Largest Drawdown Detection in Price Series | Dataford Interview Questions

Given a price/return series, write code that returns the n largest drawdowns.

Asked in the Online assessment stage. This was the programming part of the OA, and the candidate called it hard.

Implement largest_drawdowns(prices, n). A drawdown starts when prices fall below a running high, reaches its lowest trough, and ends when the prior high is recovered. Return up to n records as [peak_index, trough_index, recovery_index, drawdown_fraction], sorted from largest to smallest. Use -1 for an unrecovered drawdown. Exclude zero drawdowns. For ties, sort by earlier peak index.

Example: prices = [100, 95, 90, 96, 101, 98, 80, 102], n = 2 returns [[4, 6, 7, 0.2079207921], [0, 2, 4, 0.1]].

Constraints

1 <= prices.length <= 10000
1 <= n <= 10000
prices contains positive real numbers
A drawdown is measured as (peak_price - trough_price) / peak_price
Drawdown episodes are non-overlapping and end at the first recovery of their peak

Function Signature

def largest_drawdowns(prices, n):

Your question is Largest Drawdown Detection in Price Series. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem