What works, and what doesn't | Page 3 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

variety of resources what works and what doesn't 98 replies

EA works on one broker, doesn't on another 4 replies

If it works, it works... 3 replies

EA Works, Then Doesn't 1 reply

Why an EA works for some currancies and not others? 10 replies

Attachments: What works, and what doesn't
Exit Attachments

What works, and what doesn't

  • #41
  • Oct 5, 2015 7:55am Oct 5, 2015 7:55am
My system isn't about following a trend.. Its about playing volatility. Shorting pops and and buying drops, and playing the liquidity and lack of liquidity.
The only system that will work is one designed by and for yourself.
  • #42
  • Oct 5, 2015 8:07am Oct 5, 2015 8:07am
Quoting hanover
To whatever extent backtesting is impotent because markets change their color over time, and hence past data doesn't reflect what will happen henceforth
Markets are always changing, true. Backtesting allows one to very bluntly determine the versatility of a system while estimating risk vs reward.
Quoting hanover
... and also systems that are simple enough to be coded into an EA. Between 2009-2012 I coded maybe 20-30 EAs by handpicking the seemingly best ideas from the dozens of requests I received from (in some cases, the most highly followed) FF members, who all insisted that their strategies were profitable. When this proved not to be the case, I modified these EAs, adding their suggestions for filters to eliminate entry signals that resulted in losses, different SLs, TPs, trailing stops, MMs ..... you name it. As far as I know, NONE of these EAs are...
Im unfamiliar with 'EA's but im very familiar with mechanical systems. Generally, simpler rules are better than complicated ones. This is because overcomplicated systems tend to be more 'sensitive' to variances in data and profitable results have a higher likelihood of being due to curve-fitting.

A 'simple' EA being unprofitable is more likely due to flawed design than lack of complexity.
Quote
No matter how you slice it, an experienced human trader will always have more intelligence than even the most sophisticated EA. And, moreover, it stands to reason that if the markets could be conquered by a single, objectively definable system, that everybody would use such a system.
No, a mechanical system is still only a belief about markets put into a standardized tradable form. You can have as many different systems as you have people with different beliefs. Systems do pretty well in providing a framework for managing risk and letting one put his human decision-making abilities to better use at the portfolio level.
  • #43
  • Oct 5, 2015 8:26am Oct 5, 2015 8:26am
Quoting FXEZ
3) An adaptive component in each system to allow the fixed system rules to change over time and to thus generalize better.
I agree that good systems have a dynamic component that shifts with market character. However, the idea of machine learning that you allude to is a flawed concept in trading system design.

Having system rules and degrees of freedom that change all the time; one cannot confidently determine the system's robustness even if the backtest proved profitable.

If you happen to be trading on such a system and doing well, then ride your winner. Perhaps you could also pay attention to the fact that such a system does well when it is exposed to strong trends and a rote algorithm that simply rides such trends would also do fine.
  • #44
  • Oct 5, 2015 8:41am Oct 5, 2015 8:41am
Quoting Atokys
{quote} I agree that good systems have a dynamic component that shifts with market character.
Do you think the market changes its character gradually or are there distinct shifts at points in time? For example a major news event.
  • #45
  • Oct 5, 2015 8:48am Oct 5, 2015 8:48am
Quoting BigOldFella
{quote} Do you think the market changes its character gradually or are there distinct shifts at points in time? For example a major news event.
It can be literally anything. Even the term 'character' that we use to ascribe a quality to market activity is subjective and different people will perceive this shift in a different way from their personal trading metaphor.

This doesn't only apply to individual markets but also to the correlations between markets.
  • #46
  • Oct 5, 2015 9:11am Oct 5, 2015 9:11am
Quoting bloodpoodle
{image} Todays trade.. Thin run up = weak.. Hit area of liquidty which was cap from NFP and came right back down. odds were in my favor.. quick run up was weak.. and it was overbought.. Sell high.. don't buy high.
What is 'thin run' ? you mean fast price move up ?
  • #47
  • Oct 5, 2015 10:18am Oct 5, 2015 10:18am
a thin quick run up leaves a gap in liquidty... nobody puts orders in the middle of huge candles.. So price falls right back down to the base of the big candle.
The only system that will work is one designed by and for yourself.
  • #48
  • Oct 5, 2015 10:24am Oct 5, 2015 10:24am
Quoting MathTrader7
{quote} Interaction of bulls and bears is the essential concept based on which price moves up/down; you are using this market concept to analyze a price chart in order to find entry/exit points. How did you get 75% success rate for this analysis?

Hey MT7,

Not to highjack the thread here but can you please revisit the RENKO thread periodically. I feel the strategy has potential but I am running into some minor issues with the EA you created. Thanks in advance.
  • #49
  • Oct 5, 2015 11:05am Oct 5, 2015 11:05am
Quoting dkrock
In a trending market, momentum shifts the market gradually.
I notice that your definition for a 'trending market' doesn't include trending scenarios marked by extreme directional volatility and range expansion.
  • #50
  • Oct 5, 2015 11:19am Oct 5, 2015 11:19am
Quoting dkrock
{quote} Is that another definition of "trend"? I have no idea what you are trying to complicate things. If the market is not reversing for profit, then it is in a trend. Duh.
Well, I guess I shouldn't judge your definition. 'Trends' take many shapes and forms to me. Go back long enough on a daily price chart and you will see anything and everything.
  • #51
  • Oct 5, 2015 12:07pm Oct 5, 2015 12:07pm
Quoting dkrock
{quote} I trade momentum, so for me, it is super simple. If you trade lines, then it is probably complicated. I don't know. I don't trade lines. It is 2015. I don't trade daily price charts either so I will assume you have seen anything and everything. Like I already said, a time frame is a time frame. Knowing that tidbit is the second half of the secret to success. In this screenshot, I would call trades ONE, TWO, and FOUR range trades, and trade THREE a trend trade. {image} Your mission, should you choose to accept, is to figure out how I know...
So when do you enter and where is your sl? and when exit position ?
  • #52
  • Oct 5, 2015 12:29pm Oct 5, 2015 12:29pm
I can assure you trendlines do work its how you see them according to correct market structure. 99% of traders in the market do not know where the pressure points are just like on your own body majority of people don't know where they are.

Gann Boxes and arcs are my strongest indicator after price action extreme points.

You need to have a holistic view on the markets.

I find daily candles the best to interpret it rallies 35-65 points during previous days NY and today Asian session maybe 1-2 hour of london open before a reversal or on the news time.

Hardest part of trading is cutting your losses and letting your profits run for the full adr.
Find your Favourite Pattern
  • #53
  • Oct 5, 2015 12:37pm Oct 5, 2015 12:37pm
Quoting bloodpoodle
My system isn't about following a trend.. Its about playing volatility. Shorting pops and and buying drops, and playing the liquidity and lack of liquidity.
Quote people so they know who you're responding to.
  • #54
  • Oct 5, 2015 1:03pm Oct 5, 2015 1:03pm
Quoting bloodpoodle
What does work... 1. Market structure... 2. Liquidty Reading liquidty...
Also, in addition to the forms of market structure (or liquidity, whatever floats your boat) trades you've already listed, consider Big Round Numbers in your analysis (1.1200, 1.1300, 116.00, 117.00, etc). I've read in a research paper a while back that somewhere around 25% to 40% of all market participants place their orders anywhere in a 20 pip range around BRNs. Often times their stops are also within that 20 pip range. I've used this info to determine some of my targets and entries with good success. I read this at least 5 years ago and it still to this day contributes to my successful entries and exits.

Attached Image (click to enlarge)
Click to Enlarge Name: A showcase of BRNS.png Size: 81 KB


Of course it takes analysis to see if this adds value to your methods of trading.
  • #55
  • Oct 5, 2015 1:35pm Oct 5, 2015 1:35pm
I developed many EA's in 5 years, some of them by my analysis , some of them by other strategies, i tried machine learning and many of indicators, patterns , news trading but any of them do not work. Market is highly impacted by traders psychics and so technical indicator may be good signs but you will found that most of time does not. You can take technical indicators as "advice" but in my experience best entries which i made was by my analysis and opinion. Market is every time in different conditions so "hard code" entries in EA is almost impossible. But what i found, and it makes me profitable is EA which is managing risk, exits and managing stoplosses. But these stoplosses and strategy is not hardcoded. I will not post how my EA works in detail because it would be really long thread, but in summary i put trade by myself with some parameters and EA is managing stoplosses / lots, trailing stops and etc by myself. With this strategy i can isolate my emotions psychic and constantly taking loss/profits on some predefined conditions. Most failed orders from me was where i thinked it will "turn" good , and i was movin stoploss, or i was reloading to "hope" that market turns. One thing i learned is that hope does not work, if you are already in open trade , only thing that works are strict rules when to take stop loss , when to take take/profit, when to start breakeven SL and so on, and best technique is let to do machine this. Machine does not have emotions , feelings, just strict rules. This can be achieved by yourself but its hard to be 24/5 inside metatrader, when you need to put BE SL in the middle of night. So do not focus how to make EA that can open trade good, but focus on EA which is isolating your bad habits. EA that is thinking instead of you does not exists. If you dont know "something" about trading , you can barely make profitable EA. So from my point of view best and sucess strategy is make what you have (brain) manually, and eliminate what you dont have (perfect managment on your emotions and psychic) by machine. Machine can support you , but i dont think that you can force it to think instead of you.
  • #56
  • Oct 5, 2015 2:00pm Oct 5, 2015 2:00pm
Quoting DrakeFX
One thing i learned is that hope does not work, if you are already in open trade , only thing that works are strict rules when to take stop loss , when to take take/profit, when to start breakeven SL and so on, and best technique is let to do machine this. Machine does not have emotions , feelings, just strict rules. This can be achieved by yourself but its hard to be 24/5 inside metatrader, when you need to put BE SL in the middle of night.
Great point.
  • #57
  • Edited 2:17pm Oct 5, 2015 2:03pm | Edited 2:17pm
Quoting FXEZ
As recently as 5 years ago I probably would have taken exception with this statement.
FXEZ,

None of the EA requests that I received included anything remotely like the 3 concepts you described. Many simply incorporated indicator based entries, some support/resistance, and/or scaling in/out in grid-like fashion. And some involved recovery trading and martingale variants.

I also wrote an EA for an ex-bank trader who nowadays trades for a living. She is very profitable; the EA is not (at least not without human intervention). The human intervention part involves a nuanced view of the markets, including some fundamental analysis, and items that simply can't be coded (e.g. assessing whether a Yellen speech is more hawkish or dovish than the market was anticipating). Hence my conclusion that an experienced human trader will almost certainly outperform all of the EAs requested on trading forums like this one.

Many thanks for sharing your knowledge. I learn a lot from your posts. My knowledge of things like algorithmic trading, machine learning, and advanced statistical modelling is not great (and I've not read any of the books like those that Copernicus listed). Another reason why I've gravitated more toward 'discretionary' rather than 'mechanical' approaches.

David
  • #58
  • Oct 5, 2015 2:07pm Oct 5, 2015 2:07pm
Quoting hanover
{quote} FXEZ, None of the EA requests that I received included anything remotely like the 3 concepts you described. I also wrote an EA for an ex-bank trader who nowadays trades for a living. She is very profitable; the EA is not (at least not without human intervention). The human intervention part involves a nuanced view of the markets, including some fundamental analysis, and items that simply can't be coded (e.g. assessing whether a Yellen speech is more hawkish or dovish than the market was anticipating). Hence my conclusion that an experienced...
Yes this is exactly my point , human intervention is a must to EA. You cant hard code everything on every condition to EA. As i mentioned in my previous post, EA can help you a lot to not make you slave of your trades when you sleep , or living normal life. But cant take part in decisions, if this would be possible every 3rd programmer in world will be millionaire with some EA
  • #59
  • Oct 5, 2015 2:19pm Oct 5, 2015 2:19pm
Quoting Atokys
A 'simple' EA being unprofitable is more likely due to flawed design than lack of complexity.
Yes, I suspect that none of the EAs I was referring to had a genuinely robust 'edge'.

Quoting DrakeFX
{quote} Yes this is exactly my point , human intervention is a must to EA. You cant hard code everything on every condition to EA. As i mentioned in my previous post, EA can help you a lot to not make you slave of your trades when you sleep , or living normal life. But cant take part in decisions, if this would be possible every 3rd programmer in world will be millionaire with some EA
Agree , but I think we must also acknowledge that contributors like Atokys, FXEZ and Copernicus are talking about different concepts -- e.g. algorithmic trading, machine learning, statistical modelling -- concepts that go way beyond the average forum trader's worldview of technical analysis (indicators, line studies, patterns, levels, etc) and possibly fundamental analysis (macroeconomics, fair price, analysis of news etc).
  • #60
  • Oct 5, 2015 4:39pm Oct 5, 2015 4:39pm
Quoting hanover
e.g. assessing whether a Yellen speech is more hawkish or dovish than the market was anticipating
Hi David,
I have 3 idea to write EA.
Sadly I can't write down the rules for exit conditions for EA
In another trade system, the idea owner can't tell either in words, which are translatable to MT4.

My idea is to simulate or skip the news parsing it is to write an indicator-like engine in EA, which will indicate "to many ticks on this M1 bar, so probably is a heavy bull-bear fight stay out a bit" or "to many pips in this M1 bar already, so it is a high impact news, because or the bull or the bear side is missing from market"
I wrote an indicator "Volatility speed indicator", I just should add the "skip candles with open=close price to not make noise.
Those candles which has MathAbs( Close-Open) much higher then the average ( 3-6x) are the news spikes mostly.
Sadly I don't know what to do if I found a candle like that, because sometimes they are trend starters and no real pullback to any Fibo, and other times more then 70% pullback if not 100%.

For me the EA role it isn't that one to catch 50 pip from a 100 pip movement and get quick rich, but to make money with my minimal intervention. If it catch 10 pip pullback after 100 pip movement, that is perfect for me.
Like I would say "sell HH from now 20h or while I cancel" or "buy LL 20 days, when it happens" , because I go to sleep / holiday.

Also there it was a post in this thread: the EA should close without fear and hope the positions.
Sadly very probably when I wake up I will see the EA close in loss 2h ago and now it would be in major profit and I will stop the EA, but this happens with manual trading too.
Attached Image (click to enlarge)
Click to Enlarge Name: Backtest.png Size: 1.9 MB
I want to buy PATIENCE
1 trader viewing now