TC,
New to the thread, and certainly curious about your strategy. The results are indeed impressive. Thought I might contribute a little on a nuts and bolts issue.
Also, have you tinkered with 15-minute charts with 20 and 32 MAs to stretch out the signal? Would this actually be the same thing?
No they are not the same, though very close. And I'm afraid I'm at a loss to explain why...anyone?
Indeed they are not the same, and the difference is likely to be most pronounced following a trend. It is perhaps best illustrated by a simple example. If you average the numbers 1 through 32, the result is 16.5, but if you average the eight numbers 4 through 32 in increments of 4, the result is 18. For any trending sequence of numbers (stock prices), using more and smaller increments to "sample" the trend results in giving weight to earlier parts of the sequence. In other words, each of your 1-hour bars, which in the 8 period 1-hour average is represented by the single closing value of each bar, effectively gets represented by an average value for that bar, a point somewhere near the middle. If the closing price of every 1-hour bar was near the middle of the bar, the averages would be very close. But, during a trend the bars tend to close near the end; the 32 period 15-minute average will always lag the 8 period 1-hour bar average when the price is trending for several hours, at the times of the 1-hour updates.
The exponential average is of course a bit more complex, but the principle is the same. More samples give rise to effectively averaging the middle of the bar instead of the close of the bar. If instead of using the close price for your average you used something like a "pivot point", (high + low + close)/3, the differences between the averages from the two sample rates would be much smaller.
Dan |
