Hi,
I'm developing an EA based on rule (works on daily chart):
1. If yesterday hi-lo more than _MinimumRange_ then do following step otherwise skip
2. Get entry level by multiplying yesterday hi-lo with _FiboLevel_
3. Add/substract the number to
3a. Yesterday's low if (yesterday close > yesterday open) [Add]
3b. Yesterday's high if (yesterday close < yesterday open) [Sub]
4. If today price hit entry level calculated in step 2 then
4a. if [yesterday close > yesterday open] then go long
4b. if [yesterday close < yesterday open] then go short
note: position opened only when there's no same position (long/short) has not been closed
5. Every position opened closed only when hit sl or tp
6. sl and tp can be set from _slmarg_ and _tpmarg_ parameter
note:
every variables in _xxx_ format are EA's parameter
To get optimum parameter combination, I ran optimizer with 3 variables varied: _slmarg_, _tpmarg_, and _MinimumRange_, and
get the following result (attached as Donny_01_Optimizer.xls in zip file)
and the top five most 'profitable' result are:
No Net_Profit Total_Trades Profit_Factor Expected_Payoff DrawDown %Drawdown param1 param2 param3 param4 param5 Net_Profit/Max_Drawdown
122 32392.6 298 1.24 108.7 11483.4 39.64% slmarg=120 tpmarg=90 MinimumRange=170 FiboLevel=0 lots=1 2.820819618
31 32596.4 407 1.15 80.09 18463.7 56.69% slmarg=120 tpmarg=110 MinimumRange=150 FiboLevel=0 lots=1 1.765431631
259 34172.4 211 1.38 161.95 9365.4 22.64% slmarg=120 tpmarg=90 MinimumRange=190 FiboLevel=0 lots=1 3.648792363
44 36006.1 394 1.16 91.39 20349.6 56.87% slmarg=120 tpmarg=130 MinimumRange=150 FiboLevel=0 lots=1 1.769376302
38 36085.2 401 1.16 89.99 17072.5 53.12% slmarg=120 tpmarg=120 MinimumRange=150 FiboLevel=0 lots=1 2.11364475
considering max drawdown I try rerun iteration combination no 259, which net profit is 34172 and %DD is 22.65%
here is the result for that param combination (file Donny10.htm)
I ran with historical data from {link removed} starting from 1 Jan 2002 to 1 Feb 2007
it makes me wonder, if I'm doing something wrong, questions are:
1. Could what I did considered as 'curve fitting' ?
2. How high the probability that the strategy with that combination will make good performance with real
(forward testing) data?
Please advise as I still learning how to testing strategy using EA.
thanks,
I'm developing an EA based on rule (works on daily chart):
1. If yesterday hi-lo more than _MinimumRange_ then do following step otherwise skip
2. Get entry level by multiplying yesterday hi-lo with _FiboLevel_
3. Add/substract the number to
3a. Yesterday's low if (yesterday close > yesterday open) [Add]
3b. Yesterday's high if (yesterday close < yesterday open) [Sub]
4. If today price hit entry level calculated in step 2 then
4a. if [yesterday close > yesterday open] then go long
4b. if [yesterday close < yesterday open] then go short
note: position opened only when there's no same position (long/short) has not been closed
5. Every position opened closed only when hit sl or tp
6. sl and tp can be set from _slmarg_ and _tpmarg_ parameter
note:
every variables in _xxx_ format are EA's parameter
To get optimum parameter combination, I ran optimizer with 3 variables varied: _slmarg_, _tpmarg_, and _MinimumRange_, and
get the following result (attached as Donny_01_Optimizer.xls in zip file)
and the top five most 'profitable' result are:
No Net_Profit Total_Trades Profit_Factor Expected_Payoff DrawDown %Drawdown param1 param2 param3 param4 param5 Net_Profit/Max_Drawdown
122 32392.6 298 1.24 108.7 11483.4 39.64% slmarg=120 tpmarg=90 MinimumRange=170 FiboLevel=0 lots=1 2.820819618
31 32596.4 407 1.15 80.09 18463.7 56.69% slmarg=120 tpmarg=110 MinimumRange=150 FiboLevel=0 lots=1 1.765431631
259 34172.4 211 1.38 161.95 9365.4 22.64% slmarg=120 tpmarg=90 MinimumRange=190 FiboLevel=0 lots=1 3.648792363
44 36006.1 394 1.16 91.39 20349.6 56.87% slmarg=120 tpmarg=130 MinimumRange=150 FiboLevel=0 lots=1 1.769376302
38 36085.2 401 1.16 89.99 17072.5 53.12% slmarg=120 tpmarg=120 MinimumRange=150 FiboLevel=0 lots=1 2.11364475
considering max drawdown I try rerun iteration combination no 259, which net profit is 34172 and %DD is 22.65%
here is the result for that param combination (file Donny10.htm)
I ran with historical data from {link removed} starting from 1 Jan 2002 to 1 Feb 2007
it makes me wonder, if I'm doing something wrong, questions are:
1. Could what I did considered as 'curve fitting' ?
2. How high the probability that the strategy with that combination will make good performance with real
(forward testing) data?
Please advise as I still learning how to testing strategy using EA.
thanks,
Attached File(s)