Links

Drawdown

Definition

Drawdown refers to the maximum difference between the portfolio's maximum value and any point thereafter. We will be observing the relationship between the net profit and the drawdown of the algorithm, and how these two can be used to determine performance against the underlying. Drawdown can be calculated as follows:
DrawdownPercentage=PortfolioMaximumValueLowestSubsequentPortfolioValueDrawdownPercentage=\frac{PortfolioMaximumValue}{LowestSubsequentPortfolioValue}

Backtests

In this section, we will be discussing the drawdown of the 6-year, 2-year, and 1-year backtests. Based on our backtests, we get the following drawdowns for the tested securities:
Security
2016-2022
2020-2022
2021-2022
AAPL
24.34%
24.34%
12.49%
NVDA
27.67%
23.43%
23%
AMD
46.82%
46.82%
46.82%
QQQ
24.75%
15.09%
11.11%
AMZN
18.47%
17.3%
17.3%
From these recorded values, we can calculate the average drawdown per backtest to be as follows:
Timeframe
Drawdown
2016-2020
28.41%
2020-2022
25.40%
2021-2022
22.14%

Drawdown Observations

Overall the drawdown remained relatively consistent throughout the timeframes tested, and actually decreased by about 6% as the backtesting length shortened. This helps support our initial claim of the algorithm performing stronger in more liquid market conditions. Overall, these numbers can be scaled down with relation to the net profit, as these tests were performed using 100% of the account balance for clearer results. With future iterations of the algorithm, as with all categories, we hope to improve the drawdown especially during less liquid market cycles.