Friday, August 17, 2007

Volatility Breakouts

Volatility Breakouts

Rules: if c > 12 week ema of close, then buy @ 1 week high + 1 tick stop.

exit long at Volatility_Stop stop

Volatility_Stop = c[1] - average true range of past 4 weeks * ATR factor (ATR in tables below). Caveat: Volatility_Stop is only increased when long, never decreased.

Dow Jones Industrial Average, 2/88 - 12/98:

ATR

Net Profit

PF

n

%

Profit/trade:

.50

2153.41

1.58

98*

42

21.97

1.00

2455.99

1.73

64

47

38.37

1.50

4082.09

3.60

34

47

120.06

2.00

3555.30

2.80

25

44

142.21

2.50

4603.28*

6.08*

16

63*

287.70*

Note here that the wider the swing allowed, the more netprofitability and profit/trade. However, 1.50 seemed superior to 2, although it generated less per trade.

For comparison, the results of the 1 week/ 3 week 12 week EMA system are shown below (buy stop at high of week stop, sell stop at trailing 3 week low, but only go long if the close is > 12 week EMA). Increasing the ATR factor also decreases the number of trades that need to be made.

Net Profit

PF

n

%

Profit/trade:

2713.53

2.01

50

38

54.27

Note that any ATR factor above 1.0 is superior to the breakout system, and all are superior in terms of percent of trades profitable. Anything above 1 ATR is superior in terms of profit/trade.

Note that the MACD (12/26/6) system (buy stop at high of week if MACD > xaverage of MACD, sell stop at low if MACD <>

Net Profit

PF

n

%

Profit/trade:

4232.17

12.25

17

71

248.95

Best Buy 1/5/90 - 11/27/98:

MACD 12 / 26 /6:

Net Profit

PF

n

%

Profit/trade:

26.69

3.87

14

64

1.91

1 week/ 3 week breakout with 12 week EMA filter:

50.95

6.15

25

56

2.04

Volatility System:

ATM

Net Profit

PF

n

%

Profit/trade:

.50

40.16

2.95

55

49

.73

1.00

33.02

2.26

36

50

.92

1.50

32.30

2.39

23

52

1.40

2.00

29.35

2.51

16

56

1.83

2.50

42.87*

7.64*

12

58*

3.57*

3.00

34.71

5.28

12

58

2.89

3.50

7.08

1.96

9

44

.79

4.00

7.92

2.68

8

50

.99

4.50

5.20

1.93

8

50

.65

5.00

6.55

2.01

6

33

1.09

5.50

5.51

2.03

5

40

1.10

6.00

3.81

1.63

5

40

.76

6.50

2.11

1.31

5

40

.42

7.00

.41

1.06

5

40

0.08

7.50

3.74

2.19

4

50

.93

8.00

2.46

1.70

4

50

.61

8.50

1.24

1.32

4

50

.31

9.00

.28

1.07

4

50

0.07

9.50

-.62

.86

4

50

-.16

10.00

.85

1.17

3

33

.28

Note that increasing the ATM factor increases the profit factor, but that net profitability actually drops until ATM increases above 2, then drops off after 2.5. Percent profitability rises in direct correlation with ATM factor and the profit/trade also increases. Anything in the 2-3 range is superior to the breakout system in terms of profit/trade, but the breakout system is superior to all volatility systems in terms of total profitability.

Micron Technology, 6/8/84 - 11/27/98:

ATM

Net Profit

Profit Factor:

n

%

Profit/trade:

.50

40.60

2.02

84

49

.48

1.00

52.10*

2.44

50

50

1.04

1.50

50.53

3.21

33

61*

1.53

2.00

51.25

2.95

24

58

2.14

2.50

42.74

2.70

20

60

2.14

3.00

47.81

5.82*

15

60

3.19

3.50

44.65

4.66

11

45

4.06*

4.00

34.35

3.36

11

45

3.12

4.50

25.22

2.52

10

40

2.52

5.00

15.27

1.79

10

30

1.53

5.50

20.73

3.58

7

29

2.96

6.00

14.64

2.52

5

40

2.93

6.50

8.41

1.71

5

40

1.68

7.00

4.44

1.31

5

40

.89

7.50

11.27

3.36

3

67

3.76

8.00

-4.65

0.06

2

50

-2.33

8.50

-5.00

0.03

2

50

-2.50

9.00

-5.34

.00

2

0

-2.67

9.50

-5.69

.00

2

0

-2.84

10.00

-6.03

.00

2

0

-3.01

Note that anything in the 1-3 ATM range was good as far as net profitability; the most profit/trade was generated with 3.5 ATM and the biggest profit factor was with 3 ATM.

Just to insure the 12 week EMA screen wasn't filtering out good trades, I took it out to see how the system would perform "raw":

ATM

Net Profit

Profit Factor:

n

%

Profit/trade:

.50

52.03

1.82

136

43

.38

1.00

51.22

1.79

87

46

.59

1.50

36.76

1.60

57

47

.64

2.00

26.46

1.39

39

44

.68

2.50

21.56

1.35

34

44

.63

3.00

14.67

1.26

23

48

.64

3.50

27.10

1.68

15

40

1.81

4.00

33.94

2.35

14

43

2.42

4.50

24.10

1.87

13

38

1.85

5.00

15.00

1.51

12

42

1.25

5.50

12.22

1.51

9

56

1.36

6.00

22.29

4.33

6

50

3.72

6.50

16.47

4.07

5

60

3.29

7.00

12.62

3.22

5

60

2.52

7.50

11.37

3.75

3

33

3.79

8.00

-4.48

.00

2

0

-2.24

8.50

-4.82

.00

2

0

-2.41

9.00

-5.17

.00

2

0

-2.58

9.50

-5.51

.00

2

0

-2.76

10.00

-4.61

0.05

2

50

-2.30

Note here that the 12 week EMA filter does a good job of filtering out poor trades and that the net profitability drops off sharply as the ATM rises above .5. The profitability per trade is horrible, as is the profit factor. Conclusion: the EMA works as a nice filter.

Breakout System (1w/3w) + 12 week EMA:

Net Profit

PF:

n

%

Profit/trade:

71.17

3.75

51

51

1.40

Note that although the breakout system manages to capture more net profits, its profitability per trade is less than the volatility system. It took 51 trades to capture the profits that the ATM system with a factor of 2 caught in 24. Taking commisions into account (which this testing didn't) not to mention psychological wear and tear, and I'm sure the volatility system is superior.

MACD 12/26/6:

Net Profit

PF:

n

%

Profit/trade:

17.89

1.42

30

37

.60

The MACD system did very poorly, both in terms of net profitability, profit factor, and profits per trade. The problem with the MACD is that it is a derivative of a derivative of price. It therefore lags badly and is not in synch with the market. It must be optimized, which always makes me suspicious, whereas the breakout system and volatility system both seem rather robust.

Fidelity Select Electronics:

Net Profit

PF

n

%

Profit/trade:

MACD

15.45

2.21

13

46

1.19

Breakout

23.03

3.68

29

52*

.79

Volatility

24.98*

7.31*

16

50

1.56*

ATM = 2

Note the volatility system is superior in all counts except for percent profitability. It not only earns more net profitability, but does so with fewer trades.

An optimization report follows (the ATM of 2 was arbitrarily chosen above, not optimized):

ATM

Net Profit

PF

n

%

Profit/trade:

.50

20.62

3.11

44

55

.47

1.00

20.61

3.32

36

50

.57

1.50

22.90

3.88

25

52

.92

2.00

24.98*

7.31*

16

50

1.56

2.50

23.96

6.64

11

55

2.18*

3.00

19.65

4.23

11

55

1.79

3.50

16.89

3.16

8

63

2.11

4.00

13.96

2.62

8

63

1.75

4.50

13.86

2.18

7

57

1.98

5.00

11.33

1.87

7

57

1.62

5.50

17.95

3.20

5

60

3.59

6.00

18.00

3.17

4

50

4.50

6.50

17.10

3.04

4

50

4.28

7.00

23.97

100.00

2

100

11.99

7.50

22.34

100.00

2

100

11.17

8.00

21.78

100.00

2

100

10.89

8.50

21.22

100.00

2

100

10.61

9.00

20.66

100.00

2

100

10.33

9.50

19.95

100.00

2

100

9.98

10.00

21.32

100.00

1

100

21.32

2 ATM seems a reasonable number to use for a trailing stop. Of course, a variation could be to sell half at 2, another half at 2.5, etc., but this needs to be tested.

Fidelity Select Home Finance, 7/90 - 11/98:

ATM

Net Profit

PF

n

%

Profit/trade:

Breakout 1w/3w

34.97

4.27

34

65

1.03

MACD

34.05

15.55

14

79

2.43

Volatility Breakout

36.86*

5.86

23

78

1.60

ATF = 2.00

Once again, the volatility breakout system is superior in terms of net profitability and profit/trade compared to the raw breakout system. Note that although MACD shows a higher profit per trade, MACD does not capture as much total profit, nor does it offer a meaningful stop at the beginning of the trade. Also, FSVLX is a very trend-worthy fund, so the MACD's usefulness is perhaps overstated here.

An optimization table follows (note that the results are slightly different here since a different Max bars back setting was used).

ATM

Net Profit

PF

n

%

Profit/trade:

.50

44.41

6.55

57

58

.78

1.00

41.25

7.05

41

68

1.01

1.50

43.15

10.39

28

71

1.54

2.00

38.30

6.05

23

78

1.67

2.50

40.09

9.22

18

83

2.23

3.00

37.97

8.32

13

77

2.92

3.50

39.69

16.91

10

80

3.97

4.00

38.37

13.87

10

80

3.84

4.50

38.55

12.62

8

88

4.82

5.00

48.80

100.00

4

100

12.20

5.50

48.28

100.00

3

100

16.09

6.00

47.84

100.00

3

100

15.95

6.50

47.31

100.00

3

100

15.77

7.00

46.62

100.00

3

100

15.54

7.50

46.03

100.00

3

100

15.34

8.00

45.39

100.00

3

100

15.13

8.50

45.09

100.00

3

100

15.03

9.00

44.61

100.00

3

100

14.87

9.50

44.12

100.00

3

100

14.71

10.00

44.35

100.00

2

100

22.17

Note that the 2-3 ATM factor area is very respectable and profitable.

TelMex (TMX) 5/17/91 - 11/27/98:

Here the 12 week EMA filter hurt results badly, so I removed it in testing.

Net Profit

PF

n

%

Profit/trade:

Breakout 1w/ 3w

29.67

1.43

44

48

.67

MACD 12/26/6

20.16

1.87

14

43

1.44

Volatility Breakout

26.53

1.68

23

52

1.15

ATM = 2

Note that although the raw breakout system has a higher net profitability, it generates this with twice as many trades (44 v. 23) and has a much lower profit/trade (.67 v. 1.15). After commissions, the volatility breakout system would most likely be superior.

The MACD system performed well here, with the highest profit factor and profit/trade, but left about a third of the profits on the table that the other systems captured.

An optimization table follows:

ATM

Net Profit

PF

n

%

Profit/trade:

.50

26.13

1.34

73

38

.36

1.00

17.92

1.21

52

38

.34

1.50

14.28

1.22

32

44

.45

2.00

26.53

1.68

23

52

1.15

2.50

32.69

2.09

18

44

1.82

3.00

26.48

1.89

13

46

2.04

3.50

15.69

1.47

12

50

1.31

4.00

10.38

1.33

11

55

.94

4.50

-2.52

.93

11

45

-.23

5.00

-9.12

.80

10

40

-.91

5.50

-1.52

.95

8

50

-.19

6.00

12.47

1.67

5

60

2.49

6.50

26.97

18.80

4

75

6.74

7.00

22.41

6.74

4

50

5.60

7.50

19.59

5.01

4

50

4.90

8.00

15.03

2.97

4

50

3.76

8.50

18.04

7.48

3

67

6.01

9.00

15.43

5.33

3

67

5.14

9.50

12.82

3.95

3

67

4.27

10.00

10.21

2.99

3

67

3.40

Note that the 2-3 ATM region performed well both on a net profit and profit per trade basis.

General Motors 2/26/93 - 11/27/98

ATM

Net Profit

PF

n

%

Profit/trade:

Profit/trade:

Breakout 1w/ 3w

11.32

1.22

34

38

.33

Plus 12 week EMA:

15.96

1.57

20

50

.80

MACD 12/26/6:

24.68

11.64

7

71

3.53

Volatility Bko 2

7.62

1.21

16

50

.48

Plus 12 week EMA:

9.95

1.62

11

64

.90

Both the breakout system and volatility system were improved by the 12 week EMA filter, but both underperformed MACD on a net profit basis and on a per trade basis. However, the volatility system was superior to the breakout system on a profit/trade basis, both with (.90 v. .80 per trade) and without (.48 v. .33) the 12 week EMA filter.

An optimization table for the Volatility breakout + moving average system follows:

ATM

Net Profit

PF

n

%

Profit/trade:

.50

12.51

1.38

31

39

.40

1.00

14.48

1.54

21

52

.69

1.50

7.60

1.27

16

56

.48

2.00

9.95

1.62

11

64

.90

2.50

17.66

5.59

7

86

2.52

3.00

11.33

3.14

6

83

1.89

3.50

6.16

1.61

5

60

1.23

4.00

5.37

1.52

4

75

1.34

4.50

11.32

100.00

3

100

3.77

5.00

7.29

11.48

3

67

2.43

5.50

10.18

5.57

2

50

5.09

6.00

7.21

2.99

2

50

3.60

6.50

4.54

1.96

2

50

2.27

7.00

1.88

1.32

2

50

.94

7.50

-.79

.89

2

50

-.39

8.00

-8.00

.00

1

0

-8.00

8.50

-9.10

.00

1

0

-9.10

9.00

-10.19

.00

1

0

-10.19

9.50

.00

100.00

0

0

.00

Note that these results are somewhat unstable, but the 2-3 ATM region continues to be one of the most profitable.

Dell, 6/10/94 - 11/27/98

ATM

Net Profit

PF

n

%

Profit/trade:

.50

36.59

3.89

33

58

1.11

1.00

37.16

5.45

19

74

1.96

1.50

31.93

2.97

11

45

2.90

2.00

42.40*

10.35

7

71

6.06

2.50

41.04

100.00

4

100

10.26

3.00

36.46

100.00

3

100

12.15*

3.50

18.04

100.00

2

100

9.02

4.00

17.10

100.00

2

100

8.55

4.50

16.17

100.00

2

100

8.09

5.00

.22

100.00

1

100

.22

5.50

.00

100.00

0

0

.00

Note that Dell performs extraordinarily well around the 2-3 area with a maximum net profit at 2 ATM and maximum per trade profit at 3 ATM. 2 ATM gave outstanding results.

Moving Average High Low (Rolling Stock)

Rules: buy limit @ EMA of lows of lomalen (12w).

sell limit @ EMA of highs of himalen (12w).

1. Chevron:

Note that this system performs just as well as buy and hold with only a fraction of the exposure. An extraordinary 92% of trades are profitable with the average losing trade much smaller than the average trade.

MovAvg Hi/Lo Rolling CHEVRON CP n-Weekly 04/12/91 - 12/25/98

Performance Summary: All Trades

Total net profit

$ 59.76

Open position P/L

$ 0.00

Gross profit

$ 62.24

Gross loss

$ -2.48

Total # of trades

32

Percent profitable

91%

Number winning trades

29

Number losing trades

3

Largest winning trade

$ 5.16

Largest losing trade

$ -1.53

Average winning trade

$ 2.15

Average losing trade

$ -0.83

Ratio avg win/avg loss

2.60

Avg trade(win & loss)

$ 1.87

Max consec. winners

19

Max consec. losers

1

Avg # bars in winners

3

Avg # bars in losers

11

Max intraday drawdown

$ -12.83


Profit factor

25.14

Max # contracts held

1

Account size required

$ 12.83

Return on account

466%

Asset:










Buy:


Sell:



Total bars =>


791



Price:

Date:

Price:

Date:

P/L:

Total:

B&H:

Bars:

Net Comm:

10,000

31.69

04/03/92

32.19

04/03/92

1.6%

1.6%

1.6%

0

1.0%

10,098

33.92

06/26/92

34.80

07/17/92

2.6%

4.2%

9.8%

21

3.0%

10,299

33.65

07/24/92

34.81

07/31/92

3.4%

7.8%

9.8%

7

5.9%

10,593

35.81

10/30/92

35.24

01/29/93

-1.6%

6.1%

11.2%

91

3.6%

10,364

42.04

07/09/93

43.45

07/23/93

3.4%

9.6%

37.1%

14

6.5%

10,652

45.72

11/12/93

45.34

01/07/94

-0.8%

8.7%

43.1%

56

5.0%

10,505

44.45

02/18/94

45.37

03/18/94

2.1%

11.0%

43.2%

28

6.6%

10,663

43.89

04/01/94

45.05

04/15/94

2.7%

13.9%

42.2%

14

8.9%

10,886

43.39

05/06/94

44.89

06/10/94

3.4%

17.9%

41.6%

35

12.0%

11,200

43.34

06/17/94

44.30

07/15/94

2.2%

20.4%

39.8%

28

13.9%

11,386

42.82

07/29/94

44.63

08/05/94

4.2%

25.5%

40.8%

7

18.1%

11,808

42.69

08/05/94

43.17

10/28/94

1.1%

26.9%

36.2%

84

18.8%

11,881

42.20

11/11/94

43.79

11/25/94

3.8%

31.7%

38.2%

14

22.7%

12,267

42.02

11/25/94

43.90

12/02/94

4.5%

37.6%

38.5%

7

27.5%

12,755

45.76

04/07/95

47.14

04/28/95

3.0%

41.8%

48.8%

21

30.8%

13,080

46.81

06/23/95

48.66

07/21/95

3.9%

47.3%

53.5%

28

35.4%

13,536

47.17

08/18/95

48.93

09/01/95

3.7%

52.8%

54.4%

14

39.8%

13,981

48.05

10/20/95

49.08

11/17/95

2.1%

56.1%

54.9%

28

42.2%

14,219

54.54

05/10/96

57.75

05/17/96

5.9%

65.3%

82.2%

7

50.0%

14,997

57.81

07/19/96

59.80

08/09/96

3.4%

71.0%

88.7%

21

54.5%

15,453

57.36

08/16/96

59.71

08/23/96

4.1%

78.0%

88.4%

7

60.3%

16,026

63.76

12/06/96

66.00

12/20/96

3.5%

84.3%

108.3%

14

65.3%

16,531

64.70

02/28/97

67.71

03/14/97

4.7%

92.9%

113.7%

14

72.4%

17,241

65.64

04/04/97

67.96

05/02/97

3.5%

99.6%

114.4%

28

77.9%

17,789

80.53

10/31/97

84.63

11/07/97

5.1%

109.8%

167.1%

7

86.3%

18,635

80.58

11/28/97

79.04

02/20/98

-1.9%

105.8%

149.4%

84

82.2%

18,220

78.60

04/17/98

83.19

04/17/98

5.8%

117.8%

162.5%

0

92.2%

19,224

80.80

05/22/98

83.75

06/19/98

3.6%

125.8%

164.3%

28

98.6%

19,864

80.27

07/17/98

84.32

07/24/98

5.0%

137.2%

166.1%

7

108.1%

20,806

80.73

08/07/98

82.51

09/11/98

2.2%

142.4%

160.4%

35

112.1%

21,205

80.42

10/23/98

84.66

11/27/98

5.3%

155.2%

167.2%

35

122.6%

22,265

79.57

12/04/98

84.73

12/11/98

6.5%

171.7%

167.4%

7

136.5%

23,649

No comments:

Post a Comment