First of all sorry for the long time passed from your reply.
Quote:
Originally Posted by dbbpj
What I like about the idea you put forth in your article is the overall pluggable management of the whole process relating to time based analysis. This basic shell seems like a perfect container to insert a time based trend detection model which can have two parameters... entry time(X) and exit time(Y).
|
Using the code in the article you can play with different entry times, opening a time windows for enter trades and also set a maximum holding period. During this time I had tried a lot of different combinations without having any particular interesting result. Every time I decided to close a trade after a particular time all results seemed worst than those without any limit. Maybe an exit strategy different from a TP or trailing stop should be more effective.
Quote:
|
Exit time... (TP) can also be an indicative guide which can optionally be replaced by a price % run adjustable setting per unit time which is in turn determined by the pair's volatility expectation; again, determined by the original distance between time X and time Y.
|
Actually I have reached this situation: 3 different reference price, each based on actual volatility (not expected, which are your suggestions ?). These are: a First Target price, a Stop Loss and a Trailing Stop to be entered when first target price is matched.
Now First Target Price become next Target Price and the process trace the price until the trade is closed by the stop loss.
Again some attempt to create a dependence with time in the target price, for example reducing by a progressive factor while time passes, but with no interesting results.
Quote:
|
In terms of risk management (stop level, etc), the key is still with the detection model's ability to select optimum RR and (optionally) hit rate scenarios. What are your thoughts..?
|
About risk management I think that to tune up a trading system for better performance you must reduce as much as possible the number of variable ad parameters you are controlling.
Stopping the time at a single hour for each trade reduce the complexity of you system by one dimension and let you concentrate on other occurrence. Introducing a simple risk management that vary your lot size, during optimization, can lead to very uncertain results, maybe during a period better performance give you good results that are amplified by the risk management module and close your eyes to other period in which your system does not perform as you should expect. So better always leave as much as possible every test field with as much parameter unchanged as possible.
Other exit strategy should be investigated, I tried some exit signal but with no good results, so I should find something else.
I like your idea of volatility expectation. Let's figure out which direction to search.
ciao,
Giampiero