S&P 500 And Gold Rotation Strategy (SPY And GLD Rotation)
Last Updated on April 18, 2023
Many traders look at the gold and S&P 500 ratio. The ratio can be used to create sector rotation strategies are popular. One of the reasons is the anticipation of better and more efficient capital allocation, and perhaps a second aim is to reduce the drawdown. Today we backtest a strategy that rotates between the S&P 500 and gold: S&P 500 and gold rotation strategy.
When the S&P 500 and gold ratio is above its moving average, ie. when stocks are performing better than gold, we are long S&P 500 the coming month. When the ratio falls below the moving average we switch to gold. Pretty simple, but as it turns out, it works pretty well.
What is a rotation strategy?
A rotation strategy seeks to allocate capital efficiently between two or more asset classes or within the same asset class.
For example, as in this article, the strategy aims to increase the return by switching in and out of the S&P 500 and gold by using the gold and S&P 500 ratio. When the quantified rotation criteria say the next time period favors gold, we buy gold. When the criteria change, we sell gold and buy the S&P 500.
The logic is easy to grasp, but to make a profit is a whole other ballgame. Rotation strategies are no easy money.
How do you make a rotation strategy?
In order to make a rotation strategy, you need to start with a plan and an idea. We recommend always having one main principle in the back of your mind: simplicity is important and should be achieved. We have yet to see a rotation strategy that performs well if it includes many criteria.
Do sector rotation strategies work?
Some sector rotation strategies work, some don’t. Rotation strategies are no easy road to riches, as no strategies are.
However, as you’ll see in this article, our very simple rotation strategy between the S&P 500 and gold seems to work pretty well.
Furthermore, we have published sector rotation and momentum strategies in the past that show huge potential despite their simplicity:
- Monthly momentum in SPY and TLT (rotation strategy S&P 500 and Treasury bonds)
- Monthly momentum in ETFs (sector rotation in EEM, SPY, and TLT)
Our S&P 500 and gold rotation strategy:
We tested our strategy by using the monthly spot gold price and the S&P 500 data from Yahoo!finance (^GSPC). The latter is, to our knowledge, not dividend-adjusted, thus it underrates the performance. Dividend reinvestments are a major part of the total return over time.
The S&P 500 and gold rotation strategy criteria:
We tested the following:
If the S&P 500 and gold ratio (S&P 500 divided by the gold price) is above its 20-month simple moving average, we are long the S&P 500 from the close of the month until next month’s close.
If at next month’s close the S&P 500 and gold ratio is below its moving average, we sell the S&P 500 and buy gold for the next month.
Rinse and repeat at the close of every month. We tested from 1985 until July 2020.
The ratio and its moving average looks like this:
When the blue line (the ratio) is above the red line (the 20-month average), we are long the S&P 500 the next month. Vice versa and we are long gold.
How did the S&P 500/gold rotation strategy perform?
Here is the equity chart showing the strategy compared to buy and hold for both the S&P 500 and the gold price:
The portfolio started with 100 000 in July 1986 and shows the compounded results.
The strategy made a CAGR of 9.9% with a max drawdown of 22%, significantly better than the max drawdown of 55% for the S&P 500.
The rotation strategy underperformed the S&P 500 all the way until the GFC in 2008/09, but “took off” after that. One of the reasons for that is a much lower drawdown during the crisis: when the S&P 500 made a new low in March 2009, the portfolio made a new high.
This means the strategy could start compounding on a much higher base after the crisis, something that paid off handsomely.
Which asset class is contributing the most to the overall performance? Gold or the S&P 500?
It turns out it’s pretty equal. The chart above shows the accumulated percentage gains divided by each asset:
The blues line shows the accumulated percentage gains in the S&P 500 while the red line is the gains in gold.
We remind you about our monthly Trading Edges where we publish an original trading idea/edge/strategy 12 times a year. The annual subscription gives you access to the past 12 Trading Edges plus the next 12 Trading edges – 24 in total:
S&P 500 and gold rotation strategy – conclusion:
The S&P 500 and gold rotation strategy, which is based on the gold & S&P 500 ratio, seems to generate pretty impressive results: both beating the buy and hold of the S&P 500 and gold.
We only tested the 20-month moving average. Perhaps it works better or worse with another average?
Keep in mind that dividends are not included, something that underrates the buy and hold CAGR of the S&P 500, but at the same time this also underrates the strategy’s result somewhat.
Hi Oddmund. I have tested this strategy with Yahoo Finance monthly data since 1/1/2005 till 1/6/21 and I find that you get the best results with a 7-month moving average.
In my analysis with a 20-SMA, if you invest 100.000 in 2005 you get 500.000 in 2021, but if you switch to a 7-SMA you get 700.000, that to say allmost a 50% more.
Could you confirm this please?
Yes, but you need to check the results for all months. For example, if it differs a lot to 8 and 9 months, you are most likely curve fitting.
I have done and it seems that you are right. These are different values for SMA different SMA and you can see you have a extrange “peak” at 7. But the good part is that one can see it is very stable. Isnt it?
2 563512,6544
3 457843,5218
4 571753,2852
5 574922,6936
6 563598,5016
7 706316,0367
8 515800,1153
9 477366,3791
10 428185,5819
11 461750,7161
12 453165,5842
13 419678,4674
14 414054,9825
15 487337,9768
16 499526,0704
17 498694,0744
18 545464,1124
18 545464,1124
19 503594,5401
20 508983,3858
21 555884,9295
22 496724,3882
23 496747,3525
24 498702,4185
25 415847,0649
26 421237,6624
27 425109,0696
28 432866,5153
29 440580,9887
Yes, it looks reasonable stable 🙂