Scalper1 News
Summary How to build a robust long side. Which ETF on the short side. How to make it IRA-compliant. The first article of the series described the investor profile to hold a market neutral portfolio, some characteristics of this investing style. The second one explained the benefit of sector diversification, with examples. This one simulates solutions for the hedging position with various ETFs: leveraged and non leveraged, inverse and regular. People implementing an equity market neutral strategy usually have two balanced sets of individual stocks on both sides (long and short). Before going to the point, I want to come back on the reason why I prefer a single index ETF position on the short side. My opinion is that ‘Market Neutral’ is for risk-averse investors. Therefore it is also better to avoid a potentially unlimited risk that is not related to the market: being trapped in a short squeeze. People who think that this risk is limited to penny stocks and small caps have a short memory, or don’t know some cases. My preferred example is the ‘mother of all short squeezes’ that happened in 2008 when Volkswagen AG became briefly the highest capitalization in the world after its share price was multiplied by five in 2 days. Then it fell back to its initial level even more quickly. In the interval, investors and traders on the short side covered their positions at any price with huge losses, in panic or forced by their brokers. Whatever the reason (in this case a corner engineered by a major shareholder), and the consequences (at least a suicide has been attributed to that), I prefer avoiding by design this kind of event. Even absorbed in a diversified portfolio, such a shock hurts and may trigger a margin call for leveraged investors. On the long side… The quantitative models used for the long side of my real market neutral portfolio will not be disclosed here. However, I want to share some of its characteristics that may be reused by readers in another context. The portfolio is based on 5 different models: 2 with defensive stocks, 2 with cyclical stocks, 1 based on growth and valuation with no sector limitation. All models are based on rankings using fundamental factors. 24 stocks are selected: 14 in the S&P 500 index, 5 in the Russell 1000 index, 5 in the Russell 3000 index. The number of stocks has been chosen to limit the idiosyncratic risk. The sector diversification pattern should help beat the hedge in most phases of the market cycle. The diversification in rankings across models should limit the risk of over-optimization. The focus on large capitalizations is a choice of comfort (for myself) and ethic (for subscribers). Russell 3000 stocks are filtered on their average dollar daily volume. The portfolio is rebalanced weekly, but backtests show that a bi-weekly rebalancing doesn’t hurt the long-term performance. However, the hedge should always be rebalanced weekly. The next chart shows the simulation of this 24-stock portfolio (long side only) since 1999, with a 0.3% transaction cost and a 2-week rebalancing: (click to enlarge) Past performance, real or simulated, is never a guarantee of future returns. However, for a diversified portfolio like this one, it gives some clues about the robustness. Especially when robustness has been integrated from the design process, not just as the result of backtest optimization. On the short side… Some readers will be scared if I tell them abruptly that I use a leveraged 3x inverse ETF. Most people who are afraid of leveraged ETFs don’t really understand where their ‘decay’ comes from. If you exclude the management fee (under 1% a year), the decay has two names: roll-over cost and beta-slippage. The holdings of leveraged S&P 500 ETFs (inverse and regular) are swaps for the biggest part, and futures in second position. Rollover costs are close to zero for such contracts on the S&P 500. For beta slippage, some of my old articles have already explained what it is , and why I don’t fear it on S&P 500 leveraged ETFs. In short: most leveraged ETFs are harmful as long term holdings, but not all of them. The next table is a summary of backtests for the portfolio with different hedges, period 1/1/1999 to 11/29/2014 (weekly rebalancing). The ETF used are the ProShares Short S&P 500 ETF ( SH), the ProShares UltraShort S&P 500 ETF ( SDS), the ProShares UltraPro Short S&P 500 ETF ( SPXU) and the ProShares UltraPro S&P 500 ETF ( UPRO). For most cases it shows the performance without leverage, and with a leverage factor corresponding to holding the stocks on capital and the hedge on margin. Price data are synthetic before the inception dates (calculated by data provider). Hedge Leverage An.Ret. (%) DD (%) DL (weeks) K (%) No no 28 36 103 24 SH no 10 10 54 25 SH 2 23 23 54 25 SDS no 14 12 54 28 SDS 1.5 21 17 54 28 SPXU no 15 9 54 30 SPXU 1.33 21 11 54 30 UPRO (short) no 16 8 51 33 UPRO (short) 1.33 22 10 51 33 SPXU 50% no 20 17 51 34 SPXU 50% 1.167 24 19 51 34 SPXU 75% no 17 11 49 33 SPXU 75% 1.25 23 14 50 33 SPXU Timed no 25 15 50 34 SPXU HalfTimed no 20 10 48 36 SPXU HalfTimed 1.33 28 13 49 36 An.Ret.: annualized return DD: max drawdown depth on rebalancing (it may be deeper intra-week) DL: max drawdown length K: Kelly criterion of the weekly game, an indicator of probabilistic robustness The ‘Timed’ version uses a signal based on the 3-month momentum of the aggregate S&P 500 EPS and the U.S. unemployment rate. ‘Half Timed’ means that 50% of the hedging position is permanent, the other 50% is timed. Among the 100% market neutral versions, shorting UPRO looks better at first sight… but it is not after taking into account the borrowing rate (4.48% last time I had a look at UPRO properties in InteractiveBrokers platform). As it represents 25% of the total portfolio, the drag on the portfolio annual return is about 1%, which gives the same performance as with SPXU. I prefer buying SPXU and eliminating the inherent risk of short selling. Moreover, U.S. tax-payers can implement this kind of strategy in an IRA account if they use SPXU. Such a portfolio can be traded without leverage, but cash and IRA accounts usually have a 3-day settlement period. It is recommended trading at a broker offering a limited margin IRA feature waiving the settlement period and the risk of free-riding. It seems that Interactive Brokers and TD Ameritrade do that (and maybe others). Inform yourself carefully. Data and charts: Portfolio123 Additional disclosure: Long SPXU as a hedge. Scalper1 News
Scalper1 News