Critique my backtesting results of this EA | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

meanVel's Trade Critique Corner 1 reply

critique my EA 18 replies

Critique my system please 12 replies

Trendline only system- Please critique 14 replies

Beginner requesting plan critique 5 replies

Attachments: Critique my backtesting results of this EA
Exit Attachments

Critique my backtesting results of this EA

  • #21
  • May 8, 2016 5:29am May 8, 2016 5:29am
Quoting FX-Flux
Sorry, i have recently added another symbol that the strategy works on. Additionally, this test includes the last couple weeks of data. here are the results with just 4 symbols {image} {image}
with limitid information/data you share here (which you not need to share more) this code seem very good
you have average 35% dd with 0.1 lot trading initial deposit $1000 tested for ~15 years on 4 symbol with 100% annual gain with basic lot sizing
if ı was you ı wouldnt touch it and forward test it with 0.05 lot with $1000 initial deposit fix lot
good work
  • #22
  • May 8, 2016 5:31am May 8, 2016 5:31am
Quoting FX-Flux
{quote} What do you think makes a good EA? Yeah i look forward to three months time, worth a gamble .
A good EA works on any major or cross pair, with the same setting. Because the pattern of EU today is the pattern of AJ tomorrow.
  • #23
  • May 8, 2016 8:52am May 8, 2016 8:52am
I don't know if you can do this in MT, but in NinjaTrader you can add a secondary data series to the backtest program so that your entries, SLs and TPs work on the secondary series (I recommend 1-minute bars, not tick), even though the strategy is based purely on higher timeframes such as daily bars.

You still have to watch out for ambiguities, such as entry and/or SL and/or TP being triggered on the same 1-min bar in the case of high-impact news events, but as long as you report them and treat them with a negative bias then your backtest results will be fairly accurate and won't give you any false hopes.

Also, this probably goes without saying, but don't forget to include commission and slippage costs in the backtest. These have significant impact on the P/L, unless your backtest doesn't take many trades. But then... how can you be sure it's accurate based on such a small sample size? There are no easy answers unfortunately, and it takes a lot of work to produce a useful backtest (in my opinion).
You reap what you sow.
  • #24
  • May 8, 2016 11:11am May 8, 2016 11:11am
Quoting NorthTrader
I don't know if you can do this in MT, but in NinjaTrader you can add a secondary data series to the backtest program so that your entries, SLs and TPs work on the secondary series (I recommend 1-minute bars, not tick), even though the strategy is based purely on higher timeframes such as daily bars. You still have to watch out for ambiguities, such as entry and/or SL and/or TP being triggered on the same 1-min bar in the case of high-impact news events, but as long as you report them and treat them with a negative bias then your backtest results...

Yeah you can do that on MT4, you set your EA to do calculations on higher timeframes and then feed it lower timeframe data.

Quoting mvp77
{quote} A good EA works on any major or cross pair, with the same setting. Because the pattern of EU today is the pattern of AJ tomorrow.
Well i thought i would try my strategy on XAUUSD, all i did was change the TP and SL to suit the point value of Gold, and the results for entire history available are as follows (No optimisation required) fixed 0.1 lot size, 1000$ starting:
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Screen Shot 2016-05-08 at 16.54.59.png Size: 89 KB
Click to Enlarge Name: Screen Shot 2016-05-08 at 16.54.51.png Size: 87 KB
So many in the Commercial section want to make me rich! :o
  • #25
  • May 8, 2016 12:42pm May 8, 2016 12:42pm
From what i read from your post ....You are doing Breakout trading.
Now , if your method is entering in Intrabar ( that is in current bar without waiting it to close. ) ...than results could be significantly different than actual results.

For example ...kind of back testing using breakout methods for entering trades during high volatility news.
Its one of the biggest hurdle for testing this kind of systems in MT4.

If you strictly used OHLC data in your backtesting ....than you definitely have edge. ( Result might not be exactly same ..but your expectancy is good enough to make you money. )

If you are doing above mentioned Intrabar back testing ..my suggestion would be for you is to create your strategy in jforex platform and back test with tick data with actual historical spreads. It will allow you to test your strategy as realistically as possible and only thing left to be accounted will be latency of trades. Which is not a big issue if your strategy is not too short term ..like scalping or HFTs .Also this kind of slippage works both ways if your broker is genuine .
( Assuming you will not be trading big enough lots anytime soon to factor slippage due to size ).
  • #26
  • May 8, 2016 12:46pm May 8, 2016 12:46pm
Quoting FX-Flux
{quote} Yeah you can do that on MT4, you set your EA to do calculations on higher timeframes and then feed it lower timeframe data. {quote} Well i thought i would try my strategy on XAUUSD, all i did was change the TP and SL to suit the point value of Gold, and the results for entire history available are as follows (No optimisation required) fixed 0.1 lot size, 1000$ starting: {image} {image}
after seeing code worked on gold on first time tested.ı have now dubts
  • #27
  • May 8, 2016 12:55pm May 8, 2016 12:55pm
I programmed the EA to do calculations on higher time frames and use 1m OHLC for testing.

I find that as this is not looking to take small profits or losses it is perfectly acceptable to use the control point method for backtesting. I think most of the time the difference between the TP and SL is about 120+ pips, and so there would never be any uncertainty to weather a 1m bar was 50/50 going to take out either the stop or profit level, the price would have to travel a good distance before hitting either.


Ive tested with tick data and found there to be barely any difference in the results.
So many in the Commercial section want to make me rich! :o
  • #28
  • May 8, 2016 12:55pm May 8, 2016 12:55pm
Quoting vasily
{quote} after seeing code worked on gold on first time tested.ı have now dubts
and why is that surprising? haha
So many in the Commercial section want to make me rich! :o
  • #29
  • May 8, 2016 1:02pm May 8, 2016 1:02pm
Quoting FX-Flux
{quote} and why is that surprising? haha
if same algorithm work on usdjpy eurjpy usdchf eurusd xauusd with no diference and say 10 years there is something wrong with it
  • #30
  • May 8, 2016 1:07pm May 8, 2016 1:07pm
Quoting FX-Flux
{quote} and why is that surprising? haha
equity line (green) always below balance line and margin spikes save those losers.so you probably saving losers with high lot size trades.correct?
  • #31
  • May 8, 2016 1:15pm May 8, 2016 1:15pm
Quoting vasily
{quote} equity line (green) always below balance line and margin spikes save those losers.so you probably saving losers with high lot size trades.correct?
No martingale or anything like that. I believe those may be trades that were mainly all positive and then got stopped out or the other way round - i don't believe all trades go in the desired direction straight away, they may go negative before closing.
So many in the Commercial section want to make me rich! :o
  • #32
  • May 8, 2016 1:19pm May 8, 2016 1:19pm
Quoting FX-Flux
{quote} No martingale or anything like that. I believe those may be trades that were mainly all positive and then got stopped out or the other way round - i don't believe all trades go in the desired direction straight away, they may go negative before closing.
ok.ı dont specualte more with limited information.greenpips
  • #33
  • May 8, 2016 1:28pm May 8, 2016 1:28pm
Quoting vasily
{quote} ok.ı dont specualte more with limited information.greenpips
Thank you for your input :nerd:
So many in the Commercial section want to make me rich! :o
  • #34
  • May 8, 2016 1:35pm May 8, 2016 1:35pm
oh yes EAs
Attached Image (click to enlarge)
Click to Enlarge Name: image.jpeg Size: 73 KB
Learn, a forex trader must, unlearn and relearn he will.
  • #35
  • May 8, 2016 1:48pm May 8, 2016 1:48pm
Quoting COGSx86
oh yes EAs {image}
Yes, use compounding lot sizes for dramatic effect Guilty. however, it gets people talking.
So many in the Commercial section want to make me rich! :o
  • #36
  • May 8, 2016 8:57pm May 8, 2016 8:57pm
Quoting FX-Flux
{quote} Yes, use compounding lot sizes for dramatic effect Guilty. however, it gets people talking.

and fixed lots ? (same period)
Attached Image (click to enlarge)
Click to Enlarge Name: fixed lots.png Size: 23 KB
Learn, a forex trader must, unlearn and relearn he will.
  • #37
  • May 8, 2016 9:21pm May 8, 2016 9:21pm
Quoting FX-Flux
Sorry, i have recently added another symbol that the strategy works on. Additionally, this test includes the last couple weeks of data. here are the results with just 4 symbols {image} {image}
Looks nice, but its hard to tell how you would have handled the period from 2004-2008. That's pretty much one computer cycle (you'd likely need to buy a new computer to run the EA after the 4 years) with pretty much no increase in equity. You'd need to be able to push through those times and all the while stay positive about the EA.
  • #38
  • May 8, 2016 11:30pm May 8, 2016 11:30pm
Quoting Sciurus
{quote} Looks nice, but its hard to tell how you would have handled the period from 2004-2008. That's pretty much one computer cycle (you'd likely need to buy a new computer to run the EA after the 4 years) with pretty much no increase in equity. You'd need to be able to push through those times and all the while stay positive about the EA.
Very good point, Sciurus. If you were unlucky enough to switch it on at the start of 2004, you would have had about two years of downtrending equity. Not many people would have kept it going after those first two years, despite the backtest looking good just before you switched it on. "Something's changed.", "It must be broken", and so on. And even if you had kept it going for another two years, your equity would be pretty much back to where it was at the start of 2004. And this is using several instruments!

The scurge of trading robots. These things can drive you mad if you're not careful.
You reap what you sow.
  • #39
  • May 9, 2016 2:02am May 9, 2016 2:02am
Quoting NorthTrader
{quote} Very good point, Sciurus. If you were unlucky enough to switch it on at the start of 2004, you would have had about two years of downtrending equity. Not many people would have kept it going after those first two years, despite the backtest looking good just before you switched it on. "Something's changed.", "It must be broken", and so on. And even if you had kept it going for another two years, your equity would be pretty much back to where it was at the start of 2004. And this is using several instruments! The scurge of trading robots....
You're right. Would you say this is the same for trading manually though? If you started trading and didn't make any money for 4 years you would probably stop. As for testing this EA I should probably have a look at 2004-08 and see what it was doing.
So many in the Commercial section want to make me rich! :o
  • #40
  • Last Post: May 9, 2016 6:42am May 9, 2016 6:42am
Quoting FX-Flux
{quote} You're right. Would you say this is the same for trading manually though? If you started trading and didn't make any money for 4 years you would probably stop. As for testing this EA I should probably have a look at 2004-08 and see what it was doing.
I wouldn't say that not being able to stop is the problem. Stopping or changing tactics is easy, regardless of whether you are using an EA or trading manually. But your backtest did not test how altering your strategy based upon most recent performance affects the equity curve, so without that data we don't know if it would be more effective than just staying the course.

As is, even with the 4 year flatline, I can see using the strategy as a long-term investment tool. Think retirement, not income. The other thing you can do is to look at the performance by Month. Really dig into it to the monthly data to see if you are comfortable with the variation in the monthly performance. Then ask yourself how much you are willing to invest in the strategy knowing based on the negative months.
1 trader viewing now