall Street (1973), Burton Malkiel wrote that a blindfolded monkey throwing darts at the stock pages could pick a portfolio as well as the experts. The audit makes that a test you can run. Statisticians call it a permutation test: shuffle the thing you claim matters (here, the timing of your trades), keep everything else, and see how often chance alone does as well as you did. If chance matches you often, your result is no evidence of skill.

The 90th percentile is a generous bar. Pure chance clears it one time in ten, which in a science paper would not be enough (those usually ask for one in twenty or one in a hundred). It is a game, and the couch can afford to be kind.

What it does not check

The audit judges the timing, not the exposure. A monkey that sits out eight months of a rising market loses for the same reason you did, so a loss can come with a decent percentile and a win with a poor one. It also judges one game. Someone who plays a hundred times will collect a few TIMED THE MARKET stamps by chance alone, which is why the ledger counts every game, not the best ones. And it checks only the game's world: no fees, no taxes, no spreads, all of which would make real timing harder (see the fine print).

What the ledger says

Across every game ever filed, 31,504 ended as LUCKY wins and 14,789 as TIMED THE MARKET, out of 120,800 games. The couch won 62%. Most people who beat the couch did it the way most of the monkeys do: by accident.

Frequently asked

How do you tell skill from luck in market timing?

Compare your result with random traders who had the same exposure. Beat the Couch replays your days in cash on random dates 1,000 times (a permutation test). If you beat 90% of those random schedules, the win counts as skill; otherwise it is stamped LUCKY.

What is a permutation test in trading?

A test that shuffles the timing of a trader's positions while keeping time in the market and the number of trades fixed, then checks how often the shuffled versions do as well as the real one. If they often do, the timing shows no skill.

What does LUCKY mean in Beat the Couch?

You finished more than 2% ahead of the couch, but fewer than 90% of 1,000 random monkeys with your same days in cash did worse. The win counts on the ledger, in the lucky column.

Is the 1987 crash timeable?

In hindsight, yes: selling a week before Black Monday and buying back in December beat 99% of random schedules in our run. Selling the day after the crash lost to the couch.

BEAT THE COUCH · the game · the archive · the daily hands · the evidence · the data · for teachers · the monkey audit · the fine print · embed the ledger · Privacy · Terms

Not financial advice. The couch is not licensed for anything.