Armored Portfolios: The Hedged Portfolio Method

An Armored Portfolio

Picture your portfolio like the piggy bank in the image below – protected by armor, but still collecting coins. That illustrates the goal of the hedged portfolio method: To generate competitive returns while strictly limiting risk.

The Hedged Portfolio Method

The goal of the hedged portfolio method is to generate competitive returns for an investor while strictly limiting his risk. For example, an investor unwilling to risk more than a 15% drawdown over the next six months (i.e. an investor whose "threshold," in our parlance, was 15%) could invest in a hedged portfolio structured to maximize his expected return while insuring that, in the worst-case scenario (if each of his underlying securities went to zero), his portfolio would decline no more than 15% over that time period. In a series of rigorous backtests we determined that the hedged portfolio method, using our security selection process, could achieve competitive returns while strictly limiting risk. Below, for example, is a chart showing the performance of a series of 21% threshold hedged portfolios from Jan. 2, 2003 to April 30, 2014, versus the performance of the SPDR S&P 500 tracking ETF SPY over the same period.

In broad strokes, this is the hedged portfolio method:

There are a few additional fine-tuning steps we employ to minimize hedging costs and maximize potential returns, but that's the basic idea.

A Sample Hedged Portfolio

Below is an example of what an "armored", or hedged portfolio looks like. This one was created as of the market close on Wednesday, October 21st, and was designed for someone with $200,000 to invest who is unwilling to risk a drawdown of more than 22% over the next six months:

At the portfolio level, the worst-case scenario for an investor over the next six months, the "max drawdown," was 20.7%; the best-case scenario, how the portfolio would perform if each security met or exceeded its potential return, the "net potential return," was 16.21%; and the "expected return," a more likely scenario based on the historic relationship between our calculated potential returns and actual returns, was 6.29%.

At the position level, that portfolio includes, as primary securities, some of the stocks with highest potential returns in Portfolio Armor's universe as of that date: Activision Blizzard ATVI, Salesforce.com CRM,Global Payments GPN, and Manhattan Associates MANH. We aim to add equal dollar amounts of each primary security but round down the amounts as we go to make sure we have round lots of each (in order to reduce hedging costs). In a fine-tuning step, we replaced most of the rounded-down cash with Regeneron REGN, which was added as a "cash substitute": that's a security collared with a tight cap (1% or the current yield on a leading money market fund, whichever is higher) in an attempt to capture a better-than-cash return while keeping the investor's downside limited according to his specifications.

The Importance Of Calculating Potential Returns

It's clear that the first step in creating a hedged portfolio, calculating potential returns, is crucial, for two reasons: first, if you can't select securities with the potential to generate alpha, your hedged portfolio returns will lag; second, in order to know if it's worth the cost of hedging the securities, you need to have an idea of how accurate your potential returns are. So, before backtesting the hedged portfolio method as a whole, first we backtested our security selection process.

Backtesting Our Security Selection Process

Every trading day, Portfolio Armor generates high-end estimates of how more than 3,000 stocks and exchange-traded products will perform over approximately the next six months. These estimates are based on analysis of historical returns as well as option market sentiment, which provides a forward-looking element. Essentially, we start with the assumption that a security's returns over the next six months will start to revert from its performance over the previous six months to its long-term average six-month performance. We then test that assumption by looking at the prices of options on the security expiring approximately six months out. We call this high-end estimate a security's potential return.

Essentially, it's how the security might perform over the next six months in a bullish scenario. We backtested this method by running our analysis every trading day from Jan. 2, 2003 to Oct. 31, 2013 and then looking at the actual returns of the securities with the highest potential returns on our daily scans over the next six months. Over that 11-year period, we conducted 25,412 comparisons of our calculated potential returns to actual returns, an average of 9.4 top-ranked securities each trading day. The average potential return we calculated was 22.4%. The average actual return over the next six months, unhedged, was 6.84%. Since the average actual return was 0.3x the average potential return, we use that 0.3x multiple to derive expected returns from our potential returns. While a potential return represents a bullish upside, an expected return is the more likely result.

A subset of our top-ranked securities – 5,202 of them, or about 20% of them – had an even higher average actual return: 9.35%. All of our top-ranked securities were hedgeable with optimal collars, but the securities in this subset were also hedgeable with optimal puts (we call these AHP securities, for short). There aren't always AHP securities available, but when there are, our portfolio construction algorithm gives preference to them proportional to their higher average returns in our tests. Specifically, we increase their potential returns by 37%, since 9.35% is 1.37x 6.84%.

The security returns mentioned above were unhedged; we also tested gross returns (i.e. not net of hedging costs) of our security selection method while hedging against greater-than-9% declines. When doing so with optimal puts, the average gross return was 12.08% over six months. The average gross return for the same securities when hedged with optimal collars capped at their potential returns was about half as much, 6.25%. This illustrates to what extent the average actual returns of the securities hedged with optimal puts were driven by outliers –securities that appreciated beyond our calculated potential returns. We adjust for the impact of potential outliers during the portfolio construction process, by only hedging with optimal collars when the net potential return is greater than 1.93x that of the same security when hedged with an optimal put (since 12.08/6.25 = 1.93).

Backtesting The Hedged Portfolio Method

To backtest the hedged portfolio method, we started searching for a hedged portfolio at each threshold on Jan. 2, 2003. To avoid survivorship bias, we started with the complete universe of hedgeable U.S. securities at each point in time during the backtest. That universe included stocks that didn't survive to the present, such as Lehman Brothers, Bear Stearns, etc. Hedged portfolios were run for six months, or until all positions had been exited, whichever came first, and then the ending dollar amount of the first portfolio was used as the starting dollar amount of the second sequential portfolio, and so on, until the end of our data series on April 30, 2014.

When there were no securities available with positive net potential returns (usually, because hedging costs were too high), the last portfolio ending dollar amount was held as cash until the start of the next hedged portfolio. During those periods, we treated the cash as if it were held in a noninterest-bearing account, so the dollar amount invested remains constant until the start of the next hedged portfolio. Within hedged portfolios, residual cash positions were treated as if they were invested in a money market fund, earning the yield prevailing at the time (during much of this time period, that yield was negligible).

Within hedged portfolios, positions in underlying securities were entered at their unadjusted closing prices, with trading commissions of $7.95 deducted. To facilitate performance tracking, the dollar amounts allocated to these underlying securities were converted to the equivalent numbers of each security at its adjusted closing price on the start date of the portfolio. Underlying security positions were exited at their adjusted closing prices, with trading commissions of $7.95 deducted.

During the simulation, to be conservative, puts were purchased at the closing ask price, and calls were sold at the closing bid price; options were exited at the midpoint of the closing bid-ask spread or their intrinsic value, whichever was higher ("last" prices weren't used because in many cases with options, the last price might be weeks old). Each time options positions were entered or exited, a trading fee of $7.95 plus 75 cents per option contract was deducted (the trading fees were the ones charged by Fidelity at the time).

Results Of The Hedged Portfolio Backtests

In general, the higher the threshold was, the higher the Compound Average Growth Rate (CAGR) was. CAGRs ranged from 3.26% at a 2% threshold, to 11.06% at a 22% threshold.

Results at those thresholds and three other thresholds in between, and interactive graphs showing hedged portfolio holdings at each threshold during the backtesting period, can be found at this link (incidentally, if you go to the 2% threshold backtest interactive graph at that link, click to show positions, and then move your mouse pointer over the graph to June 22, 2006, you'll see Lehman Brothers (LEH) was a holding in a hedged portfolio then).

Image Credit: Public Domain