No more BS - What have you learned? | Page 30 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

What have you learned? 231 replies

What I have learned trading & How I developed my system 6 replies

It's been 3 years what have I learned? 82 replies

What Have I Learned Since 2006?? 1 reply

Attachments: No more BS - What have you learned?
Exit Attachments

No more BS - What have you learned?

  • #581
  • Sep 23, 2010 12:20pm Sep 23, 2010 12:20pm
Quoting pipmutt
Sounds like a mechanical strategy to me and you're never going to get comparable results, TA indicators are notoriously lagging.

I'm sure there are thousands of simple automated strategies plodding away producing mediocre returns, I guess it depends on what you want out of the market and how much time and effort you're prepared to invest.

FXCM's just enabled EA's with C# from the look of it, might give more flexibility than Meta 3.51 I was writing it with years back.

I hate C though, not something I'd dev in for fun.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #582
  • Sep 23, 2010 1:16pm Sep 23, 2010 1:16pm
Quoting Turveyd

FXCM's just enabled EA's with C# from the look of it, might give more flexibility than Meta 3.51 I was writing it with years back.

I hate C though, not something I'd dev in for fun.
No good talking to me about programming, I'm an IT dummy, doing a spreadsheet in Excel is about my limit!

Anytime I need anything written for MT4 I commission someone from Eastern Europe, they're pretty good at this stuff and amazingly cheap!

Mind you the language barrier can be a problem sometimes!
  • #583
  • Sep 23, 2010 1:44pm Sep 23, 2010 1:44pm
Quoting pipmutt
No good talking to me about programming, I'm an IT dummy, doing a spreadsheet in Excel is about my limit!

Anytime I need anything written for MT4 I commission someone from Eastern Europe, they're pretty good at this stuff and amazingly cheap!

Mind you the language barrier can be a problem sometimes!

Keep it that way IT especially having to explain things to IT numpties for a living is a real pain in the brain.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #584
  • Sep 23, 2010 3:43pm Sep 23, 2010 3:43pm
Quoting pipmutt
Herculean, I would say impossible!

I suppose it's a remote possibility for a purely mechanical strategy but taking into account fundamentals, unexpected news and data, geopolitical events, central bank interventions, risk aversion.....and a host of other influencing factors, it's a non-starter as far as I can see. How can the unexpected be programmed in?
I'm not certain to what extent the prior discussion, which I was replying to (Chess vs Go, tactical vs strategic, etc), was referring to technicals, or fundamentals. Anyway, don't fundamentals amount to data that is being processed by the human trader? If that's so, I don't see why they can't potentially be coded. Coincidentally, I'm currently writing an EA for a client that, among other things, reads the FRB, RBA and RBNZ websites to gather interest rate data.

An 'unexpected' event presumably 'surprises' a human just as much as an EA. If there is a rationale behind the human response, again the thought processes can potentially be bottled. If the unexpected is a black swan, once-in-a-lifetime occurrence, then a (say) 1% loss - assuming some risk management - once every (say) 50 years is not going to have significant effect.

The way I see it, any rational process can ultimately be quantified, but intuitive ones can't. To a machine, 'intuition' is effectively a random (that terrible word again, LOL) process, i.e. any response is as good or as bad as any other. I would argue that the same applies with the human trader's process, i.e. if the 'guesswork' is in some way educated, then it likely relates to recognition of some prior pattern, or behavior, even if it's been processed subsconsciously; and, in any event, 'educated' implies some kind of underlying rationale.

Through a process of incremental refinement, an EA could simulate an increasing number of discretionary trader's nuances. The gap between the human's knowledge and the EA's would gradually narrow. That raises the question as to how much margin for error might be permitted, before any inaccuracies or omissions would render the human's strategy unprofitable.

Well, I did say herculean, LOL. And there are some very smart programmers out there. Given the liquidity that's available in forex, there might be some incentive to recruit an army of them.......

Perhaps one of the biggest problems might be the possibility of a kind of 'cat and mouse' game. If the heavyweights and 'manipulators' change their game plan, for whatever reason, the EA's pattern matching ability would likely be compromised. Again though, that is the same problem faced by a human trader. And I guess it's possible to argue that offsetting that, it takes two sides to seal a forex trade, and human psychology is irrepressibly predictable to some extent (e.g. the majority trapped traders always seem to exit immediately their trades return to breakeven), creating some kind of 'protective' immutability. Likewise with certain fundamentals that, everything else being equal, create predictable imbalances between supply and demand.

The possibility of a changing game plan raises the issue of adaptability. Neural nets are capable of adapting themselves to pattern changes. Hence the question might be: to what extent is the algo capable of adapting as intelligently as a human? The human must also presumably have some kind of (even if subconscious) statistical reason for varying his strategy, which brings us back to 'rationale'........

Of course much of this is pure speculation, and many of my points are nothing more than generalizations. But I've enjoyed pondering some of the possibilities.
  • #585
  • Sep 23, 2010 4:07pm Sep 23, 2010 4:07pm
Quoting Fudomyo
Here's an interesting question for you.

Is trading more like Chess or Go?


In 1995 IBM created Deep Blue, a Chess computer which competed against...
Fudomyo,

I love reading your post and you posses a level of intelligence not usually found on forex threads. Looking for the Holy Grail of EAs is as you say a waste of time as it will never be found. Who in their right mind would give it away.

Being a writer of EAs I will tell you that very good EAs do exist and they exist at a level far beyond your wildest dream. At the heart of the very best is your basic trend line which was used by traders before computers even existed. When used with the higher time frames I can assure you will never lose. The comments I've seen here on the use of the higher time frames are all false but debating with traders that have this mind set or lack the funds to trade the Daily is an effort in futility.

Thanks again for you posts,

Johnny
  • #586
  • Sep 23, 2010 4:43pm Sep 23, 2010 4:43pm
Quoting hanover
I'm not certain to what extent the prior discussion, which I was replying to (Chess vs Go, tactical vs strategic, etc), was referring to technicals, or fundamentals. Anyway, don't fundamentals amount to data that is being processed by the human trader? If that's so, I don't see why they can't potentially be coded. Coincidentally, I'm currently writing an EA for a client that, among other things, reads the FRB, RBA and RBNZ websites to gather interest rate data.

An 'unexpected' event presumably 'surprises' a human just as much as an EA....
I'm just waiting for some brilliant programmer to invent the next dominant "lifeform"

And I know some people say oh that would never happen... well some people's egos no know bounds... even if it means bringing about their own destruction to marvel at their creation for a small moment.

And then vampires will appear... etc... the wars will begin lmao so off topic I love it.

But yes I think there are already algos that are capable of interpreting sentiment. Wouldn't surprise me at all if there are some incredibly advanced ones.
  • #587
  • Sep 23, 2010 4:58pm Sep 23, 2010 4:58pm
Quoting Deusomega

But yes I think there are already algos that are capable of interpreting sentiment. Wouldn't surprise me at all if there are some incredibly advanced ones.
The banks and the like wouldn't employ very highly paid traders and low paid ones and everything inbetween ( the high paid 1 are carrots for the low paid 1 to attempt to achieve so work there asses off and very rare ) if EA existed.


But they do use some basic logic to get there 1000 lots into the market when required, but only as instructed with the rules setup by there masters.

True AI is likely 1000 years out, very very unlikely to be there in our life times.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #588
  • Sep 23, 2010 8:40pm Sep 23, 2010 8:40pm
Quoting hanover
I'm not certain to what extent the prior discussion, which I was replying to (Chess vs Go, tactical vs strategic, etc), was referring to technicals, or fundamentals. Anyway, don't fundamentals amount to data that is being processed by the human trader? If that's so, I don't see why they can't potentially be coded. Coincidentally, I'm currently writing an EA for a client that, among other things, reads the FRB, RBA and RBNZ websites to gather interest rate data.

An 'unexpected' event presumably 'surprises' a human just as much as an EA. If...
hanover,

I thought this way as well at one time and wrote an adaptable EA that weighed every tic of a chart and every variable I could think of only to find out it was totally inferior to a very simplistic approach.

The reason is the summation of all the variables is already on your charts and some that are known now won't immediately effect the trend in place for days until the market has time to absorb and mull over all the news. The fact that news is traded beyond it's real effect and back to a more reasonable price is one of the many reasons money can be made here.

The fact that one group of traders goes long and another group goes short at the same time will always make the true value of a currency pair a moving target. We can only find the mean value with tools like the Linear Regression Line. What ever the heavy weights do to manipulate the market such as quantitative easing by the Central Banks only puts more volatility into the market which adds to the trading opportunities.

Must get back to work. Happy trades.

Johnny
  • #589
  • Sep 23, 2010 11:02pm Sep 23, 2010 11:02pm
Quoting hanover

Anyway, don't fundamentals amount to data that is being processed by the human trader? If that's so, I don't see why they can't potentially be coded. Coincidentally, I'm currently writing an EA for a client that, among other things, reads the FRB, RBA and RBNZ websites to gather interest rate data.
Interest rate expectations and announcements are a great example of how context is more important than actual data, the market interprets and prices in, often in a quite obscure fashion. I'm not sure if/how that pricing-in could effectively be translated into code, there are just way too many variables. Imagine trying to write a program which has the ability to interpret, for example, Trichet's statements, replies, innuendo, and rhetoric at an ECB press conference or a Bernanke one-liner at a cocktail party, especially as it's often what is not said which has the most significance!

When it comes to fundamentals, the driving force behind market sentiment and ultimately price, most events tend to be unique, in a unique set of circumstances. To be in any way effective an automated strategy would need all those constantly changing variables and data to be interpreted and input manually which kind of defeats the purpose.

I'm not sure I'd call market behaviour intuition or educated guesswork, it's more of a rational response to available information, but information which is unique so prior market behaviour patterns in similar (but not identical) circumstances would have very little relevance.

I think the most an EA could do is to react to what the market is doing so it's never going to be as effective or productive as a human trader.

We're not extinct just yet, lol
  • #590
  • Sep 24, 2010 3:40am Sep 24, 2010 3:40am
dont know with you guys..

I learn this too:
for me ==> chart #2 more emotional than chart #1
Attached Image(s) (click to enlarge)
Click to Enlarge Name: chart1.png Size: 55 KB
Click to Enlarge Name: chart2.PNG Size: 52 KB
...
  • #591
  • Sep 24, 2010 4:23am Sep 24, 2010 4:23am
Quoting bandung
dont know with you guys..

I learn this too:
for me ==> chart #2 more emotional than chart #1
I don't even have candles, there pointless and only really useable to spot spikes in the market, I agree much less emotional.


Candles being BS and all!!
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #592
  • Sep 24, 2010 5:15am Sep 24, 2010 5:15am
I never consciously thought of own stuff in this perspective.

Another way is to bring the candles closer together, so that you only read the larger patterns they create.
You'd need candles that are contractable, though, such as here. A lower TF may also be necessary.
  • #593
  • Sep 24, 2010 8:55am Sep 24, 2010 8:55am
@JohnnyBSmart, pipmutt: thanks for your thoughtful replies, I see where you are coming from. I can see now that, even though rationale may be applied to interpret certain data, it would be too difficult for a machine to make sense of it (the Trichet speech is a fine example). Most of my analysis is technical, hence I tend to overlook fundamentals, quite probably to my detriment. Anyway, thanks for pointing this out.
  • #594
  • Sep 24, 2010 3:20pm Sep 24, 2010 3:20pm
Quoting Turveyd
I don't even have candles, there pointless and only really useable to spot spikes in the market, I agree much less emotional.


Candles being BS and all!!
So what are you using now?

I've also dropped candles in the custom back-tester that I'm writing. The reason for me was that not even one research paper written by a quant used candles. I was afraid that I won't be seeing as much information if I drop candles, but I've had the pleasant surprise of actually seeing more this way. For example, look at how easily you can identify swift moves ("thin" lines) versus chop periods ("thick" lines).

The TF toolbar is actually stuff left behind from when I used candles. It will be replaced with some kind of zooming control.

http://dl.dropbox.com/u/190212/ff/veloce_no_candles.png
  • #595
  • Sep 24, 2010 3:41pm Sep 24, 2010 3:41pm
Quoting Adal
So what are you using now?

I've also dropped candles in the custom back-tester that I'm writing. The reason for me was that not even one research paper written by a quant used candles. I was afraid that I won't be seeing as much information if I drop candles, but I've had the pleasant surprise of actually seeing more this way. For example, look at how easily you can identify swift moves ("thin" lines) versus chop periods ("thick" lines).

The TF toolbar is actually stuff left behind from when I used candles. It will be replaced with some kind...
Candles zoomed out so there just lines with no detail just like yours.

Candles are formed by a start and end time, if you move the start and end time you get different random candles there pointless.

A candle that reads as a big drop then another a big climb is exactly the same market move as a pin bar but the time clicked over while the market was at the bottom rather than after the move thats all.

The market does not wait for the 4H candle to close before doing something, and different charts can give different 4H starts and closes so different 4H candle patterns.

Ohhhh how I laugh when the idiots are waiting for the 4H to close to get a indication of the direction to come.

Your brain can only handle so much detail so keep the detail for what is needed and real.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #596
  • Edited 10:06pm Sep 24, 2010 8:33pm | Edited 10:06pm
Quoting Turveyd
The market does not wait for the 4H candle to close before doing something, and different charts can give different 4H starts and closes so different 4H candle patterns.

.
This is very true. But the one hour is a different animal.

When I first started reading here at FF, was about the time the "Bunnygirl Cross" was all the rage. Not that I ever used the system but one piece of advice that I took away from that thread was, ALWAYS BE CAUTIOUS OF TAKING A TRADE IN THE LAST 5 MINUTES OF THE HOUR.

For a long time I made it almost a routine, if I was at my puter, to watch those last few minutes of the hour and the first couple of minutes of the next hour.

If you ever study the market around this time you will see how the big boys play their little games. Those with the money available to manipulate price at will, didn't get that money by being dumb.

I'm not saying it happens all the time, just with regular monotony.

It's possible, with a movement of just 4 pips to create the impression that the market is moving in a particular direction, not for just one hour, but for Three hours.


Roy
Polishing my crystal ball
  • #597
  • Sep 24, 2010 8:54pm Sep 24, 2010 8:54pm
Quoting Turveyd
A candle that reads as a big drop then another a big climb is exactly the same market move as a pin bar
That is true, however it is not only important what happened but also how fast. While in your example the 2 candles would amount to the same end result, the speed at which it happened is different. If the price reaches a certain level, then retraces sharply within the same candle, it might be saying that a lot of traders reacted a certain way, and their reaction may be telling you something. Many times (for a variety of reasons) the selling/buying euphoria settles in and prices get pushed further then they should be. Quick pull offs from those levels result in pin bars. I am not trading them myself, but there is definitely some useful(interesting) info that can be gathered there. Provided they are being looked at with an open mind...
Stubbornly persistent
  • #598
  • Sep 24, 2010 11:37pm Sep 24, 2010 11:37pm
I've learned that I disagree with most that has been said in this thread.

Just a few counterpoints to posts I disagree with on the first few pages:

1. Determination is not enough.
2. Time and positive expectancy do not inevitably lead to success.
3. The market is always worth trading, it's just a matter of identifying the opportunity. At any moment in time, someone is in a good trade.
4. We can learn from demo, but for the vast majority of people, losing on it is a different beast to losing in real life.
5. Compounding is basic maths, not the 8th wonder of the world!
6. A good system does not equal a good trade, and breaking the rules does not necessarily lead to a bad trade. The trades should be judged on their own merit, not just by the system.
7. The size of wins and losses are equally important, as both are required for positive expectancy.
8. By definition, every action we take is a gamble.
9. Many may do okay setting a target profit, but on a per trade basis, skilled discretionary exits will always be superior.
10. TA is not a coinflip. Certain methods may be useless, but the chart represents the underlying action of the market, and is as useful as we can make it.
11. If the underlying action can be exploited, naturally, price patterns can be exploited.
12. There is always a reason behind a trend: price moves more in one direction than it retraces in the other.
13. Slow profits do not make for a fast fortune. Marathon runners don't sprint, but neither do they walk.
14. Stating that the exit is the single most significant part of trading is putting the cart before the horse. We can't take any money off the table without first making the money, and this is achieved through entry.
  • #599
  • Sep 25, 2010 3:12am Sep 25, 2010 3:12am
Quoting 30XTCi
This is very true. But the one hour is a different animal.

When I first started reading here at FF, was about the time the "Bunnygirl Cross" was all the rage. Not that I ever used the system but one piece of advice that I took away from that thread was, ALWAYS BE CAUTIOUS OF TAKING A TRADE IN THE LAST 5 MINUTES OF THE HOUR.

For a long time I made it almost a routine, if I was at my puter, to watch those last few minutes of the hour and the first couple of minutes of the next hour.

If you ever study the market around this time you will see...
2 Trades on 10s and 8 months off.

Yes on the 8months off, to much work to trade.

2 trades ?? 36 currently with a 74% win rate and account +60% area.

And I don't see the last 5minutes being different to any other 5mins the big boys don't care about making pretty candles.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #600
  • Sep 25, 2010 3:15am Sep 25, 2010 3:15am
Quoting sergiu
That is true, however it is not only important what happened but also how fast. While in your example the 2 candles would amount to the same end result, the speed at which it happened is different. If the price reaches a certain level, then retraces sharply within the same candle, it might be saying that a lot of traders reacted a certain way, and their reaction may be telling you something. Many times (for a variety of reasons) the selling/buying euphoria settles in and prices get pushed further then they should be. Quick pull offs from those levels...

Nothing to do with speed, it's just when the clock ticks over and starts another candle.

and no Randomly formed and worthless, look at the people that trade 4H, they have a almost perfect 50% win / lose rate, +200 yeaaaahhh, -200 dohhh!!! and have had for years and still keep plugging away at the same candle readings.

You'd be equally as well off reading tea leaves, which is also ridiculous.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
0 traders viewing now