Can the Golden Cross Rule be Improved Upon using Relative Strength?
The golden cross strategy is a very simple one – buy if the 50 day moving average is above the 200 day and sell if below. In my previous blog post I showed that this simple rule can produce decent returns on the Nasdaq 100 index if applied over the long term. However, the volatility and drawdowns make it a challenging strategy to stay invested.
In this post I will look at how we can improve upon the basic system by looking at the following:
- Add a relative strength rule
- Add risk management: volatility normalization
Add a Relative Strength Rule
Relative strength is an investment method that compares multiple assets and picks the strongest performing one. The trailing performance is re-computed at a set time interval and the holding is updated when another asset’s performance rises to the top.
In our case, we use a two-asset relative strength. We compare the traded asset, the Nasdaq 100, versus the S&P 500. The idea is to be invested in the Nasdaq 100 when its recent performance is greater than that of the S&P 500 and to be in cash when the Nasdaq’s performance is less than the S&P 500.
The S&P 500, being a broader index than the Nasdaq, serves as a good performance benchmark for the Nasdaq 100 which is more highly concentrated in technology and innovative companies. Here is a heatmap comparing the 12 month rolling returns of the two indexes from 2020 to 2024:
The above heatmap shows that when the Nasdaq 100 is darker green than the S&P 500, there is a likelihood of positive performance in the near term. It also shows that when the Nasdaq 100 is darker red than the S&P 500, there is a likelihood of negative performance in the near term.
Below are statistics comparing the Nasdaq 100’s one month performance after the occurrence of 3 different scenarios (from 1991 – 2024):
This result shows that the Nasdaq 100’s relative strength over the S&P 500 can serve as a reasonable predictor of Nasdaq 100’s near term future performance.
For the purposes of the augmenting the Golden Cross system, we use a slightly quicker relative strength system than the 12 month rule. Here is the rule:
In plain English, if the Nasdaq 100 price divided by its 60 day average is greater than the S&P 500 price divided by its 60 day average, then be long the Nasdaq 100, else be in cash. Here are the backtests:
Golden Cross System:
Relative Strength System:
70% Golden Cross + 30% Relative Strength:
We can see from the above that adding the Relative Strength system at a 30% weight has a synergistic effect. The Maximum Drawdown decreases from -35.7% to -28.5% and the Sharpe goes from 0.76 to 0.79 and the Return/Drawdown ratio goes from 0.44 to 0.50
Volatility Normalization
To further improve the system, I look at a risk management rule which I call ‘volatility normalization’. What it does is it reduces the exposure when volatility in the Nasdaq 100 rises above a certain threshold. This has the effect of decreasing drawdowns and increasing overall consistency of returns during volatile environments.
The rule is as follows:
The above rule reduces the exposure once the rolling 50 day volatility of Nasdaq 100 returns exceeds an annualized value of 30%, and the further it goes above that level, the less exposure it takes. Here is a graph of what the exposure would look like historically:
From the above graph we see that exposure was reduced to around 35% during the high volatility environment of the dot-com collapse. Similar reductions are seen in the 2008 collapse and the COVID crash of 2020.
Here is the Golden Cross + Relative Strength system before and after the volatility reduction rule:
We see an improvement in the risk parameters – the max drawdown goes from -28.5% to -25.3%. The Sharpe ratio goes from 0.79 to 0.81 and the Return/Drawdown ratio goes from 0.50 to 0.54.
Conclusions
By combining the basic Golden Cross system with a Relative Strength strategy, we were able to improve the Sharpe Ratio and Return/Drawdown ratios. Then upon adding the volatility reduction rule, we further improved these risk parameters resulting in a strategy that might be usable.
The annualized return of the final strategy, however, is a bit low for my liking at 13.6%. In future analysis I will look at ways of increasing the rate of return while keeping the drawdowns at a reasonable level.
Excel Spreadsheet
I’ve created a spreadsheet where you can adjust the various parameters discussed in this blog post and generate backtested equity curves.
