Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

VennD - Any Pairs - Any Style - Any Time 740 replies

Need backtesting results for this system 6 replies

No backtesting results 0 replies

Backtesting Results Attached - Not Selling Anything 4 replies

Backtesting results with different timezones 5 replies

Is backtesting any good? Ridiculous Results

  • Post #1
  • Quote
  • First Post: Edited 12:59pm Jul 8, 2006 12:43pm | Edited 12:59pm
For starters i use metastock profx as my software package to back test information. I made an expert advisor for my strategy and backtested it on a 5 min interval which i usually use. I set a 10 pip take profit and a 20 pip stoploss. The backtest was for the last 4 months and i tested the main pairs.
(GBP/USD, EUR/USD, USD/JPY, AUD/USD, EUR/JPY) The results turned out to be ridiculously good.(I'm embarrassed to even say the number of total pips because i will be put into question right away) However the win % was 70%. The total number of trades was also ridiculous around 15k trades in that time frame. Is this any reliable because this strategy worked for me but i realized i couldn't watch everything at once so bought this software just so i get alerts and the fact that i didn't want to learn the MT4 language and didn't really want to give my strategy away and Oanda's API is too expensive? Feedback appreciated.
Quoting ProdiG
For starters i use metastock profx as my software package to back test information. I made an expert advisor for my strategy and backtested it on a 5 min interval which i usually use. I set a 10 pip take profit and a 20 pip stoploss. The backtest was for the last 4 months and i tested the main pairs.
(GBP/USD, EUR/USD, USD/JPY, AUD/USD, EUR/JPY) The results turned out to be ridiculously good.(I'm embarrassed to even say the number of total pips because i will be put into question right away) However the win % was 70%. The total number of trades was also ridiculous around 15k trades in that time frame. Is this any reliable because this strategy worked for me but i realized i couldn't watch everything at once so bought this software just so i get alerts and the fact that i didn't want to learn the MT4 language and didn't really want to give my strategy away and Oanda's API is too expensive? Feedback appreciated.
Once I'd be positive that I didn t deceive myself through my coding or through MStocks backtesting procedures, one of the 1st thing I would look at is the Average Pip per Win. If it is low, it s unlikely that the strategy would be a winner on the long term. And...does the test include the spread? With ~15k trades, forgetting spread will make a huge difference...
As far as avg winning pips the avg is 12(Assuming spread is 2) and avg loss is 18 pips(since im already down 2 when entering). I adjust for the spread according to pair but spreads are generally between 2-3 on oanda anyway. As for the coding i checked it and it is correct.
Back testing can give you an idea of results, but I have found there are inconsistancies, so authenticate your data with forward testing on a demo account....JMHO
Same Whore .... Different Dress
Let it run on a demo for a few weeks, on several currency pairs, trending and ranging market. If you picked a single currency pair your results might be biased. That would mean that effectively, you just picked the best results while ignoring the ones that didn't meet your expectations. That happens all the time.

Anyway, a few weeks should show you if your strategy is any good. The longer the better. Let us know how it went.

Good luck!

Dane
Well I already use this strategy for trading and usually scalp 5-10 pips everytime. This usually works a couple of times a day and it basically works during more calm times, i have a separate strategy for news releases. I was just amazed at the results i got after backtesting it. My problem was i needed alerts cause i can watch 6 pairs at once and would only make like 20 pips a day when i could make more. I was wondering if it might be a good idea to set up an API for this formula and have it trade like 1 mini lot?
Quoting ProdiG
Well I already use this strategy for trading and usually scalp 5-10 pips everytime. This usually works a couple of times a day and it basically works during more calm times, i have a separate strategy for news releases. I was just amazed at the results i got after backtesting it. My problem was i needed alerts cause i can watch 6 pairs at once and would only make like 20 pips a day when i could make more. I was wondering if it might be a good idea to set up an API for this formula and have it trade like 1 mini lot?
I think watching more than 4-6 pairs is counterproductive, since they are so corelated. You might be better off just increasing your trade size on the pairs that you are actually able to watch. With more signals you will be more prone to mistakes, as - as you said - the EA gives you big number of trades.

Just my 2 pips.

Dane
what i realized trading this is that there will be alot of false alerts since the metastock datafeed is much different to that of brokers. so i guess autotrade wont really be an option. i even modified it and got even more alerts so no way can i have over 20k alerts go off and trade automatically unless its like one mini lot but the results were also better. but i learned one thing when it gives an alert and oanda chart agrees get in right away.
Hows the forward testing going?
Same Whore .... Different Dress
  • Post #10
  • Quote
  • Last Post: Jul 11, 2006 12:12am Jul 11, 2006 12:12am
I couldn't agree with Dane more. Bring the amount of pairs down. Heck, anything that is dollar-based pretty much all moves the same when one of them move a lot, so if you have four positions open on a dollar pair, you could lose 4x...

Just my experience... I scalp one pair and that's EUR/USD because I need the lowest spread I can get.
Mr. Trend
0 traders viewing now