spotdespot wrote:

michalkraus wrote:
I have one idea : what if we set some size of range that is too much for setting the trade. For example lets say that zone must have range at most 200 pips, when it is bigger, we wont trade it. I am not saying 200 pips is that correct number it will probably vary, but might work as a good filter....

Thanks Phil

Just arrived here from Philippe's link from the Marylin thread.

I am not sure if this point above was ever taken in to consideration with the testing as I think it is pretty important. I trade manually from Frankfurt open onwards and I have a simple mental rule that if the "Asian" range (I define this as NY close to Frankfurt open) is greater than half the ADR then only take cast iron trades - basically don't trade that pair for the day unless a trade absolutely smacks you between the eyes The reason being that in my experience price often just bounces around, ranging with little clear direction and often presents dodgy setups that reverse against you.

I think this same principle would save this strategy some pain too. My "rule" is a mental one but I suspect there is no reason why a more mathematically solid % of ADR rule for the Asian session to prevent trading couldn't be tested?

Cheers,
Dave.

SpiderX wrote:

phil_trade wrote: Hi

I will publish FT code when ready. I guess Tommaso could translate in MQL.

But yes, before that, we need to learn why such DD in these period cause I've learn something with Forex... worse come sooner than you expect it

Hi Phil,

The good thing is that these periods are fairly extended, so it might be possible to do some trade management on equity curve, to either keep out of the market during that period or lower risk until equity curve starts improving.
One possibility is to draw a 10 EMA on the equity curve and stay out of the market when equity curve is below 10 EMA.

Of course equity management method will fail if equity curve fluctuates up and down in a short period of time.


It might be unrealistic for us to expect a strategy to work in ALL market conditions, but at least we can know when things are going south and cut our risk accordingly.


Cheers

SpiderX wrote:Hi Philippe,

I think the way this works is that EA need to keep track of cumulative profit/loss on a daily basis.
Based on cumulative P/L values, EA calculates a moving average of the PL.
A simple rule can be:
if ( cumulative PL < MA of PL && MA of PL is in downtrend)
then
reduce lots.. or stop trading

by the way, my ideas come from here:
http://capstonetradingsystems.com/money ... algorithms

the implementation in here is on Tradestation, where they can still "dummy trade" and track the recovery of the equity curve.
Not sure how this can work out on Empty4.

However, this looks to be a good way to avoid extended periods of unfavorable conditions and wait for the market to get to better conditions.

Cheers

phil_trade wrote:
Ok, I understand the principle. I will think about it.

garyfritz wrote:Trading the equity curve doesn't usually work very well. It's **exactly** analogous to writing an EA that trades when the market is above an MA, and goes flat when it goes below. That kind of EA doesn't generally work and equity-curve trading generally doesn't either. See e.g. my post at http://www.stevehopwoodforex.com/phpBB3 ... 780#p28780.

BUT... in this case it might help. This system seems to produce fairly long streaks of wins and losses. So I did some experimentation with it. Looking at each individual pair, I said "don't take this week's trade unless this pair was profitable in the last 6 weeks." That didn't produce any more profits but it smoothed the results out a fair amount, and got rid of the major drawdowns.

Phillipe, I don't know if you could test this in Forextester or not. I can't in Tradestation. You have to run the system, taking all trades, so you know what the equity curve is. Then you have to determine whether you actually should have taken the trades. I did the analysis by dumping my trades to a file and checking it in Excel.

phil_trade wrote:
Hi Gary

I have to write first the module of "fake" trades to run calculation of MA during "no trade" period.

What do you think about run normal trade 0.1 lot and decrease to 0.01 lot when 6 weeks unprofitable. It's easy to do and should give about same result ?

garyfritz wrote:You have to simulate all of the trades with the SAME position size in order to accurately calculate the equity-curve-MA or the 6-weeks-profitable or whatever you're using.

Dropping down to 0.01 would mess up the equity curve calculations, unless you also kept track of the results from the 0.1-size trades. And if you're doing that anyway, you can drop to any size you want -- 0.01, or 0.00, or whatever.