In Oct. 2015, I wrote A Simple SPY Top-Off Portfolio, proposing a simple 2-fund strategy that should outperform the S&P 500 by a few percentage points a year. The idea is simple: allocate 1/3 of your assets to a 3x daily S&P ETF (e.g. the ProShares UltraPro S&P 500 ETF (NYSEARCA:UPRO)) and the remaining 2/3 to some sort of bond fund to generate additional returns.
In that first article, I suggested using a short-term bond fund like Vanguard’s Short-Term Bond Index Fund (MUTF:VBISX) to get a safe 1-3% annual advantage over the S&P, but also noted that you could "up the ante" by using a longer duration bond fund. The trade-off is that you’re more likely to occasionally underperform the S&P, since longer duration bond funds are more volatile than short-duration ones.
I actually implemented a 1/3 UPRO, 2/3 long-term bond fund (VBLTX) strategy for quite a while in my own portfolio, with excellent results. I recently decided to split the 2/3 allocation between VBLTX and the Vanguard High Yield Corporate Fund (MUTF:VWEHX), but that’s another story.
Performance vs. all Vanguard funds and ETFs
I’m still really high on the 1/3 UPRO, 2/3 bond idea, so I decided to look at how its performance compares to Vanguard's entire lineup of mutual funds and ETFs. UPRO was introduced on June 25, 2009, so I’ll look at performance since then.
For Vanguard mutual funds and ETFs, I got a comprehensive list from Vanguard’s website: Of the 176 products (55 ETFs and 121 non-money market mutual funds), 129 operated during the time period of interest and were thus included.
Figure 1 shows compound annualized growth rate vs. maximum drawdown for my 1/3 UPRO, 2/3 VBLTX strategy and the 129 Vanguard products (funds in blue, ETFs in red).
Right off the bat, you can see that UPRO/VBLTX had better raw returns than all 129 of the Vanguard products. Its CAGR was 21.3%; the top 3 Vanguard CAGRs were the Vanguard Consumer Discretionary ETF (NYSEARCA:VCR) at 19.8%, Vanguard Information Technology ETF (NYSEARCA:VGT) at 18.3%, and Vanguard Industrials ETF (NYSEARCA:VIS) at 17.7%.
As for max drawdown, only 39 of the 129 Vanguard products had a smaller MDD. At 14.3%, UPRO/VBLTX actually had a considerably better MDD than the Vanguard S&P 500 ETF Index Fund (MUTF:VFIAX) (18.7%).
My UPRO/VBLTX strategy also had an excellent Sharpe ratio of 0.097. That was better than all but 16 of the 129 Vanguard products. Notably, the 16 products with better Sharpe ratios all had much smaller raw returns than UPRO/VBLTX. The Vanguard Wellesley Income Fund (MUTF:VWINX) had the highest CAGR at 12.2%.
A natural criticism of these results is that I developed the strategy in Oct. 2015, but the time period examined here goes back to June 2009. In general, it’s fairly easy to develop strategies that beat pretty much everything looking backwards. Prospective outperformance is harder to achieve.
Figure 2 shows growth of $10k for my strategy vs. Vanguard's S&P fund VFIAX since I first published the UPRO/VBLTX idea (10/2/15 - 10/20/17).
Compared to VFIAX, the UPRO/VBLTX strategy had a higher CAGR (20.6% vs. 16.9%), smaller MDD (11.3% vs. 12.7%), and better Sharpe ratio (0.111 vs. 0.091).
Compared to all 129 Vanguard funds, UPRO/VBLTX had the 9th highest CAGR, 49th smallest MDD, and 4th highest Sharpe ratio. Again, none of the products with a higher Sharpe ratio had anywhere near the CAGR of UPRO/VBLTX (VWINX highest at 8.8%).
I would also add that this strategy is very simple and, in my view, theoretically sound. All I’m doing is using a 3x daily ETF to free up 2/3 of my assets to generate additional returns. It does not require "optimizing" any parameters using historical data that might be sub-optimal going forward. I suppose the choice of VBLTX is somewhat tied to the backtested time period, but VBLTX is not some obscure fund that happened to perform well since 2009; it is one of three Vanguard bond funds that targets a particular duration.
Driven by a bull market?
Skeptical readers might also point to the fact that we’ve been in a raging bull market in both stocks and bonds since 2009. Conventional wisdom says it’s easier to beat the market during a bull market than during a bear market, and I would agree with that.
I want to point out that while my strategy uses a leveraged ETF, its net leverage is no greater than the S&P’s. In fact, its net beta is slightly less than 1, since VBLTX has negative beta. So it’s not simply that the markets have been hot and therefore my high-beta strategy outperformed. If that were the case, my risk-adjusted returns would probably not be better than the S&P’s, and certainly not better than 87.6% of Vanguard's products.
Fixing what isn’t broken?
As I mentioned earlier, I recently decided to split up my 2/3 bond allocation between VBLTX and the high-yield VWEHX. The rationale is robustness; it’s less likely for two weakly correlated (0.05) bond funds to hit a rough patch than it is for one. One drawback of VWEHX is that it isn’t nearly as negatively correlated with stocks as VBLTX is. However, it has a higher dividend yield than VBLTX (SEC yield 4.41% vs. 3.49%).
Performance over the time period examined was very similar for UPRO/VBLTX and UPRO/VBLTX/VWEHX. The three-fund version had a slightly higher CAGR (21.4% vs. 21.3%), but also a higher MDD (15.6% vs. 14.3%) and slightly worse Sharpe ratio (0.091 vs. 0.097).
I couldn’t be happier with the performance of UPRO/VBLTX going back to 2009 and particularly since I started using the strategy a year and a half ago. It’s a simple idea that I believe will consistently beat the market (and thus the vast majority of funds and strategies) going forward.
Disclosure: I am/we are long UPRO, VBLTX, VWEHX. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: The author used Yahoo! Finance to obtain historical stock prices and used R (including the "quantmod", "stocks", and "dvmisc" packages) to analyze the data and generate figures. Any opinion, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.