Sunday, October 31, 2010

Nasdaq and VXN Rising

There is no question that the NASDAQ ($COMP) is leading this market's latest uptrend that began on September 1.  Many software stocks have moved into a leadership role during this time.  The NASDAQ has been up 8 straight days, while the $VXN has risen over that same period (especially this past week, up 8.9%).  Also, two instances this past week (Tues and Fri), the $COMP and $VXN both rose for 2 consecutive days. I did a little study earlier this week on what happens when the $SPX and $VIX both rise 2 consecutive days. For this NASDAQ test, I added the criteria that the index closes above its 200 day moving average, which it did Friday (even without this filter, the results are almost the same).


When VXN and NASDAQ both rise 2 consecutive days, buy NASDAQ on close > 200 day MA;  sell 'n' days later. $100,000 per trade; 1995 to present.
Exit n
 days
Net Profit
# Trades
# of winners
% of Winners
Max. Trade % DD
Avg % P/L
W. Avg. Profit
L. Avg. Loss
Profit Factor
Payoff Ratio
1
      7,540.43
60
31
51.67
-2.68
0.13
     839.26
       (637.12)
1.41
1.32
2
    22,624.18
54
33
61.11
-3.24
0.42
  1,318.51
       (994.61)
2.08
1.33
3
    33,117.40
52
37
71.15
-3.24
0.64
  1,439.15
    (1,342.09)
2.65
1.07
4
    35,241.59
48
31
64.58
-6.41
0.73
  2,121.11
    (1,794.87)
2.15
1.18
5
    45,219.66
48
33
68.75
-10.18
0.94
  2,360.02
    (2,177.40)
2.38
1.08
6
    41,750.05
48
33
68.75
-10.18
0.87
  2,367.41
    (2,424.96)
2.15
0.98
7
    43,996.20
48
33
68.75
-14.90
0.92
  2,539.97
    (2,654.86)
2.10
0.96
8
    41,523.03
46
31
67.39
-16.40
0.90
  2,928.46
    (3,283.96)
1.84
0.89
9
    28,408.65
46
30
65.22
-16.40
0.62
  2,760.14
    (3,399.73)
1.52
0.81
10
    46,445.00
44
29
65.91
-16.40
1.06
  3,062.82
    (2,825.13)
2.10
1.08
11
    51,125.58
44
31
70.45
-16.40
1.16
  2,764.85
    (2,660.37)
2.48
1.04
12
    53,802.95
44
28
63.64
-16.40
1.22
  3,375.99
    (2,545.29)
2.32
1.33
13
    54,239.45
44
29
65.91
-18.28
1.23
  3,374.55
    (2,908.17)
2.24
1.16
14
    60,001.91
44
30
68.18
-24.05
1.36
  3,669.35
    (3,577.04)
2.20
1.03
15
    56,300.82
44
30
68.18
-25.92
1.28
  3,770.02
    (4,057.13)
1.99
0.93
16
    43,738.20
42
31
73.81
-33.08
1.04
  3,650.85
    (6,312.57)
1.63
0.58
17
    41,979.42
40
28
70.00
-33.08
1.05
  3,912.71
    (5,631.38)
1.62
0.69
18
    46,751.33
40
27
67.50
-33.08
1.17
  4,121.32
    (4,963.41)
1.72
0.83
19
    51,250.93
40
28
70.00
-33.08
1.28
  4,161.59
    (5,439.46)
1.79
0.77
20
    59,083.29
40
28
70.00
-33.08
1.48
  4,541.55
    (5,673.35)
1.87
0.80


This wasn't what I thought it would be. There basically is some kind of upside bias between 1 and 20 days out, although the max trade drawdown gets a little scary after a week.
-------------------------------------------------------------------------------------------------------------

To me, the Nasdaq looked tired this week. It has gone up 8 straight days, and volume has gradually been increasing the last 5 days while prices have hardly moved. It looked liked churning, which I thought was bearish. So I wanted to test an idea: what happens when you buy the Nasdaq on the close after it has risen for 8 straight days, which it did on Friday 10-29-10?



When NASDAQ rises 8 consecutive days, buy on close > 200 day MA; sell 'n' days later. $100,000 per trade; 1995 to present.
Net Profit
# Trades
# of winners
% of Winners
Max. Trade % DD
Avg % P/L
W. Avg. Profit
L. Avg. Loss
Profit Factor
Payoff Ratio
1
         276.25
18
9
50.00
-1.78
0.02
     732.99
       (702.30)
1.04
1.04
2
    12,472.36
15
12
80.00
-1.74
0.83
  1,259.96
       (882.38)
5.71
1.43
3
    10,293.41
14
11
78.57
-2.21
0.74
  1,292.94
    (1,309.66)
3.62
0.99
4
    10,856.26
13
10
76.92
-3.92
0.84
  1,670.45
    (1,949.40)
2.86
0.86
5
    14,690.67
13
9
69.23
-4.13
1.13
  2,270.72
    (1,436.46)
3.56
1.58
6
    12,086.10
13
8
61.54
-5.55
0.93
  2,369.37
    (1,373.78)
2.76
1.72
7
      6,644.75
13
8
61.54
-6.14
0.51
  2,199.60
    (2,190.41)
1.61
1.00
8
      5,719.95
13
9
69.23
-9.78
0.44
  2,219.64
    (3,564.21)
1.40
0.62
9
    14,627.39
13
9
69.23
-9.78
1.13
  2,606.46
    (2,207.68)
2.66
1.18
10
    20,579.55
13
11
84.62
-9.78
1.58
  2,494.38
    (3,429.31)
4.00
0.73
11
    22,734.70
13
10
76.92
-9.78
1.75
  3,206.61
    (3,110.48)
3.44
1.03
12
    25,796.62
13
11
84.62
-11.35
1.98
  3,641.58
    (7,130.38)
2.81
0.51
13
    31,348.43
13
11
84.62
-11.35
2.41
  3,889.58
    (5,718.49)
3.74
0.68
14
    34,566.36
13
12
92.31
-11.35
2.66
  3,624.15
    (8,923.42)
4.87
0.41
15
    37,864.92
13
12
92.31
-11.35
2.91
  3,827.43
    (8,064.18)
5.70
0.47
16
    37,719.07
13
12
92.31
-11.35
2.90
  3,846.63
    (8,440.53)
5.47
0.46
17
    36,131.61
13
11
84.62
-11.35
2.78
  4,287.54
    (5,515.68)
4.28
0.78
18
    41,211.18
12
10
83.33
-11.35
3.43
  5,071.84
    (4,753.60)
5.33
1.07
19
    36,912.98
12
11
91.67
-11.35
3.08
  4,289.00
  (10,266.03)
4.60
0.42
20
    40,277.44
12
11
91.67
-11.35
3.36
  4,650.75
  (10,880.84)
4.70
0.43


There aren't too many instances, but I think there are enough to get the idea. Basically, when this index goes up for 8 days in a row, there is a good chance that it will continue to go up. Even when I loosened the time frame to 1970 - present, there are 5 times more trades, every exit period 1 through 20 days was net positive, and the success rate averaged about 65%. So much for my bearish intuition here. That's what makes these studies so great; they usually defy my natural logic.

Interesting results to the tests. I really expected a bearish bias to appear at some point, but it is nice to be surprised in this game.  That's how we learn. One thing to note is that the test does not factor many things: technical chart analysis (overhead resistance at the 2500 - 2530 level), macro-economic events (FED meeting, Jobs report), political events (huge mid-term elections on Tues.) and global/social issues (terrorism thwart on Friday). This was just a study of an idea, but is does provide something to think about for the next couple weeks.

Footnotes

Tests run in Amibroker using data from Norgate. I also ran re-ran variations of the tests, interchanging $VXN, $VIX, $COMP, and $NDX, and all results were similar.