← All lessons

Reshuffled trade sequences

The same trades in another order tell a very different story. · 7 min

A backtest gives one list of trades in one order. The order was luck. Dealing the same trades in random orders shows the drawdowns and streaks you could just as easily have lived through.

Predict first

A backtest of 100 trades made +29.9% at 1% risk with a maximum drawdown of 7.6%. Reshuffle the same trades 1,000 times. What happens to the final result?

Same trades, different order

One hundred trades from a backtest, dealt in 1,000 random orders. Every order ends at the same place; look at the road.

Running the numbers…

final equity = Π (1 + risk × Rᵢ): the same in every order
drawdown and losing streaks: depend on the order

Worked example

Three trades at 10% risk: +2R, −1R, −1R.

  • Order A: 1.2 × 0.9 × 0.9 = 0.972. Worst drawdown: from 1.2 to 0.972 = 19%.
  • Order B: 0.9 × 0.9 × 1.2 = 0.972. Worst drawdown: from 1 to 0.81 = 19%.
  • Order C: 0.9 × 1.2 × 0.9 = 0.972. Worst drawdown: 10%.

Same end, drawdowns from 10% to 19%.

Common mistake

“My backtest’s maximum drawdown is my risk.” It is one sample of the drawdowns your trades can produce, and often a mild one. Plan for the bad end of the reshuffled distribution: here, one order in twenty reached 12.7% against the backtest’s 7.6%, and the longest losing streak reached 10 trades.

Reshuffling trades changes which of these?
A reshuffle shows a 1-in-20 drawdown of 24%; the backtest’s was 11%. What should guide your risk per trade?

Closing round

Size it from the fan

A strategy’s next 100 trades at 1% risk, 600 times.

Building a fresh case…

Further reading: Davey, Building Winning Algorithmic Trading Systems (Monte Carlo analysis).