LoRio
Sergeant
- Messages
- 484
Hello FPA .
I started developing a web application aimed at providing training in forex trading utilizing past data.
Why is it needed:
There are some ways to learn .
A.Backtesting with an EA that simulates your strategy.
B.ForwardTesting on demo account.
C.ForwardTesting on live account.
D.Utilizing market simulation functions of trading software.
E.Paper Trading
I have tried all the above and here is my thesis:
A.The reliability of backtests may be a headache especially if its a scalping strategy.
Most software will backtest on data that is either full of gaps , has limited past time span to back test on , or uses tick interpolation for simulation(random replay of ticks using High,Low,Open,Close and volume data)
Furthermore ,adjusting an EA to enter on the same terms as you would is a hassle.
Also , indicator values sometimes cannot be utilized ,and if so , their values are unreliable since a more stable strategy would utilize only closed candle indicator values.
B+C.This is by far the best method , my only issue here is having to wait for our strategy's setup to form.It immerses you well into the psychology needed (on a live account) but many approaches will prematurely hit the trash due to impatience or "losing money while out" phenomena
D.I love market simulation software.But , i hate the fact i can see the following candles before i indicate the starting candle of the simulation.
Furthermore ,its a hassle to set up a simulation account that properly simulates event flows you meet on a live account.(margin call,insuficient balance,leverage modification other than a drop down and select)
E.Same as the above with paper trading if it's performed in the past.
Not fully immersive if its performed as a forwardtest.
So heres my approach.
I have downloaded tick data (dukascopy thankyou) for the past 3 years,and H4,D1,W,MN data as far back as available.
Same for news prints starting from 2010(investing dot com thanks).
I intend to build an application which simulates exactly a live account on MetaTrader 4 (visually),has accurate tick data (accurate to the extent it came from the same source)
simulates margin correctly and is flexible in terms of creating a virtual account EXACTLY the same as the account you'll get when you deposit money with your broker.
Also , before entering simulation you will not take an accidental peek at what followed on that currency pair.
I cant simulate :
-variable spreads
-swaps
-i dont have access to CB chairman interviews ,times of war , looming war etc.
news go to the extent of prints with the previous forecast and actual numbers only.!
Alleged Pros:
-i want to be able to sit my bottom down before an NFP print , open up the simulator and have a little go at ALL the available NFP prints stored in the app's history , simulated .Something like a warm up before the match.
-i want to be able to sit and select a trading session(or timespan) and get my training on that specific timeframe.
-i want to be able to train specifically on volatile conditions and the opposite(although its kind of hindsight dangerous , meaning , you select volatile conditions training , and thus you know there will be volatility)
-i want to be able to train specifically on ranging -trending conditions (although hindsight is crucial here too)
This cannot go public due to the fact i own none of this data,and it definetely wont be for sale.
If someone from Investing.Com or DukasCopy reads this , and has an objection , they are more than welcome to tell me "Dude , please use the data only for your self".
If FPA believes this can harm their site in any way ,they are welcome to erase it(or make it private),its their website after all.
I'm thinking a local server on my spare pc which will be open for use -when i am not in need of bandwidth on my home time-is the way to go.
Also , i wont lie , if there is something i have not considered , you are welcome to contribute as this may help me develop a more immersive environment as far as "preparing for trading"/"learning" is concerned.
Thanks for reading.
I started developing a web application aimed at providing training in forex trading utilizing past data.
Why is it needed:
There are some ways to learn .
A.Backtesting with an EA that simulates your strategy.
B.ForwardTesting on demo account.
C.ForwardTesting on live account.
D.Utilizing market simulation functions of trading software.
E.Paper Trading
I have tried all the above and here is my thesis:
A.The reliability of backtests may be a headache especially if its a scalping strategy.
Most software will backtest on data that is either full of gaps , has limited past time span to back test on , or uses tick interpolation for simulation(random replay of ticks using High,Low,Open,Close and volume data)
Furthermore ,adjusting an EA to enter on the same terms as you would is a hassle.
Also , indicator values sometimes cannot be utilized ,and if so , their values are unreliable since a more stable strategy would utilize only closed candle indicator values.
B+C.This is by far the best method , my only issue here is having to wait for our strategy's setup to form.It immerses you well into the psychology needed (on a live account) but many approaches will prematurely hit the trash due to impatience or "losing money while out" phenomena
D.I love market simulation software.But , i hate the fact i can see the following candles before i indicate the starting candle of the simulation.
Furthermore ,its a hassle to set up a simulation account that properly simulates event flows you meet on a live account.(margin call,insuficient balance,leverage modification other than a drop down and select)
E.Same as the above with paper trading if it's performed in the past.
Not fully immersive if its performed as a forwardtest.
So heres my approach.
I have downloaded tick data (dukascopy thankyou) for the past 3 years,and H4,D1,W,MN data as far back as available.
Same for news prints starting from 2010(investing dot com thanks).
I intend to build an application which simulates exactly a live account on MetaTrader 4 (visually),has accurate tick data (accurate to the extent it came from the same source)
simulates margin correctly and is flexible in terms of creating a virtual account EXACTLY the same as the account you'll get when you deposit money with your broker.
Also , before entering simulation you will not take an accidental peek at what followed on that currency pair.
I cant simulate :
-variable spreads
-swaps
-i dont have access to CB chairman interviews ,times of war , looming war etc.
news go to the extent of prints with the previous forecast and actual numbers only.!
Alleged Pros:
-i want to be able to sit my bottom down before an NFP print , open up the simulator and have a little go at ALL the available NFP prints stored in the app's history , simulated .Something like a warm up before the match.
-i want to be able to sit and select a trading session(or timespan) and get my training on that specific timeframe.
-i want to be able to train specifically on volatile conditions and the opposite(although its kind of hindsight dangerous , meaning , you select volatile conditions training , and thus you know there will be volatility)
-i want to be able to train specifically on ranging -trending conditions (although hindsight is crucial here too)
This cannot go public due to the fact i own none of this data,and it definetely wont be for sale.
If someone from Investing.Com or DukasCopy reads this , and has an objection , they are more than welcome to tell me "Dude , please use the data only for your self".
If FPA believes this can harm their site in any way ,they are welcome to erase it(or make it private),its their website after all.
I'm thinking a local server on my spare pc which will be open for use -when i am not in need of bandwidth on my home time-is the way to go.
Also , i wont lie , if there is something i have not considered , you are welcome to contribute as this may help me develop a more immersive environment as far as "preparing for trading"/"learning" is concerned.
Thanks for reading.
