Holy-Grail!! At-Last - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I use indicators macd, cci, stochastic, vagor and bollinger bands. All indicators are used in trend mode as even bollinger bands does not pass overbought/sold in backtests optimizing. I've tested all indicators that comes pre-loaded on mt4 with 1hr time-frame and cannot get others to meet my criterion. I'm not saying indicators don't work on 30m,15m or even 1m data, its just not practical time-wise for me to optimize. And even when I tried in the past, the results were not satisfactory.
I also employ counter-trend, gaps and scheduled news releases which are Not based on indicators. The high win-rate you're looking at in that old back test is because the gap system and bollinger bands systems were trading more than once per bar. Because their take profits are short, they're closing out and reopening themselves, hence the high win-rate. I've since learn how to fix that problem which helped the profit factor and draw-down but killed the win-rate. Also somehow, my test data since then got changed as the one time very attractive gap system is not doing so well in resent times.
I'm currently working on fixing the gap system by letting it wait for the cushion before the gaps starts to close. Next project is writing a logic into the program to close opposing order if it's a Sell and the new orders are Buy, then close the old order. I don't know if this would help or kill the system but should act like a filter. If anyone knows of a good filter or combination of filters, please let me know.
profit factor of 1.25 with a drawdown of only 10% is damn good i hope it continues giving results like that in the future too
SDC 2010.07.02 06:09
profit factor of 1.25 with a drawdown of only 10% is damn good i hope it continues giving results like that in the future too
Thanks SDC, my original intend was developing a system with a profit factor of 2.0. The article I was going by said pf of 1.50 or better with max draw-down <20%. We should always be looking for ways of making better systems but if I cannot break the 1.50 barrier, No harm, No foul, I'll just polish what I've got and dive in live.
marcoblbr 2010.07.02 05:03:
interesting... so are the buy / sell orders based on any indicator or you created a new method ? i'm building my ea, but i couldn't find a good method to open orders. i can use money management to lower the losses and i know the moment to close positions based on a trailing stop (moving stop loss) i developed, but it is still really hard to know the right time to open positions!
Well, I've explained what my system is made of. As far as the time-to-open, what I've observed is most of my optimization coincide with key support/resistance lines. IMO this is because of the Average Movement/Behavior the EUR/USD its optimized on. For example lets take my favorite MACD. If you try to use the Author of that indicators default setting on forex, I can almost bet its gonna fail. Those coordinates were generated to coincide with stock market support/resistance.
What I'm trying to say is Good entries in my opinion are when price break out of Key support and resistance levels for trend followers. And for counter trend when price bounce off those levels. If you don't want to sit at your computer drawing good support and resistance levels then use an indicator. Again this is all just my Observation.
ubzen:
I just put this out for fun and teasing purposes ...
one thing i don't understand is why people always complain about the strategy tester in mt4 ? why it isn't reliable and is there a way to make a test and make sure it would run exactly the same in a live account (using mt4 or mt5) ?
i know that in a broker the values could be a little bit different, but supposing the broker gives the values of the strategy tester, shouldn't the results be right?
one thing i don't understand is why people always complain about the strategy tester in mt4 ? why it isn't reliable and is there a way to make a test and make sure it would run exactly the same in a live account (using mt4 or mt5) ?
i know that in a broker the values could be a little bit different, but supposing the broker gives the values of the strategy tester, shouldn't the results be right?
If it's to make sure programatically the EA does what it's supposed to do, then fine.... but many use it as a hope generator.... curve fit EA's to known market movements, get an acceptable result and then wonder why it's not giving the same results when live. Of course the tester can't help with lag, persistance, variable spread etc.
V
ubzen, people are beginning to take you seriously, it's time to come clean and stop the "fun and teasing".
engcomp, I am coming clean ;). But I'm also being very honest. Most of the information I'm giving away is public knowledge. I'm always the first to tell everyone I'm a Forex newbie and have No real trading experience. This outlet is kinda like my journal, workspace and education source. If someone in real-life ask me "oh, you wanna be a fx-trader, who do you know that fx-trades" (i actually got that question yesterday). I can always say engcomp :) lol.
I'm leaving hints of what I've got here and there so that other people may also feel free to do the same. My systems are by no means satisfactory by my own standards. If I cannot put real money on it, I doubt someone else would trade it for 5 months, be at a loss and still continue to trade it.
one thing i don't understand is why people always complain about the strategy tester in mt4 ? why it isn't reliable and is there a way to make a test and make sure it would run exactly the same in a live account (using mt4 or mt5) ?
i know that in a broker the values could be a little bit different, but supposing the broker gives the values of the strategy tester, shouldn't the results be right?
Here's my take on the whole Curve-Fit___No-Good Tester dilemma. These are just my Opinions, draw your own conclusions. I'm gonna make 2 points to begin with. 1) No system curve-fitted or non-curve fitted should be used as a Hope generator. 2) Meta-trader has a bug in Back-testing which could allow 100% win testing. I ran into that bug last week and didn't even bother to post the results as a prank because that would have just not been right.
Starting with point #1, over the past 3 months of me studying Eur/Usd, I've noted that charts only have 6 elements to work with in Forex currently. Price(H-L-O-C), Volume, and Time. There's no fundamental information like supply and demand here. In other markets, people use open-interest alongside volume to tell them allot of information. Therefore, you can discount Volume in forex because it only tell you how active people are and Not if they are active Bulls or Bears. Time is another data that can be safely discounted (unless you playing the interest rate game) here in forex. Other markets use Time indicators because they have expiration dates, delivery dates and Interest rate date-time. Again here in forex time just like volume is only good for telling you When everybody is ready to play Not the direction.
So what are we left with, good old Technical Analysis in it's native form Price(H-L-O-C). My conclusion on TA is that it's a science of Pattern and Probability. The pattern and behavior of price in Gold or Stocks are not likely to be the same as the one's found on EUR/USD and as such the EUR/USD is not likely to resemble the EUR/GBP hence the need to Optimize. Those people who would not like my approach usually fall into two camps, a) those who don't like curve fitting or b) those who don't like losing.
The A) types are usually hard numbers Mathematically Inclined/Interest Rate/Arbitrage along those lines type traders. They could put there formulas on the table and tell you if 1+1 does not =2 then they gonna make a profit. But even these god-like traders cannot feel invincible because markets corrects 1+1 !=2 fast. If the market doesn't it'll be broken and even non math types would ketch on. The B) types are those who set No-Stop-Loses/Hedge-same-Currency and when they lose, they Martingale Stack their positions knowing very well the markets cannot keep going up-nor-down forever. This type of strategy requires huge reserves and nerves of steel. These guys too shouldn't feel invincible for obvious reasons. Humm... I forgot where I was going with this lols.
Anyways, we all have Risks as we (a, b or c types) are betting on a pattern that worked in the past to continue working. The point I'm trying to make is, if the back-tests are bug free, meaning it's doing what you intend && it's going to do what you intend in Live trading, Winning or Losing in Live trading becomes irrelevant for validating the back-testing results. See... the pattern changed but your system didn't. You're no different from A)type or B)type and shouldn't have been counting your eggs before they hatched.
For all the dis-trust about back-tester, let me close by saying "I've never created a system in back-tester which didn't perform in DEMO testing the way it did in back-testing". Ok, for one exception, the Bug. Now, I'm not Naive-enough to think that back-testing environment is the same as Live testing environment. I cannot stress this point enough, If your broker re-quotes you, fails to close your orders at stops, freezes up when volatility rises...the list goes on. So... you Optimized and Demo tested in an ideal environment but when you Live test the Pattern changes... what do I call that, A Bug. All this in my humble opinion.
Therefore, you can discount Volume in forex because it only tell you how active people are and Not if they are active Bulls or Bears.
[...] 2) Meta-trader has a bug in Back-testing which could allow 100% win testing. [...]