Great Breakout System with Unfinished EA | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Another great Breakout GBP/JPY System (Dachel's Simple system) 62 replies

Great Posts By Great Traders 79 replies

Average system + great discipline =great results? 44 replies

Arrow - Unfinished 1 reply

Attachments: Great Breakout System with Unfinished EA
Exit Attachments

Great Breakout System with Unfinished EA

  • #21
  • Aug 4, 2013 8:08am Aug 4, 2013 8:08am
here's what im getting with spread as current (i use "offline" MT4, and latest spread was 21 or 2.1 pip)

http://i.imgur.com/S42Hi4i.png

result is positive generally but still pretty much different with OP result.

(all settings following his settings)

Keep in mind im testing from late March 2010 until late June 2013 so maybe thats why
  • #22
  • Aug 4, 2013 8:11am Aug 4, 2013 8:11am
Quoting rsg10z
i also get positive result on testing EURUSD M30 for data from 27 March 2010 until June 2013. Keep in mind im using 1 min paid data from forextester, which they collect directly from broker like InterbankFX so i guess the accuracy is way better than any other free data (because i have compared the bar closed/open/high/low in details with my real IBFX account and everything matched perfectly) However, i was testing with spread "25", considering that i use a 5 digit broker should I put 2 or 20 there? 25 means 25 pips or 2.5 pips? i am still new...
I'm using MT4 data so your paid data is probably much better than mine, the spread setting for a 5 digit broker should be 20, I've been using 2, but that would be .2 on my broker so I'm rerunning the backtest to see if it makes a difference and I'll post the results. Current probably won't work because the market is closed and the current spread is likely to be huge.
  • #23
  • Aug 4, 2013 8:35am Aug 4, 2013 8:35am
Sometimes running tests on weekends when market not open does weird stuff to the historical data. I have experienced that myself
  • #24
  • Aug 4, 2013 8:48am Aug 4, 2013 8:48am
Quoting broketrader
{quote} Hi Jurex, try to follow this procedure. 1) Close all your EURUSD charts 2) Go to history center (press F2) 3) Open EURUSD pair (double-click) and for each time frame, double-click on it, select all records (click on first, hold shift key, scroll down and click on last) and delete them. 4) When all records on all time frames deleted, double click again on the 1 min time frame, you should have nothing there as everything has just been deleted. 5) Click on the import button and select a csv file downloaded from histdata.com you have to do so...

Great , thanks broketrader for your explanation .. ...
  • #25
  • Aug 4, 2013 9:05am Aug 4, 2013 9:05am
Quoting Jurex
{quote} Great , thanks broketrader for your explanation .. ...
You are welcome.

sheppy, weekends have no influence on historical data only on spread.

I have to say that I trust pretty much my tests as I have run dozens if not hundreds of them on another EA and that EA is now trading live since 3 months and live results are in line with tests. I mean, live results are almost the same than tests for that same period but I remain intrigued by the differences between yours and mine...

Could anyone, jaws810 or rsg10z or both, run a small test lets say for january 2013 (01.01.2013 to 31.01.2013) and post the journal here so I can compare with mine and try to look at what's going on. We could be on something here, on your side or on mine.

Thank you.
Simplicity is the Ultimate Sophistication.
  • #26
  • Aug 4, 2013 9:28am Aug 4, 2013 9:28am
hey broketrader, you sure the tick data you downloaded earlier was good enough? i only know how to download tickdata from dukascopy, is it the same with tick data you are talking about?

this is btw how i use tick data:

http://forums.babypips.com/expert-ad...-100-free.html

and i dont think weekend has anything to do with spread. Im testing the EA in MT4 terminal which I use solely for backtesting purpose (be it manual or for EA) hence why i never login with this MT4 terminal so everything stays offline and spread stays at 21 point (2.1 pip)
  • #27
  • Aug 4, 2013 9:33am Aug 4, 2013 9:33am
Quoting rsg10z
hey broketrader, you sure the tick data you downloaded earlier was good enough? i only know how to download tickdata from dukascopy, is it the same with tick data you are talking about? this is btw how i use tick data: http://forums.babypips.com/expert-ad...-100-free.html and i dont think weekend has anything to do with spread. Im testing the EA in MT4 terminal which I use solely for backtesting purpose (be it manual or for EA) hence why i never login with this MT4...
I don't need tick data, 1 min bars data is enough for me as I don't scalp but trade the 5 min time frame. This produces 90% accuracy tests which is enough for me.
Simplicity is the Ultimate Sophistication.
  • #28
  • Aug 4, 2013 10:13am Aug 4, 2013 10:13am
sorry i misunderstood. i thought that histdata.com provide tick data, but it provides the same 1 min data like the one i got from forex tester. nvm
  • #29
  • Aug 4, 2013 10:14am Aug 4, 2013 10:14am
Quoting broketrader
Could anyone, jaws810 or rsg10z or both, run a small test lets say for january 2013 (01.01.2013 to 31.01.2013) and post the journal here so I can compare with mine and try to look at what's going on. We could be on something here, on your side or on mine. Thank you.
Hi,
Just ran a quick test from Jan 01 to 31 '2013 for comparison on my Alpari UK chart.
Attached Image (click to enlarge)
Click to Enlarge Name: swing_eurusd.gif Size: 7 KB

Attached File(s)
File Type: txt 20130804.txt   496 KB | 295 downloads
  • #30
  • Aug 4, 2013 10:33am Aug 4, 2013 10:33am
Dukascopy data, 2 pips spread, everything default: Doesn't look good, which in itself doesn't mean the strategy isn't good. It just means there's a serious amount of work to be done.

@ OP: Rarely (if ever) does a strategy work in any pair, any TF, or with same settings/parameters, especially fixed sl & tp.
Attached Image (click to enlarge)
Click to Enlarge Name: 2013-08-04_16-26_Strategy Tester- Swing.jpg Size: 115 KB
"To hold, you must first open your hand. Let go." - Lao Tzu
  • #31
  • Aug 4, 2013 11:11am Aug 4, 2013 11:11am
i agree. the strategy might be good but you need flexible TP or SL, for example SL at latest low or high. that kind of thing
  • #32
  • Aug 4, 2013 12:39pm Aug 4, 2013 12:39pm
May I know what exactly the problem of the EA? Couldn't produce the same equity curve like before? I will take a look after I back home.

For the spread matter in latest MT4 built. If your broker is 5 digits, the spread in tester for you to select let said is 2, then it would be 0.2 pips or 0.00002 pips spread. If your broker is 4 digits, then it would be 2 pips or 0.0002 pips spread.

I believe a profitable system will always be profitable no matter if you use any type of data(Tick or not tick) to test. I have done many backtest, either 90% or 99% modeling. Both the equity curve are almost the same, means both head to the same direction, but the return value is different. There are some factors to affect the backtesting,
1. Spread (Fixed or variable)
2. Setting
3. Historical data without gap
  • #33
  • Aug 4, 2013 2:30pm Aug 4, 2013 2:30pm
Quoting rsg10z
i agree. the strategy might be good but you need flexible TP or SL, for example SL at latest low or high. that kind of thing
You used a 15min TF, it works best on 30min TF.
  • #34
  • Edited 3:44pm Aug 4, 2013 3:30pm | Edited 3:44pm
Here is the results for 2013 from 1/1/13 till 8/1/13 with the spread set to 20 for a 5 digit broker and default settings. The results are not good it's been a rough year for the Euro with this system but check out the results for the USD/JPY during the same time period. Due to the huge drawdown it would probably be a good Idea to run this system on several pairs.

Euro results for the year.
Attached Image (click to enlarge)
Click to Enlarge Name: report.JPG Size: 78 KB
Attached Image (click to enlarge)
Click to Enlarge Name: Graph.JPG Size: 81 KB
Attached Image (click to enlarge)
Click to Enlarge Name: Settings.JPG Size: 58 KB



USD/JPY results for the year.
Attached Image (click to enlarge)
Click to Enlarge Name: JPN report.JPG Size: 77 KB
Attached Image (click to enlarge)
Click to Enlarge Name: JPN graph.JPG Size: 71 KB
Attached Image (click to enlarge)
Click to Enlarge Name: JPN settings.JPG Size: 57 KB



That's a 15.95% profit in 7 months if you ran them both with the same sized accounts. I would probably use a higher risk percentage to increase profits.
  • #35
  • Aug 4, 2013 3:36pm Aug 4, 2013 3:36pm
Quoting jaws810
Here is the results for 2013 from 1/1/13 till 8/1/13 with the spread set to 20 for a 5 digit broker and default settings. The results are not good it's been a rough year for the Euro with this system but check out the results for the USD/JPY during the same time period. Due to the huge drawdown it would probably be a good Idea to run this system on several pairs. Euro results for the year.{image}{image}{image} USD/JPY results for the year.{image}{image}{image}

Looks about as good as a Coin Toss to me . . .
20 pips a day isn't too much to ask . . .
  • #36
  • Aug 4, 2013 11:13pm Aug 4, 2013 11:13pm
Quoting jaws810
{quote} You used a 15min TF, it works best on 30min TF.
actually i ran it on 30 min.
  • #37
  • Edited 1:44am Aug 5, 2013 1:31am | Edited 1:44am
Quoting RaptorUK
{quote} Looks about as good as a Coin Toss to me . . .
Most working systems are just a bit better than a coin toss, the MM is usually the difference between wining and losing. It makes huge gains and minimizes losses. Look what happens when you ratchet up the risk percentage a bit. In this test I didn't change any of the EA settings other than the risk percentage which I set to 5 and I used the current spread instead of 2 pips since the market is open. 2 years and 7 months turns $10K into over $198K with a high water mark over $300K. If you cashed out on 8-1-13 you would have made 18.8 times what you invested or over 10% per month compounded on your initial $10,000 investment and that's after a loss of over $100K, and if you take profit every time the high water mark is exceeded you'd probably do even better.
Attached Image (click to enlarge)
Click to Enlarge Name: Graph.JPG Size: 76 KB

Attached Image (click to enlarge)
Click to Enlarge Name: report.JPG Size: 78 KB

Attached Image (click to enlarge)
Click to Enlarge Name: Settings.JPG Size: 56 KB
  • #38
  • Aug 5, 2013 1:39am Aug 5, 2013 1:39am
Quoting rsg10z
{quote} actually i ran it on 30 min.
Sorry that was meant for the post above you, I must have hit the wrong quote button, I was commenting on Vox dei's tester report.
  • #39
  • Aug 5, 2013 4:04am Aug 5, 2013 4:04am
Quoting jaws810
{quote} Most working systems are just a bit better than a coin toss, the MM is usually the difference between wining and losing. It makes huge gains and minimizes losses. Look what happens when you ratchet up the risk percentage a bit. In this test I didn't change any of the EA settings other than the risk percentage which I set to 5 and I used the current spread instead of 2 pips since the market is open. 2 years and 7 months turns $10K into over $198K with a high water mark over $300K. If you cashed out on 8-1-13 you would have made 18.8 times...
Jaws,

Agree with you that most entries are little better than 50/50, but I'd say that it's the exits rather than the MM (sizing) which provides the edge.

IMO a profit factor of 1.2 over 7030 trades is very good.

For me, assuming consistent fixed frac sizing (i.e. X% equity risk per trade), the key measure is the
F ratio = (annual return%) / (maximal drawdown%)
Anything above 1.5 is very acceptable.

In your case:
years=2+7/12;
deposit=10000;
netpl=178937;
maxdd=61.25;
annualroe=(((deposit+netpl)/deposit)^(1/years)-1)*100;
print "annual return =", annualroe, "%";
print "F ratio = ", annualroe/maxdd
gives --------------->
annual return = 211.9307571863 79 %
F ratio = 3.4600939948 7966

Hence, given the sizing you've used, we have an annual return of 212%, max DD of 61.25%, and hence a F ratio of 3.46

Personally, I'd be more comfortable at around 1/3 of that, i.e. annual return of ~ 70% and max DD of ~ 20%. Hence I'd reduce my sizes proportionally.
But risk is a personal thing: each to their own. People trading smaller accounts tend to be willing to taker larger risks, even though there's no mathematical justification for doing so.

The above is meaningful only if consistent fixed frac sizing is being used. I'm not familiar with the MM being employed by the EA.
And I'd also want to look very closely at the modeling quality, and also the nature and effect of any curve fitting that might exist, before I considered letting the EA trade my hard earned savings.

Just my 2c, following a very quick glance at the thread.

David
  • #40
  • Aug 5, 2013 5:55am Aug 5, 2013 5:55am
Quoting ~bull.bear~
May I know what exactly the problem of the EA? Couldn't produce the same equity curve like before? I will take a look after I back home.
Just tried backtest the EA with EURUSD & GBPUSD. Both are losing in year 2004 onward. However, both are winning in year 2008 onward.
0 traders viewing now