Triple Witching In September

The Stock Trader’s Almanac notes that seasonally for the last few years, Dow has done well on the triple witching day in September. Dow Jones had been up 8 of the last 10 years on the third Friday of the last month of the third quarter.

We can develop a trading strategy based upon this seasonal trend. This seasonal tendency also impacts S&P 500, NASDAQ and Russell 2000.

The original Almanac strategy goes long on the Thursday close before the triple witching day in September and exits on the Friday close. Using ETFs – DIA for Dow, SPY for S&P, QQQ for NASDAQ and IWM for Russell 2K – we get following results:

ETF # of Trades Win % Avg. Ret Avg. Avg. Win Avg. Loss Avg W / Avg L Avg Days Held Annualized
QQQ 14 71% 0.3% 14 47 (67) 0.70 1 105%
DIA 15 67% 0.1% 6 28 (38) 0.75 1 46%
SPY 20 30% (0.3)% (14) 40 (37) 1.08 1 (102)%
IWM 13 69% 0.3% 15 44 (50) 0.87 1 108%

The annualized rate is computed based upon the average return for the average days held.

Clearly, S&P 500 does not show a seasonal trend. NASDAAQ shows a stronger trend and Russell 2000 has higher volatility.

We can improve the performance of this strategy by applying some filters for entry and exit.

Entry Filter: Go long on the Thursday close before the third Friday of September only when the price is below the 5-Day SMA. This filter is applied on DIA, SPY and QQQ.

Exit Filter: Exit on the 3-Day high close. (For DIA, SPY and QQQ.)

For Russell 2k, IWM, a hybrid exit strategy is used.

Entry: Go long on the Thursday close before the triple witching day.

Exit: If IWM was above 5-Day SMA on entry then exit on the triple witching day close. If the price was below 5-day SMA then exit on the 3-Day high close.

The filters reduce the numer of trades but improve the overall result.

ETF # of Trades Win % Avg. Ret Avg. Avg. Win Avg. Loss Avg W / Avg L Avg Days Held Annualized
QQQ 7 100% 1.6% 78 78 3 209%
DIA 8 88% 0.8% 41 62 (102) 0.61 3 90%
SPY 8 75% 0.7% 36 63 (44) 1.42 3 84%
IWM 12 75% 0.5% 26 47 (36) 1.31 2 114%

Here is the list of trades using the modified strategy:

# ETF Entry Date Entry Exit Date Exit Return
1 SPY 16-Sep-93 46.03 22-Sep-93 45.41 (1.3)%
2 DIA 17-Sep-98 78.78 21-Sep-98 79.25 0.6%
3 SPY 17-Sep-98 101.46 22-Sep-98 102.33 0.9%
4 QQQ 16-Sep-99 61.47 17-Sep-99 63.31 3.0%
5 DIA 16-Sep-99 107.53 17-Sep-99 108.38 0.8%
6 SPY 16-Sep-99 131.75 17-Sep-99 133.04 1.0%
7 QQQ 14-Sep-00 93.00 19-Sep-00 93.19 0.2%
8 DIA 14-Sep-00 111.02 22-Sep-00 108.75 (2.0)%
9 IWM 14-Sep-00 53.91 15-Sep-00 52.91 (1.9)%
10 QQQ 20-Sep-01 28.97 24-Sep-01 29.61 2.2%
11 DIA 20-Sep-01 83.65 24-Sep-01 86.20 3.0%
12 SPY 20-Sep-01 98.18 24-Sep-01 100.16 2.0%
13 IWM 20-Sep-01 38.56 24-Sep-01 39.00 1.1%
14 QQQ 19-Sep-02 21.58 25-Sep-02 21.87 1.3%
15 DIA 19-Sep-02 79.64 26-Sep-02 79.81 0.2%
16 SPY 19-Sep-02 84.25 25-Sep-02 83.90 (0.4)%
17 IWM 19-Sep-02 36.40 25-Sep-02 36.45 0.1%
18 IWM 18-Sep-03 51.86 19-Sep-03 51.80 (0.1)%
19 QQQ 16-Sep-04 35.32 17-Sep-04 35.43 0.3%
20 DIA 16-Sep-04 102.75 17-Sep-04 103.01 0.3%
21 SPY 16-Sep-04 112.54 17-Sep-04 112.55 0.0%
22 IWM 16-Sep-04 57.32 17-Sep-04 57.22 (0.2)%
23 QQQ 15-Sep-05 39.16 16-Sep-05 39.40 0.6%
24 DIA 15-Sep-05 105.66 16-Sep-05 106.23 0.5%
25 SPY 15-Sep-05 122.49 16-Sep-05 122.84 0.3%
26 IWM 15-Sep-05 66.17 16-Sep-05 66.94 1.2%
27 IWM 14-Sep-06 72.49 15-Sep-06 72.66 0.2%
28 IWM 20-Sep-07 80.79 21-Sep-07 80.96 0.2%
29 QQQ 18-Sep-08 41.57 19-Sep-08 42.90 3.2%
30 DIA 18-Sep-08 109.97 19-Sep-08 113.57 3.3%
31 SPY 18-Sep-08 120.07 19-Sep-08 124.12 3.4%
32 IWM 18-Sep-08 71.80 19-Sep-08 75.00 4.5%
33 IWM 17-Sep-09 61.66 18-Sep-09 61.91 0.4%
34 IWM 16-Sep-10 64.94 17-Sep-10 65.21 0.4%
35 IWM 15-Sep-11 71.41 16-Sep-11 71.52 0.2%

The overall win percentage of the strategy is 83%.
The average return in 0.8%.
The average win to average loss ratio is 1.23.
The average hold period is 3 days.
The annualized return in 118.3%.

The strategy had only three losing years – 1993, a loss of (1.3)%; 2000, a loss of (1.2)%; 2003, a loss of (0.1)%.

Instead of 1X ETFs, one can also use ultra ETFs (2X or 3X). Using 2X ETFs, DDM for Dow, SSO for S&P, QLD for NASDAQ and UWM for Russell 2k, will double the return but will also increase the risk.

 

The Fine Print: Before employing this strategy in your live account please understand the rationale behind this seasonality pattern. Then run some back tests for the trade start-date and holding periods. Make sure that the risk level, max draw down, win ratio and average loss falls within your comfort zone and that you can withstand the draw down and the associated risk. Also read our disclaimer.

 

Print Friendly, PDF & Email