The story so far: The article on leveraged ETFs said that they were good but with a lot of volatility and the article on dollar cost averaging (DCA) said that regular investing into volatile investments can benefit from volatility. So what happens when we regularly invest into leveraged ETFs? Do we get double-good?
Leveraged ETFs on the S&P 500 such as UPRO are leveraged on a daily basis so to test investment strategies we need daily data. Unfortunately we don't have daily data for the S&P 500 Total Return going back to 1800 or even to the stock market crash of 1987. So we miss out on some periods of time when we really would like to see how the strategy performed since crashes are when we lose the most money and a crash seems to come along once a lifetime.
So we interpolate by simulating the days in between the month ends. We do this by generating a random walk with the required return and which has the required volatility as determined by fitting an EGARCH model to the monthly figures.
We show the simulated data in the chart below. The interpolated prices series in blue is indistinguishable from the monthly prices in the DCA article - this chart does not have enough resolution to show the daily values.
We also simulated the prices for the 3x leveraged ETF UPRO. The UPRO simulation assumes annual fees of 1%. It is clear that the 3x leverage produces 3 times higher volatility and much higher return. We will see the higher return in a later chart. The previous article showed that 3x leverage was profitable for the S&P 500 going back to 1950. This chart shows that it was profitable right back to 1800.
As in the previous article we will plot the annualised return as a function of leverage. In the chart below the blue line is without fees and the red line with fees deducted. The orange dots indicate ETFs of interest - the lower dot is the S&P 500 unleveraged and the upper is UPRO at 3x leverage and 1% fees.
The S&P 500 annualised return is 8.4%, UPRO is 20.4%.
Looks like we could have gone up to more than 10x leverage without doing worse than the unleveraged S&P 500. But high leverage is risky - at 5 or more leverage you would have been wiped out by the 1987 crash where the S&P 500 dropped by more than 20% in a day.
Anyway, on with the story. What about dollar cost averaging?
Suppose we invest $1 each month into a leveraged ETF what would our final value be after 200 years? We plot the log of the final value as a function of leverage below.
It looks the same as the previous chart. The peak is in the same place - about leverage = 5.8. If leverage favoured DCA then we would have expected the peak to be at a higher leverage. A bit disappointing.
This is still consistent with our DCA article where we showed that all other things being equal then DCA likes volatility. But here we don't have all other things equal because the higher volatility investment also has higher returns.
Conclusion - Dollar cost averaging is of no advantage to leveraged investing. There is no double-good. Just the single good of using leverage.
But 20.4% return over 200 years - we can count that as triple-good in anyone's book.
The fact that this data was simulated has no influence on the conclusions of this article - the results would be the same if we had the real data. But there is a subtle difference - we don't have the S&P Total Return daily prices for the 1987 crash. The S&P500 lost more than 20% in one day - that would have wiped out any 5x leveraged ETF. Our simulations don't replicate that single outlier. So our charts can go higher than 5x leverage even though in real life returns would have gone to zero for leverage greater than 5.