I study trading system ideas with my research buddies.
My team (non-pro) studied cycle period-based channel breakouts but in vain. We tested more than 100 ideas in the area. (When we do research, we usually test more than 100 ideas. But we as a team kind of gave up already.)
Here is our basic idea:
A Turtle-type of system will buy/sell when todayâs price is higher/lower than the past 20-day high/low. As you can see, it uses a fixed number, in this case 20, as a lookback period.
Probably, the market condition must be constantly changing, i.e., the current cycle period may vary all the time.
So, we think we had better use the current cycle length (which may change day by day) as a lookback period.
For instance, if todayâs measured cycle period is 25 days, a system will buy/sell when todayâs price becomes higher/lower than the past 25-day high/low.
The amended system will enter the market when the current price exceeds the past cycle high/low.
Our results were not good at all. We feel that the market does not necessarily break out after the price surpasses the past cycle high/low.
One time we added a filter to the first amended system. Only if the current price is higher /lower than an adaptive moving average, we take these buy/sell signals made by the first amended system. Otherwise, signals from the amended system are ignored. The result got better. But we feel that is mainly because of the power of the M.A. We feel cycle period-based channel breakouts can work only with some other good techniques.
As mentioned above, we may not have tested enough. In addition, we mainly tested our ideas with bond/interest rate futures. In this regard, what we did was less than half-baked.
Is there anyone who has tested something related to cycle period-based channel breakouts? Please advise. Thanks.
My team (non-pro) studied cycle period-based channel breakouts but in vain. We tested more than 100 ideas in the area. (When we do research, we usually test more than 100 ideas. But we as a team kind of gave up already.)
Here is our basic idea:
A Turtle-type of system will buy/sell when todayâs price is higher/lower than the past 20-day high/low. As you can see, it uses a fixed number, in this case 20, as a lookback period.
Probably, the market condition must be constantly changing, i.e., the current cycle period may vary all the time.
So, we think we had better use the current cycle length (which may change day by day) as a lookback period.
For instance, if todayâs measured cycle period is 25 days, a system will buy/sell when todayâs price becomes higher/lower than the past 25-day high/low.
The amended system will enter the market when the current price exceeds the past cycle high/low.
Our results were not good at all. We feel that the market does not necessarily break out after the price surpasses the past cycle high/low.
One time we added a filter to the first amended system. Only if the current price is higher /lower than an adaptive moving average, we take these buy/sell signals made by the first amended system. Otherwise, signals from the amended system are ignored. The result got better. But we feel that is mainly because of the power of the M.A. We feel cycle period-based channel breakouts can work only with some other good techniques.
As mentioned above, we may not have tested enough. In addition, we mainly tested our ideas with bond/interest rate futures. In this regard, what we did was less than half-baked.
Is there anyone who has tested something related to cycle period-based channel breakouts? Please advise. Thanks.
