|
PureBytes Links
Trading Reference Links
|
MB wrote
>COME GET SUCKED IN...
Thanks to Gary Fritz for forwarding this to the list in January - the
subject was the illusion of trend
The basic idea behind technical analysis is to look at the past in
order to predict the future. One of the simplest questions we can
answer is what happens after the market goes up in some time frame
versus what happens after it goes down. A trending market goes up
more than average after it has been going up, and down more than
average after it has been going down.
On the other hand, the S&P 500 does the opposite. "Buy the dips"
works on the S&P 500 futures.
Consider the result below and ask yourself whether you want to buy
after a rise or after a drop:
S&P_500
3233 days of data, 588.200000 last close.
Box Program, sort market into boxes by month, week, day change.
10 days, 5 days Avg change Count Total $ change
up up 21.313348 1401 29860.000000
up down -50.402174 460 -23185.000000
down up 37.985437 412 15650.000000
down down 143.618143 948 136150.000000
Best total for all boxes correct = 204.8 K dollars
20 days, 10 days Avg change Count Total $ change
up up -4.290541 1480 -6350.000000
up down 147.679814 431 63650.000000
down up 40.583554 377 15300.000000
down down 93.716143 923 86500.000000
Best total for all boxes correct = 171.8 K dollars
The Avg change column above is the average daily change in S&P 500
futures, measured in dollars, on the day after an observation that
the market moved up or down in the past N days, listed in the left 2
columns. There are 4 possibilities for any 2 time frames, since
either time may be up or down in the past.
Here is another measure of Trend, as Alex Saitta has described in
TASC. Measure the standard deviation of the market over various
times. A random walk will show a standard deviation which increases
as the square root of the number of days. A trending market will
show a standard deviation which increases faster, and an anti-
trending, or reversing market, will show a standard deviation which
increases slower than the random walk. Clearly, the S&P 500 standard
deviation increases slower than random.
S&P_500
3233 days of data, 588.200000 last close.
Trend measure: standard deviation over various times
Days Std dev Expected Actual
1 0.008899 1 1.000000
4 0.016235 2 1.824386
9 0.023002 3 2.584797
16 0.029321 4 3.294872
25 0.035648 5 4.005789
36 0.042230 6 4.745422
49 0.049262 7 5.535622
64 0.055698 8 6.258924
81 0.062575 9 7.031621
100 0.068388 10 7.684899
"Looks like a trend" is not enough. Do reds and blacks trend in
roulette? Sometimes they look like they do. But it is really random.
We must measure mathematically to answer this question properly.
And the mathematics above tell me that the S&P 500 tends to reverse
more often than it trends.
Another way to measure trend is to see whether simple trend following
systems work on the market. Moving average and breakout are the two
simplest trend following systems. They work on Pound, Coffee, Crude,
D Mark, Dollar, Eurodollar, Orange Juice, Lumber, Swiss Franc, T
Bill, T Bond, and Yen.
But moving average and breakout both lose money on S&P 500.
S&P_500
3233 days of data, 588.200000 last close.
Moving Average Trading, Buy on close crossing up, sell crossing down.
Commission is 50.00 dollars
50 day moving average, buy on close above it, sell below.
tot=-96875.00, -7613.56 dollars/year, 257 trades, 14 percent win
-2.59 percent per year return unleveraged.
Average win = 7365.13, Average loss = 1720.32
Average win/loss ratio = 4.281258
100 day moving average, buy on close above it, sell below.
tot=-77400.00, -6180.13 dollars/year, 167 trades, 13 percent win
-2.10 percent per year return unleveraged.
Average win = 10080.68, Average loss = 2063.28
Average win/loss ratio = 4.885765
150 day moving average, buy on close above it, sell below.
tot=-19025.00, -1543.74 dollars/year, 127 trades, 12 percent win
-0.52 percent per year return unleveraged.
Average win = 12278.12, Average loss = 1941.22
Average win/loss ratio = 6.324965
200 day moving average, buy on close above it, sell below.
tot=72025.00, 5940.70 dollars/year, 81 trades, 18 percent win
2.02 percent per year return unleveraged.
Average win = 12598.33, Average loss = 1771.97
Average win/loss ratio = 7.109791
250 day moving average, buy on close above it, sell below.
tot=58700.00, 4922.84 dollars/year, 83 trades, 15 percent win
1.67 percent per year return unleveraged.
Average win = 13507.69, Average loss = 1670.00
Average win/loss ratio = 8.088439
These results are dismal, compared to buy and hold. Moving average is a
failure on S&P 500. How could a trend following system fail on a trending
market? Maybe it is not a trending market.
Wanna try breakout?
S&P_500
3233 days of data, 588.200000 last close.
Breakout Intraday Trading Program, Buy above old high, sell below old low.
Commission is 100.00 dollars
10 day lookback, buy above highest high, sell below lowest low.
tot=-195175.00, -15101.75 dollars/year, 224 trades, 28 percent win
-5.13 percent per year return unleveraged.
Average win = 4864.06, Average loss = 3165.47
Average win/loss ratio = 1.536601
20 day lookback, buy above highest high, sell below lowest low.
tot=-75975.00, -5878.60 dollars/year, 114 trades, 30 percent win
-2.00 percent per year return unleveraged.
Average win = 6765.00, Average loss = 3958.86
Average win/loss ratio = 1.708825
30 day lookback, buy above highest high, sell below lowest low.
tot=-66825.00, -5170.61 dollars/year, 75 trades, 29 percent win
-1.76 percent per year return unleveraged.
Average win = 9647.73, Average loss = 5265.57
Average win/loss ratio = 1.832230
40 day lookback, buy above highest high, sell below lowest low.
tot=-92025.00, -7120.47 dollars/year, 61 trades, 27 percent win
-2.42 percent per year return unleveraged.
Average win = 10957.35, Average loss = 6325.00
Average win/loss ratio = 1.732388
50 day lookback, buy above highest high, sell below lowest low.
tot=-62425.00, -4830.16 dollars/year, 49 trades, 28 percent win
-1.64 percent per year return unleveraged.
Average win = 13950.00, Average loss = 7363.57
Average win/loss ratio = 1.894461
60 day lookback, buy above highest high, sell below lowest low.
tot=54675.00, 4230.50 dollars/year, 33 trades, 36 percent win
1.44 percent per year return unleveraged.
Average win = 16110.42, Average loss = 6602.38
Average win/loss ratio = 2.440092
70 day lookback, buy above highest high, sell below lowest low.
tot=-7325.00, -566.77 dollars/year, 31 trades, 35 percent win
-0.19 percent per year return unleveraged.
Average win = 14690.91, Average loss = 8446.25
Average win/loss ratio = 1.739341
80 day lookback, buy above highest high, sell below lowest low.
tot=9725.00, 752.48 dollars/year, 27 trades, 40 percent win
0.26 percent per year return unleveraged.
Average win = 14075.00, Average loss = 9068.75
Average win/loss ratio = 1.552033
How could breakout fail on a trending market? Maybe it's not a
trending market.
Here we have 4 totally different measures which all agree on this
conclusion: The S&P 500 does not trend, rather it tends to reverse
its past movements.
Dave Chamness
|