Prevolio

Eggs in One Basket

Split a pot across four investments, run a decade, then run it again — and see why one lucky result proves nothing.

See that investment returns are random: diversification narrows the range of outcomes without giving up much expected return, and a strategy has to be judged by its whole distribution rather than by the one run you happened to get.

  • Risk–return tradeoff
  • Diversification
  • Volatility
  • Expected value vs. realized outcome
  • Thinking in distributions

Guided walkthrough

Step 1 of 3
Predict first

Press 'All in on the hot stock', then run the decade five times over, watching the final number each time.

Before you act, predict what will happen. How much did the results differ from each other? Would you have called this a good strategy after seeing only your best run? Then do it — did your prediction match what happened?

Try a strategy:

The four numbers below are parts, not percentages. They add up to 100 right now, and each label shows the share of the pot it actually buys.

parts
parts
parts
parts
yr
Press Run to play out one random decade with this split.

What this strategy usually does

The single worst and best of 200 simulated decades of this exact split, with the typical (median) run between them. These three are extremes and the middle — the usual range below is a different thing: where most runs land.

Typical (median)
148
Average
156
Usual range (8 runs in 10)
96 – 230
Chance of losing money
11%
How this model works
AssetNormal yearYear-to-year swingCrash yearAverage year, crashes included
Cash1%none — the same every yearno crash branch1%
Bonds3%typically ± 6%, sometimes moreno crash branch3%
Index fund7%typically ± 18%, sometimes more2% chance of -35%6.2%
One hot stock14%typically ± 45%, sometimes more8% chance of -80%6.5%

Those last two columns are the whole sim in one line: a crash year replaces that year's normal draw rather than being averaged into it, so the hot stock's headline 14% is really 6.5% a year once its crash years are counted — barely more than the index fund, for wildly more risk.

Each asset's return is drawn on its own every year: nothing here is correlated, a bad year makes the next year no likelier to be bad, and there are no fees, taxes or inflation. Real markets have all four, which is why this is a model of the idea, not of a market.

At the end of every year the pot is rebalanced back to your split, so you keep holding the strategy you chose instead of letting a lucky winner take over.

In a single year an asset can lose at most 95% of its value, and a pot never falls below zero. Every run starts at 100 units.

Typical (median) is the middle run of the 200. Usual range is the middle 80% — 8 runs in 10 end inside it, and 1 in 10 ends below it. Worst run and best run are the two single most extreme of the 200, so they sit outside the usual range by definition.

These figures come from 200 simulated runs of a simplified model. They describe this model's behaviour, not a prediction about real markets — and nothing here is investment advice.

Related concepts