A truly wise man, always has more questions than answers.
Tunnel Vision. Independent Signals & Positioning Recommendations.
- #142
- Feb 3, 2006 10:49pm Feb 3, 2006 10:49pm
I think you mean Marballs
I just admire someone who can make calm and rational decisions when under pressure. I'll try to keep it G rated from now on
Forex Trading: The hardest way to make easy money.
Hi Claude,
No need to feel bad. I've been there myself more times than I care to remember 
I won't be able to get started on programming this until Monday for a comparison.
Do you know what the difference is between applying the time filter as stated and running it continuously. Does it reduce the profit a lot?
Does it work for other currencies?
Sorry for asking so many questions, so only answer if you have time to run the tests.
Best regards
Alan
Hello AlanI felt bad about the discrepancy so i am reposting the zipped file for you.
Attached is a picture of one random trade with the logic that the system uses. To calculate the profit.
I won't be able to get started on programming this until Monday for a comparison.
Do you know what the difference is between applying the time filter as stated and running it continuously. Does it reduce the profit a lot?
Does it work for other currencies?
Sorry for asking so many questions, so only answer if you have time to run the tests.
Best regards
Alan
- #146
- Feb 5, 2006 1:16am Feb 5, 2006 1:16am
- Joined Sep 2004 | Status: Trader | 1,464 Posts
sxaxlxvxaxtxixoxnxbxyxgxrxaxcxexdxoxtxoxrxgx
Hello Claude
Thanks for answering the questions.
I must admit, I don't much look at the profit factor I'm initially more interested in the % drawdown. If you have a 'reliable' small % drawdown then you can just ramp up the trade size to a comfortable % drawdown and wait for the profit to roll in!
I can see actual figure of drawdown but I couldn't see % drawdown, which would be dependent on the maximum balance so far. Could calculate it manually though if it isn't given as a figure in the reports.
On the spreadsheets I downloaded
USDCHF 3pt spread -> net profit $190,970
USDCHF 4pt spread -> net profit $82,617
Over about 1000 trades (order of magnitude)
So for the sake of 1pip, it wiped off 55% of the profit.
If for whatever reason, the broker wriggled the price by 1 pip volatility, that could wipe out all the profit and put you at a loss. That seems like a very slim margin strategy.
Quick read of document to check currencies, "We trade GBP/USD, USD/CHF, and the S&P e-mini futures contract. Each has a specialty. Mine is GBP/USD."
So it should work on GBPUSD as well.
Am I right to understand that without the modifications of Demark trendline, trade only certain candle combinations, (also haven't read it thoroughly so so not sure if you changed the moving average bits), then the system is not profitable?
And, although the $ profit is higher without the time filter, the $ with the filter is more likely the return you are going to get if you traded it for real (unless you can trade 24/5!)
With a time filter, less trades are made so it could be that you get more pip per trade which gives a better 'comfort' factor, and hopefully a smaller drawdown.
Also for true testing, you need run your testing data that has not been used to optimize the system e.g. find the best case scenario for data up to 31/12/2004, then use those parameters to see what you would get for trading 2005 to simulate trading in the future.
I've heard that MT4 backtesting isn't reliable
I use Amibroker, but I don't know how to integrate programming trendlines into a trading strategy. That bit is too complicated for me!
Best regards
Alan
Hello Allen
By applying a time filter net profit is $44,701 with a profit factor of 1.379
Without the time filter net profit is $114,062 with a profit factor of 1.239
So as you can see, the system is correct, more often with the time filter on but trades a lot less and net profit falls substantially.
As far as other currencies. The system is more or less only effective on the Swiss franc. I would not trade any other currency. I have managed to make the British pound and the euro profitable with the system. But with my modifications.
I understand modifying the system is not very popular, but the one-hour system on other currencies needed to be change slightly to increase the profit factor.
And no I did not use lagging indicators. I did modify the system so that I can recognize certain candle combinations, and only trade if it has broken a Demark trendline before crossing the tunnel. This seems to have cut out a lot of the false moves.
As far as asking questions asked away {read my signature}
I must admit, I don't much look at the profit factor
I'm initially more interested in the % drawdown. If you have a 'reliable' small % drawdown then you can just ramp up the trade size to a comfortable % drawdown and wait for the profit to roll in!I can see actual figure of drawdown but I couldn't see % drawdown, which would be dependent on the maximum balance so far. Could calculate it manually though if it isn't given as a figure in the reports.
On the spreadsheets I downloaded
USDCHF 3pt spread -> net profit $190,970
USDCHF 4pt spread -> net profit $82,617
Over about 1000 trades (order of magnitude)
So for the sake of 1pip, it wiped off 55% of the profit.
If for whatever reason, the broker wriggled the price by 1 pip volatility, that could wipe out all the profit and put you at a loss. That seems like a very slim margin strategy.
Quick read of document to check currencies, "We trade GBP/USD, USD/CHF, and the S&P e-mini futures contract. Each has a specialty. Mine is GBP/USD."
So it should work on GBPUSD as well.
Am I right to understand that without the modifications of Demark trendline, trade only certain candle combinations, (also haven't read it thoroughly so so not sure if you changed the moving average bits), then the system is not profitable?
And, although the $ profit is higher without the time filter, the $ with the filter is more likely the return you are going to get if you traded it for real (unless you can trade 24/5!)
With a time filter, less trades are made so it could be that you get more pip per trade which gives a better 'comfort' factor, and hopefully a smaller drawdown.
Also for true testing, you need run your testing data that has not been used to optimize the system e.g. find the best case scenario for data up to 31/12/2004, then use those parameters to see what you would get for trading 2005 to simulate trading in the future.
I've heard that MT4 backtesting isn't reliable
I use Amibroker, but I don't know how to integrate programming trendlines into a trading strategy. That bit is too complicated for me!Best regards
Alan
- #152
- Feb 5, 2006 11:17pm Feb 5, 2006 11:17pm
- Joined Jan 2006 | Status: Monarch o' the Glen | 5,561 Posts
Hello Claude
Thanks for answering the questions.
I must admit, I don't much look at the profit factor
I can see actual figure of drawdown but I couldn't see % drawdown, which would be dependent on the maximum balance so far. Could calculate it manually though if it isn't given as a figure in the reports.
On the spreadsheets I downloaded
USDCHF 3pt spread -> net profit $190,970
USDCHF 4pt spread -> net profit $82,617
Over about 1000 trades (order of magnitude)
So for the sake of 1pip, it wiped off 55% of the profit.
If for whatever reason, the broker wriggled the price by 1 pip volatility, that could wipe out all the profit and put you at a loss. That seems like a very slim margin strategy.
Quick read of document to check currencies, "We trade GBP/USD, USD/CHF, and the S&P e-mini futures contract. Each has a specialty. Mine is GBP/USD."
So it should work on GBPUSD as well.
Am I right to understand that without the modifications of Demark trendline, trade only certain candle combinations, (also haven't read it thoroughly so so not sure if you changed the moving average bits), then the system is not profitable?
And, although the $ profit is higher without the time filter, the $ with the filter is more likely the return you are going to get if you traded it for real (unless you can trade 24/5!)
With a time filter, less trades are made so it could be that you get more pip per trade which gives a better 'comfort' factor, and hopefully a smaller drawdown.
Also for true testing, you need run your testing data that has not been used to optimize the system e.g. find the best case scenario for data up to 31/12/2004, then use those parameters to see what you would get for trading 2005 to simulate trading in the future.
I've heard that MT4 backtesting isn't reliable
Best regards
Alan
First The percent of drawdown Is under the tab Time it falls under the category equity curve.You'll see that it is 31.34%
Also the results that I posted Are the system not optimized at all They are using the moving averages that Vegas has provided.
The only difference as I mentioned before, was to prevent lots of wipspan trades I use the moving average of the high and the moving average of the low. You can see the settings under the tab settings there you will see the two moving average and two other parameters called nCon which stands for number of contracts. And another one called decimal. This is just to change it to a pair that would only have 2 decimals as compared to 4.
The British pound was not profitable, as written. You can make it profitable without the modifications but that requires changing the time of the moving average ((optimizing)).
I understand cautioned. When optimizing. I've gone through several different methods. One of which is optimizing a random 13 weeks and then taking the average.
But I now use separate software, which shows me a 3-D graph. And I able to pick the most robust parameters.
Another way I optimize is my perfect system method. I basically build a system that is perfect using a zigzag indicator. As you are aware. It is impossible to trade with the zigzag. It looks back and changes constantly, but it does build a perfect system. If you want it to. Of course this would be impossible to trade in real-time.
You then take every trade that the perfect system took and put them into an Excel file.
You run the system that you have built through optimization process and save the best 100. You then put every trade that this run-through's have taken and put them into an Excel file.
You then run a correlation analysis and take the parameters that most closely match the perfect system. I find this to be most effective for parameters that stay true into the future. This way, you can optimize over all of your data and still get parameters that produced a straight equity curve. And in my experience, have a much higher percentage of succeeding into the future.
A truly wise man, always has more questions than answers.
- #155
- Feb 6, 2006 12:06pm Feb 6, 2006 12:06pm
Hello Claude,
Thanks, it's the first time I've seen a Tradestation report!
My mistake, I assumed you included Demark trendlines and candle formations as you mentioned them here "I did modify the system so that I can recognize certain candle combinations, and only trade if it has broken a Demark trendline before crossing the tunnel. This seems to have cut out a lot of the false moves."
Very curious as Vegas reported he was trading it profitably. But he has said the market does change.
Sounds possible. I've no experience of doing it that way.
The latest VWBII has some very interesting ideas.
Best regards
Alan
Hello Alan
First The percent of drawdown Is under the tab Time it falls under the category equity curve.You'll see that it is 31.34%
Also the results that I posted Are the system not optimized at all They are using the moving averages that Vegas has provided.
The only difference as I mentioned before, was to prevent lots of wipspan trades I use the moving average of the high and the moving average of the low. You can see the settings under the tab settings there you will see the two moving average and two other parameters called nCon which stands for number of contracts. And another one called decimal. This is just to change it to a pair that would only have 2 decimals as compared to 4.
My mistake, I assumed you included Demark trendlines and candle formations as you mentioned them here "I did modify the system so that I can recognize certain candle combinations, and only trade if it has broken a Demark trendline before crossing the tunnel. This seems to have cut out a lot of the false moves."The British pound was not profitable, as written. You can make it profitable without the modifications but that requires changing the time of the moving average ((optimizing)).
I understand cautioned. When optimizing. I've gone through several different methods. One of which is optimizing a random 13 weeks and then taking the average.
But I now use separate software, which shows me a 3-D graph. And I able to pick the most robust parameters.
Another way I optimize is my perfect system method. I basically build a system that is perfect using a zigzag indicator. As you are aware. It is impossible to trade with the zigzag. It looks back and changes constantly, but it does build a perfect system. If you want it to. Of course this would be impossible to trade in real-time.
You then take every trade that the perfect system took and put them into an Excel file.
You run the system that you have built through optimization process and save the best 100. You then put every trade that this run-through's have taken and put them into an Excel file.
You then run a correlation analysis and take the parameters that most closely match the perfect system. I find this to be most effective for parameters that stay true into the future. This way, you can optimize over all of your data and still get parameters that produced a straight equity curve. And in my experience, have a much higher percentage of succeeding into the future.
The latest VWBII has some very interesting ideas.
Best regards
Alan
- #157
- Feb 6, 2006 12:59pm Feb 6, 2006 12:59pm
- Joined Jan 2006 | Status: Monarch o' the Glen | 5,561 Posts
- #158
- Feb 6, 2006 1:00pm Feb 6, 2006 1:00pm
- Joined Sep 2004 | Status: Trader | 1,464 Posts
sxaxlxvxaxtxixoxnxbxyxgxrxaxcxexdxoxtxoxrxgx