| Author: rgearyiii |
| Subject: Re: How to go with SSO |
| Date: 1/16/2014 |
| Recommendations: 25 |
You can actually create an approximation of a 2x leveraged S&P500 ETF back to 1974. First, here`s a backtest of the market cap-weighted S&P 500 stocks (including their dividends), which I`ll call "gSPY", back to 1974: 19740104-20140114 19930129-20140114 (1) (2) (3) (4) gSPY ^GSPC gSPY SPY CAGR: 11.07 7.60 CAGR: 9.51 9.11 TR: 6508.89 1759.33 TR: 565.85 516.76 GSD(20): 17.32 17.41 GSD(20): 17.10 17.21 DD(20): 11.22 11.72 DD(20): 11.99 12.14 MDD: -54.62 -56.78 MDD: -54.62 -55.19 UI(20): 13.15 15.54 UI(20): 16.04 16.63 Sharpe(20): 0.43 0.23 Sharpe(20): 0.49 0.46 Beta(20): 1.00 1.00 Beta(20): 0.99 1.00 TI(20): 6.87 3.66 TI(20): 7.62 7.24 AT: 0.46 0.00 AT: 0.65 0.00(1) http://gtr1.backtest.org/2012/?lf-1lp-1h1::sp500.a:et1:mcp:g... (2) http://gtr1.backtest.org/2012/?h1::gprc%280%29gt0%7bU:%7b!GS... (3) http://gtr1.backtest.org/2012/?s19930129lf-1lp-1h1::sp500.a:... (4) http://gtr1.backtest.org/2012/?h1::gprc%280%29gt0%7bU:%7bSPY... I`m certainly relieved to not see the S&P 500 torpedoed! gSPY obviously beats ^GSPC because of dividends, but I don`t have a definitive asnwer right now as to why gSPY beats SPY by a little. Next, gSPY can be leveraged 2x (which I`ll call gSSO) by blending 2 parts gSPY and -1 part Cash (we`ll leave the margin premium at 0%): 19740104-20140114 20060621-20140114 (5) (6) (7) gSSO gSSO SSO CAGR: 11.34 CAGR: 6.68 6.68 TR: 7186.76 TR: 62.74 62.74 GSD(20): 38.76 GSD(20): 55.06 55.33 DD(20): 26.60 DD(20): 43.55 43.90 MDD: -87.78 MDD: -84.08 -84.66 UI(20): 33.57 UI(20): 41.49 41.99 Sharpe(20): 0.35 Sharpe(20): 0.36 0.36 Beta(20): 2.01 Beta(20): 1.95 1.95 TI(20): 5.52 TI(20): 7.35 7.37 AT: 2.81 AT: 3.22 0.00 (5) http://gtr1.backtest.org/2012/blend.cgi?p2.235::%7b2:lf-1lp-... (6) http://gtr1.backtest.org/2012/blend.cgi?s20060621p2.235::%7b... (7) http://gtr1.backtest.org/2012/?h1::gprc%280%29gt0%7bU:%7bSSO... The match between gSSO and SSO looks impressive, but that`s mainly because I tuned the margin premium (the percentage points of interest above the Cash yield paid on short positions, including Cash itself) to get CAGRs to match, which happened at 2.235%. I doubt that State Street Global Advisors pays that much in margin premium, but since gSPY beat SSO by 0.4%, we`d expect gSSO to beat SSO by 0.8% without some sort of penalty. As for why gSSO does not have much of a higher CAGR than gSPY, that`s easily explained by high margin interest rates in the 1970s and 1980s. Now to backtest klouche`s timing strategy with gSSO, simply run link (5), click "Add blend to new universe", insert "Cash" as the second screen in the Universe Builder, click Build Universe, and then click "Build a screen in new universe" (it`s actually totoally uncesseary to add Cash to the universe in this case). Then add klouche`s steps to the form, set the holding period to 5 market days like he used, and run the backtest: (8) gSSO w/ klouche timing, 19740104-20140114 Avg Min Max SD CAGR: 12.20 11.31 12.79 0.49 TR: 9944.17 7113.31 12087.56 1622.85 GSD(20): 24.93 24.21 25.86 0.62 DD(20): 16.61 15.89 17.81 0.70 MDD: -60.27 -65.76 -57.46 3.54 UI(20): 19.04 15.77 23.09 3.17 Sharpe(20): 0.41 0.38 0.43 0.02 Beta(20): 0.95 0.93 0.97 0.02 TI(20): 9.47 8.72 9.93 0.42 AT: 3.82 3.73 3.94 0.08(8) http://gtr1.backtest.org/2012/?h5::tr%281,105%29gt1:tr%281,1... (9) (10) gSSO w/ klouche timing, 20060621-20140114 SSO w/ klouche timing, 20060621-20140114 Avg Min Max SD Avg Min Max SD CAGR: 17.17 14.12 20.83 2.20 CAGR: 17.61 15.28 20.65 1.93 TR: 232.53 170.22 315.38 48.07 TR: 241.15 191.61 310.68 42.71 GSD(20): 20.26 19.74 21.27 0.57 GSD(20): 20.51 19.90 21.37 0.55 DD(20): 11.81 11.06 12.24 0.43 DD(20): 11.87 11.53 12.38 0.32 MDD: -26.60 -29.38 -23.70 1.93 MDD: -26.65 -29.56 -24.88 1.65 UI(20): 10.07 8.20 12.02 1.34 UI(20): 9.91 8.36 11.88 1.20 Sharpe(20): 0.88 0.76 1.06 0.10 Sharpe(20): 0.90 0.81 1.03 0.08 Beta(20): 0.41 0.33 0.48 0.05 Beta(20): 0.43 0.34 0.48 0.05 TI(20): 40.21 34.82 49.40 5.13 TI(20): 39.98 34.23 43.68 3.66 AT: 3.36 3.16 3.73 0.20 AT: 1.97 1.73 2.26 0.18(9) http://gtr1.backtest.org/2012/?s20060621h5::tr%281,105%29gt1... (10) http://gtr1.backtest.org/2012/?s20060621h5::tr%281,105%29gt1... The rough match between (9) and (10) lends some credibility to the use of gSSO as a substitute for SSO in a full-length backtest of the timing system. But unfortunately, gSSO with klouche`s timing (8) loses to gSPY with no timing (1) from 1974 to the present by both Sharpe ratio and maximum drawdown. It`s probably extremely difficult for any strategy involving leverage to look good in the 1970s and 1980s due to high interest rates. But it`s probably also a bit of a lesson in the dangers of curve-fitting and short backtests of timing strategies. Robbie Geary |