My market-neutral strategy
In March of last year, I presented a market-neutral strategy that uses Vanguard's S&P 500 mutual fund, VFINX, and its long-term bond fund, VBLTX. The basic idea is to estimate VBLTX's beta on the fly and adjust fund allocations to achieve a portfolio beta of zero.
The purpose of this article is to look at my strategy's prospective performance since publication, covering the time period March 22, 2016, to May 5, 2017.
Here is the strategy:
- On the first day, calculate VBLTX's beta, β, based on the previous 50 daily gains. The net beta of a VFINX/VBLTX portfolio with c allocated to VFINX is c (1) + (1 - c) (β), so the VFINX allocation necessary for net zero beta is given by -β / (1 - β).
- If β is negative, zero portfolio beta is achievable. Set fund allocations accordingly.
- If β is positive, zero portfolio beta is not achievable. Hold 100% VBLTX until β turns negative.
- Each day, re-calculate β based on the previous 50 gains, and calculate effective portfolio beta based on current allocations and current β. If holding 100% VBLTX and β has turned negative, or if the effective portfolio beta is outside [-0.15, 0.15], execute a rebalancing trade.
Figure 1 shows growth of $10k for my strategy, Vanguard's market-neutral VMNFX, and the two underlying funds for my strategy. Performance metrics are summarized in Table 1.
|Strategy/fund||Growth (%)||Max drawdown (%)||Sharpe ratio|
Overall, my market-neutral strategy performed reasonably well. It achieved 7.5% better growth than VMNFX and had a much better Sharpe ratio. However, its drawdown was worse.
Looking at Figure 1, my strategy jumped out to a nice lead over VMNFX, but lost it all when VBLTX fell 12% after the US election. After that, it recovered and finished meaningfully ahead of VMNFX.
Notably, a 12% drawdown in VBLTX is almost unheard of. Its maximum drawdown over its entire lifetime, going back to Feb. 1994, is 14.5%. So I'm happy to see my strategy still outperform VMNFX, although certainly some of that outperformance stems from VFINX's 19.5% growth.
Over the 13.5-month period, my strategy required 8 trades.
Beta and fund allocations over time
Of course, you'd like for a strategy that calls itself market neutral to maintain beta close to 0. Figure 2 shows 50-day beta for my strategy and for VMNFX over the test period.
Overall, VMNFX did a somewhat better job keeping beta around 0. The median 50-day beta was slightly better for my strategy (0.01 vs. 0.04), but the range was worse (-0.36 to 0.41 vs. -0.06 to 0.27).
One contributing factor is that when rebalancing is necessary, but VBLTX's beta is positive, my strategy allocates 100% to VBLTX. There are alternative approaches, as I discuss in the next section.
Next, let's look at how the allocation to VFINX varied over time.
The median VFINX allocation was 20.1%, and range was 0% to 42.6%. You can see where the rebalancing trades occurred on the graph.
Some changes going forward
I thought a lot about how to deal with the case where a rebalancing trade is triggered, but VBLTX's beta is positive. VFINX's beta is of course 1, so it's impossible to combine it with another positive-beta fund and achieve zero net beta. In prior articles I mentioned the following three approaches:
- Hold cash until VBLTX's beta turns negative.
- Allocate 100% to VBLTX until VBLTX's beta turns negative.
- Use an inverse S&P fund like SH in place of VFINX, which makes a portfolio beta of zero possible.
I chose (2) because VBLTX typically has negative beta, so even if it's 50-day trailing beta is positive, there's a good chance its beta going forward will be roughly zero, or negative.
Approach (1) is simple and somewhat natural, but it doesn't seem right to hold nearly 100% VBLTX when its beta is very slightly negative, and 0% VBLTX when it becomes slightly positive.
After thinking more about (3), I don't like it, because inverse S&P will tend to decline in value, and you hate to have one of your two funds working against you.
That said, I favor a slight variation on (2) going forward. Rather than 100% VBLTX, I propose combining VBLTX with cash to achieve a portfolio beta of up to 0.1. To illustrate, if VBLTX's beta is 0.05, I'd still allocate 100% to VBLTX, since the portfolio beta is still below 0.1. But if VBLTX's beta is 0.2, I'd allocate 50% to VBLTX and 50% to cash, for a portfolio beta of 0.1. As soon as VBLTX's beta turns negative again, or the effective beta exceeds 0.15, I rebalance.
In closing, I'll graph growth of $10k for my revised strategy and VMNFX since VMNFX's inception in 1998.
My revised strategy drastically outperforms VMNFX (CAGR 6.7% vs. 1.9%, MDD 15.0% vs. 26.0%, Sharpe ratio 0.055 vs. 0.017). While it closely tracks VBLTX, it had 23.4% lower volatility and a better Sharpe ratio (0.055 vs. 0.043).
In general, while I would not expect the revised VFINX/VBLTX strategy to generate 6.7% annual returns going forward, I have high confidence it will outperform VMNFX and similar funds that use pair trading to generate growth regardless of market movement.
Disclosure: I am/we are long VBLTX.
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.