Why So Many Indicators but Little Strategies in TL?

[Guest Tresor]

I'm using a few moving averages with price action. The problem is when the market becomes choppy then nothing works.

Any ideas on this?

 

Hi goinglong,

 

I have one strategy that uses ChoppyMarketIndex to determine if the market is in chop. It is called Thermostat Program. I can send it to you or you can find it in a book 'Building Winning Strategies in Tradestation''.

 

My guess would be that you can make the strategy that should use long length moving averages when market is trending and shorter moving averages when the market is in chop. And the ChoppyMarketIndex would tell you wheather the market is tranding or in chop.

 

From theoretical readings that I made also ADX can tell you if maket is in chop .

 

Regards

[Guest Tresor]

Hi Moderators,

 

Could this thread be moved to ''Automated Trading'' section of the forum? There seems to be no thread there so it might be a beginning of section for traders to discuss automated trading strategies and systems?

 

Regards

[MC 10]

The Longer You Survive In This Game The More You Will Come To The Realization That The Indicator Search Is A Futile One . The Day I Removed All The Occilating, % Rate ,bla Bla Bla , Was The Day When The True Nature Of The Market Became Apparent. Different Instruments Have Different Temperments And Will React To Information Different From One Another. Screen Time Is The Best Teacher You'll Get, And Most Have No Idea What They Are Supposed To Be Looking For In The First Place. Mp And Vsa Are Both Places One Can Start To Gain Greater Understanding Of Market Structure And The Relationship Of Auction Theory And Volume( Imo ) The Key To The Market.tape Reading At Key Levels With Time And Sales Has Been Replaced With Delta Footprint For A More Informed If Not Over Informed View Of Order Flow. One Can Gain All Information Needed To Be Sucesfull In The Market From These Few Items . The Rest Is Up To You You You. No Indicator, System, Or Other Trading Method Can Ever Replace The Time It Takes To Learn How The Intrument You Want To Trade Works. After That Ther Is The Problem Of How Much Size To Take On And Stops , And How To Manage Trades,wich Is Just As Important If Not More Than Your Entry. In No Other Buisness Can You Have A Plan , Execute It To Perfection Only To Have Your Account Drained Before Your Very Eyes. Thats Where The Time Comes In . So If Your Goal Ie To Create A System You Better At Least Take The Time To Learn The Markets The System Is To Trade , And If Thats The Case Why Not Just Trade Discressionary Because As Far As I Know Most Of The Money In This Game Is Still Made By People Not Systems Or Machines, But I Could Be Wrong.

 

OUCH MY HEAD!

Good info but very hard to read bud. The caps on every first letter and no spacing makes that very cluttered to read. I think many would benefit from what you said but wont take the time to read that as it is.

 

MC

[MC 10]

Hi Moderators,

 

Could this thread be moved to ''Automated Trading'' section of the forum? There seems to be no thread there so it might be a beginning of section for traders to discuss automated trading strategies and systems?

 

Regards

 

Done.

Good thinking.

[Guest Tresor]

OUCH MY HEAD!

Good info but very hard to read bud. The caps on every first letter and no spacing makes that very cluttered to read. I think many would benefit from what you said but wont take the time to read that as it is.

 

MC

 

Yeah, I read the first lines of the text and then gave up.

[Guest Tresor]

Done.

Good thinking.

 

Thanks MC

[ammo 10]

the reason there are no strategies is because tl is so simple, sell at res and buy at support. Now if you use monthly,then daily, then 4hr,then 2 hr ,then 30 min ,then 15 min,then 5min, and 1 min and chart all those lines you will channel the intrady chop,its still sell res and buy support

[Guest Tresor]

the reason there are no strategies is because tl is so simple, sell at res and buy at support. Now if you use monthly,then daily, then 4hr,then 2 hr ,then 30 min ,then 15 min,then 5min, and 1 min and chart all those lines you will channel the intrady chop,its still sell res and buy support

 

Hi ammo,

 

Of the whole discussion I am starting to build a picture that answers the question why so little strategies here.

 

The picture is as follows:

 

1. There are for sure strategies that are very profitable and can be automated. No doubt about this.

 

2. The very profitable automated strategies are complex and required vast intelectual and time consuming resources to be created. MC bets it takes 100 - 1000 hours just to code them, not to mention the whole period of devising the strat, its backetsting and forward testing.

 

3. People who created such profitable strategies will not share then for obvious reasons.

 

4. There is a number of people / companies that have mastered technical analysis and have coding abilities and sell strategies for hard money to newbies. I have not tested these paid strategies. Smart guys at Traders Laboratory say they are crap. I trust them and I will not buy any of such strategy. I would test it if someone makes replicas of them but I would't pay for them (since they are said to be a crap).

 

5. Smart guys at Traders Laboratory who created unique profitable strategies will not share them for obvious reasons. I wouldn't share them, too. And I will not be asking for revealing codes and logic of these strategies. I will merely post some ideas later that I have in my mind and maybe they can comment on them.

 

5. Smart guys at Traders Laboratory believe paid strategies are a waste of money. If so, that answers my question why so little strategies. If they believe paid strategies are crap, then why they should post replicas of these systems here? I believe this answers the question in the title of the thread.

 

The above is the picture I have now after reading a few comments by traders with high IQ.

 

Based on this picture I would like to know:

 

1. How many lines of code a unique profitable reversal strategy might have. 1000 thousand / 10000 thousand? I already know how much time it takes to code (100 - 1000 hrs)

 

2. How many conditions like ''begin if / then else'' conditions are there, according to your knowledge, in such a strategy. Please tell a rough number.

 

3. How many indicators are incorporated in such a fully automated strategy. Again a rough number please.

 

Regards

[Hlm 10]

I'm going to jump in here and share some thoughts/opinions. It's been awhile since I have had a good rambling on here.

 

1) It's not the indicators; it's the people who use them. For many it's like giving someone that has never built anything all the raw materials and tools to build a house. An individual not only needs a solid understanding of the final product, but how to work the tools and the correct steps to be taken as well.

 

2) Those that "only" use price action and volume are still using indicators. During thousands of hours of screen time the brain is learning and doing basic calculations (moving averages, stochastics, etc.) subconsciously. Now don't get me wrong. There are great benefits in doing this and I strongly believe everyone should go through this phase before using indicators. This process keeps your learning fluid and dynamic and in turn increases your learning speed by brute force scenarios.

 

3) One of the easiest ways to use an indicator it to filter out the times you shouldn't trade. This instantly increases your probability of trading during times when you can be successful. For example, someone mentioned moving averages. To show people what I am talking about I have them throw up a few simple moving averages (usually fib numbers). I then have them look back over the chart and tell me what they notice. The majority come back with the answer that when price nears the larger moving averages while they are flattening and coming together (many times when price is between them) price gets choppy and fades off the smaller ones. After some discussion they realize the concept that it's because they larger time frames are in their decision stage and positioning is taking place. The end result is that by just using a few simple averages to get a basic idea of several different time frames they can now estimate the potential market condition around that time. In other words depending on their strategy they have filtered out times when they should not trade. It is very important for one to know specifically what an indicator is telling them. A simple 5 3 3 stochastic is also a good example. If an individual does not pay attention to how high or low it is, but solely on the direction it is moving and how fast when compared to price movement, it gives a great internal like reading especially in instruments like Forex. You need to know how to use that power tool or else you will get hurt. Wow, now that's some rambling.

 

4) In the majority of cases one must first be a successful discretionary trader before they can move to automation. At this point the object is not to be fully automatic (that will come in time) but to simplify/automate certain aspects of your trading. You will first need a blueprint. List every single process you go through when taking a trade. Make sure to be as detailed as possible. Then order these by how much time and thought goes into answering this step (level of discretion). At this point you start from the easiest decisions and look for indicators (remember, you must understand its limitations) that answer that specific question. I believe this to be the correct way to use indicators in systems. An indicator is used to answer a specific question you have. It must have a specific purpose. As you work further and further down your list you will learn a lot about your trading style and yourself...trust me.

 

5) As stated many places, no one is going to sell a fully automatic system that works. Also, most systems you find that no longer work won't help you as well. These systems are no different then you curve fitting something over the last couple of months and it working for the next couple of weeks. Remember, there were always be some individuals that just get lucky. Also, in my opinion the only thing one should ever pay for is education. Let me be more clear...education that isn't based off of some specific system/indicator.

 

6) This may get a little controversial but I truly believe a market is a market. I disagree with individuals that say you must pick one instrument and become an expert. Of course when first learning how a market works it may be a good idea to stick to one (keeping things simple). I am not saying that an individual can't learn certain characteristics of an instrument to find an edge/arbitrage situation. However, this is where the risk comes in for that edge to disappear. By all means if you found such an edge max out your leverage and try to squeeze every single cent from it before it vanishes. The word I used above was "must". If you want to be in the game for the long term and be able to effortlessly adapt, you should also learn that concepts that every market has and how to profit from the bigger picture.

 

7) If you plan to go down this automation route strap yourself in. It's going to be a long and wild ride. Also, if you are not both a successful discretionary trader and a talented programmer it's going to get even more painful. It may be easy to have someone else program the first few items on your list, but as you get further down it gets more complicated. If you want quality work, this back and forth discussion and research translates into a crap load of $$$ being spent on a programmer. Remember, the concepts and theory may be simple, but it takes extreme creativity to keep it that way when translating it into code.

 

8) BEWARE OF VISUAL BACK TESTING. Our minds are amazing machines. However, they can be tricked very easily. In my opinion visual back testing is more dangerous than statistical back testing via automation.

 

Well...there were a couple of other things I wanted to discuss but the current nagging from my four year old has made me completely forget. :hmmmm:

 

1. How many lines of code a unique profitable reversal strategy might have. 1000 thousand / 10000 thousand? I already know how much time it takes to code (100 - 1000 hrs)

 

2. How many conditions like ''begin if / then else'' conditions are there, according to your knowledge, in such a strategy. Please tell a rough number.

 

3. How many indicators are incorporated in such a fully automated strategy. Again a rough number please.

1) This completely depends on your coding ability. A creative and talented coder could knock out in one line what another coder does in ten lines. It may be different depending on formatting as well. In a very general context unless you are a superbly talented programmer (remember, simple strategy doesn't necessarily translate into simple coding) I would guess a few hundred lines minimum with a high probability that it will be much much more. The same uncertainly goes with the time. Most likely only half your time will be coding. The other half will be in research and development with trying to find the most efficient ways to program the code. Oh...and unless you are a very talented programmer don't forget the time for debugging. :bang head:

 

2) Again, it depends on how you program it. If one takes a branching tree concept there could be dozens. However, if an individual builds a system that gathers information and derives an indicator off the statistical analysis it could just be one simple if/then statement.

 

3) I hate to sound like a broken record, but again it depends on how one programs it. Does the individual go out and look for the indicators to fill in the gaps? Or does the individual just use the basic concept of several indicators and merges them together. For example, you could make a rainbow by using 20 moving averages and then base your analysis off of the relation to one another. Or you could just create one indicator that internally does the statistical calculation across 1000 moving averages and displays it with one or two simple lines.

 

Much of this depends on how much discretion you use with your current successful strategy.

 

In my personal experience one usually starts by plugging in the holes with simple indicators. However, the further you get down that list, the greater the need is to extract the specific concept you need from the indicator and to create a more personalized/fluent indicator for analysis. I see this as the natural progression.

Edited June 7, 2008 by Hlm
Spelling. No content was added or removed.

[Guest Tresor]

HIm,

 

Many thanks for the post. I intentionally agree with everything you wrote.

 

Regards

[Guest Tresor]

I meant ''intuitively'', not ''intentionally. Sorry.

[MidKnight 12]

I think to believe it must be complex is a serious flaw in your working hypothesis. They can be complex but they needn't be. What they need is a strongly grounded fundamental principle to be built upon *hint (moving avg crossover is not a strongly grounded fundamental principle). Look at markbrown.com who has published many simple strategies back in 1997 on the omega list. Note how they explore a principal. The scope is limitless...that is what takes the time and effort.

 

If you do go down this route....it will probably take you years of exploration and is definitely a journey rather than a destination. Just like discretionary trading is.

 

With kind regards,

MK

[Kiwi 16]

A very interesting post Hlm. It's rare that I read one that long right through.

 

A couple of points I would make:

 

1 - You can often code a good idea in several hundred lines as long as its in a language made for trading. The thing that takes longest is that you think something is clearly specified until you try to code it ... and it never is.

 

2 - I don't really agree that the brain learns to do the indicators but agree you could make it do it. I think that the better indicator developers created them to model something they were thinking/seeing although the worse ones just modified earlier ideas.

 

The least sexy indicator of all in its reality is the stochastic although it does get a big following - when you examine it, all its doing is applying a horizontal HiLo channel thats a long as the primary number (6 of 6,3,3 say) and looking at how close price gets the top or bottom. Overbought is in the top 10% of the last X days. Basically a really nice way of asking the OB/OS question. If you understand that then you know that to tune it you have to look at price and say "how many bars does it have to keep going in direction x before its OS/OB."

 

The other two numbers are just smoothing of that and then smoothing of the smoothing.

 

That gives me a good idea for a new indicator. Its the KiwiStoch and it has 4 parameters. The first is the channel size for up moves. The second is the channel size for down moves. Why? Because for many markets (stock and index futures, yes, forex probably no) the market falls much faster than it rises. So the channel length should be different in up trends and down trends (do we need a 5th number for the length of a lsma or fast ma (hull or t3 say) to decide if the current "trend" is up or down)?

 

3. Some good systems will run differently up or down (or often are long only) and what System developers are doing is including that last issue in their system. One reason they use indicators in systems is simply that its easier than writing the whole price behaviour description yourself (programing short cut).

[Hlm 10]

If you do go down this route....it will probably take you years of exploration and is definitely a journey rather than a destination. Just like discretionary trading is.

Yep...strap yourself in and enjoy the ride.

 

1 - You can often code a good idea in several hundred lines as long as its in a language made for trading. The thing that takes longest is that you think something is clearly specified until you try to code it ... and it never is.

Exactly. What one believes to be a very simplistic discretionary strategy can quickly turn into a programming nightmare.

 

2 - I don't really agree that the brain learns to do the indicators but agree you could make it do it. I think that the better indicator developers created them to model something they were thinking/seeing although the worse ones just modified earlier ideas.

Well, the brain may or may not actually do the calculations exactly like a computer does (again, you are dealing with the subconscious so it makes it hard to prove) but it recognizes patterns that in reality give you the same end result. The main concept I was trying to portray was that even if you remove all indicators you are still working with lagging information unless you trade with absolutely no context.

 

The least sexy indicator of all in its reality is the stochastic although it does get a big following - when you examine it, all its doing is applying a horizontal HiLo channel thats a long as the primary number (6 of 6,3,3 say) and looking at how close price gets the top or bottom. Overbought is in the top 10% of the last X days. Basically a really nice way of asking the OB/OS question. If you understand that then you know that to tune it you have to look at price and say "how many bars does it have to keep going in direction x before its OS/OB....."

I find the importance of oversold or overbought from a stochastic to be of little use. The only aspect I care about is the direction and ease of movement in relation to price. Also, I believe a stochastic is best used as a filter instead of a trigger.

[ammo 10]

you will in the end have a robot managing your money,to me thats a scary thought,just learn to trade,as boring as it can be ,i would think programming is worse

[Guest Tresor]

7) If you plan to go down this automation route strap yourself in. It's going to be a long and wild ride. Also, if you are not both a successful discretionary trader and a talented programmer it's going to get even more painful. It may be easy to have someone else program the first few items on your list, but as you get further down it gets more complicated. If you want quality work, this back and forth discussion and research translates into a crap load of $$$ being spent on a programmer. Remember, the concepts and theory may be simple, but it takes extreme creativity to keep it that way when translating it into code.

 

Hi, I have a feeling that what you wrote in your post derives from your own experience and for sure is important. You walked the way that I am going to start to walk soon.

 

Let me give you some background of my trading and then ask you a few questions. I have been trading since August last year. I used to be an accountant, so 2 month before starting to trade a made an excel spreadsheet with financial data of 90% of all companies traded on my local exchage - I got eyes inflammation of constant working with 12.1 inch monitor 1.5 month day and night work) - see the sample spreadsheet.

 

Based on this spreadsheet and using some fundamental models of valuing companies I came to the conclution that my stock exchange should have already collapsed (the companies here were valued 3 - 4 times more higher than e.g. Google). So I shorted. I shorted 2 months too early. The markets collapsed 2 month later. I lost almost all my equity. So I threw away all my fundamental analysis knowledge that I had so far and started learning technical analisys. Till now I have quadrippled my starting equity that I had in July. I could say I am not a very good trader but I am becoming a good one. Just for explanation, I am using some exotic techniques that you guys do not use, E.g. I trade divergences between price and open interest, etc. US exchanges do not inform of open interest in ticks in real time so you simply can't know these techniques. I am not a programmer.

 

My advice to newbies is THROW AWAY ALL YOUR FUNDAMENTAL ANALYSIS.

 

That was the introduction. Now, let's continue the question section:

 

1. ''It may be easy to have someone else program the first few items on your list'' - for sure it will be easy and money saving. What should be the complexity of these first codings that I can expect to be coded for free? I am sure that if I ask someone from this community to code signals based on MA crossovers it will be okay for this person to code it for me for free but if I ask someone to code divergence between the price and an indicator or to code a divergence between indicator and indicator this might be controvertial whether such job should be coded for free. My question is: what can I ask for and what I cannot ask and should rather look for a paid-for help?

 

2. What are normal rates of programmers? Are these hourly based or I pay after the job is done.

 

3. Where can I look for a good programmer? If you can recommend someone you can send me a private message.

 

4. What are most common mistakes that I can do while co-operating with a programmer. What descriptive know-how do I have to possess in order be able to properly explain the programmer what I want from him. If you made any concrete mistakes at the beginning of your path please tell them.

 

Regards

Edited June 8, 2008 by Tresor

[Hlm 10]

A little bit of background from me to put my answers in context:

I am not a "professional" programmer, but I have been doing it across several languages for around 15 years (much longer than I have been in the markets). I also have a few very close friends that also enjoy programming as a serious hobby. The result is that I have not had to go out and look for a programmer. Only just recently did I interview a couple of local college students that were priced at the lower end of the range I give below. Due to certain reasons I have decided to hold off on outside help for the time being. In other words, my experience for hiring external help is very limited.

 

1. ''It may be easy to have someone else program the first few items on your list'' - for sure it will be easy and money saving. What should be the complexity of these first codings that I can expect to be coded for free? I am sure that if I ask someone from this community to code signals based on MA crossovers it will be okay for this person to code it for me for free but if I ask someone to code divergence between the price and an indicator or to code a divergence between indicator and indicator this might be controvertial whether such job should be coded for free. My question is: what can I ask for and what I cannot ask and should rather look for a paid-for help?

It would depend on the amount of information you wish to extract from the divergence. For example, to program an indicator that detects divergence based off a specific threshold is much easier than doing a full blown statistical analysis of the information. With a quick search online you can most likely find some code samples to detect simple divergences. The easiest way to find out if someone is willing to program the "easier" parts for free is to describe in detail what you are looking for. Many times there are programmers out there that have code laying around that might fit what you are looking for with very little tweaking. However, even though we do have programmers on this forum, there are other forums that have a larger community for programming and may be of more help.

 

2. What are normal rates of programmers? Are these hourly based or I pay after the job is done.

There is a large variety of answers to this question. Depending on the programming language, skill, native tongue, etc, you are talking anywhere from $15/hr to $300+/hr. I have also heard of people doing it per job but don't know of specific examples. That is about as for as my knowledge goes in this subject.

 

3. Where can I look for a good programmer? If you can recommend someone you can send me a private message.

I do not know of any specific places off the top of my head. The first question would be what you are wanting it to be programmed in. This would include the language and software. After this is answered others may be able to help you out more effectively.

 

4. What are most common mistakes that I can do while co-operating with a programmer. What descriptive know-how do I have to possess in order be able to properly explain the programmer what I want from him. If you made any concrete mistakes at the beginning of your path please tell them.

Make sure you have a well defined and detailed blueprint. Really take the time to sit down and cover every aspect of how you trade. Once you have all the information laid out, go back through and simplify as much as you can. Run through the list several times while reordering and combining potential areas. The better this process is done, the less headaches you will have in the future.

[Guest Tresor]

 

It would depend on the amount of information you wish to extract from the divergence. For example, to program an indicator that detects divergence based off a specific threshold is much easier than doing a full blown statistical analysis of the information. With a quick search online you can most likely find some code samples to detect simple divergences. The easiest way to find out if someone is willing to program the "easier" parts for free is to describe in detail what you are looking for. Many times there are programmers out there that have code laying around that might fit what you are looking for with very little tweaking. However, even though we do have programmers on this forum, there are other forums that have a larger community for programming and may be of more help.

 

OK,

 

Let's make some excersies that might help me and other newbies to properly write what they have in their minds. I enclose a printscreen of the futures that I trade (16 minute resolution + a TRIX subchart below. I have spoted some regular bearish and bullish divergences there. In my automated strategy I would like to use divergence as one of 5 important conditions for reversals.

 

I use MultiCharts. MC has words ''BullishDivergence'' and ''BearishDivergence'' built-in. And I would like to ask a programmer to code sell and buy signals based on divergence detected between the price and TRIX. Bullish divergence should be looked for below TRIX Zero line and Bearish divergence should be looked for above TRIX Zero line. Could a programmer from this forum tell if this is enough for him / her to do the job or they would need some additional data in this specification?

 

Regards

Edited June 9, 2008 by Tresor

[Hlm 10]

I have spoted some regular bearish and bullish divergences there. In my automated strategy I would like to use divergence as one of 5 important conditions for reversals.

 

Could a programmer from this forum tell if this is enough for him / her to do the job or they would need some additional data in this specification?

If you have set rules to recognize a divergence from a hard right edge then it should not be hard to program. In other words, without hindsight, at which bar do you consider it to be confirmed or of high probability? I personally do not use divergences but I have seen simple indicators that use a zig-zag concept as a simplistic approach to recognizing them. There might even be divergence code already out there for EasyLanguage that just needs to be tweaked to work with TRIX (Silly rabbit, Trix are for kids!). As for your question about there being enough information, etc...I would suggest starting another thread in the coding section.

 

----A Few Notes----

 

For a discussion about programming a specific indicator you should start a thread here.

 

A new thread should also be started for other discussions about automation that don't specifically pertain to the current thread title.

 

This will help keep information organized and clean.

[Guest Tresor]

 

A new thread should also be started for other discussions about automation that don't specifically pertain to the current thread title.

 

Hi Hlm,

 

My bad. The question in the thread title was already answered

 

In a few days I will start a new thread on how a specification for a coder should look like.

 

Thanks everybody for their input to this thread which actually ended.

 

Regards

[zdo 20]

4. What are most common mistakes that I can do while co-operating with a programmer.

 

“Frustrating” = hooking up with a programmer who has not traded and does not understand trading.

 

“More Frustrating” = hooking up with a programmer who has traded and does not understand trading.

 

“Even More Frustrating” = hooking up with a programmer who has traded and does share nor understand your trading style. ie will try to translate / transfer your method into / onto his or her trading constructs. This one is subtle at first… but grows increasingly blatant and ‘frustrating’

 

“Very Frustrating” = hooking up with a programmer who has no passion for the vision and task at hand...

 

"Possibly just as 'frustrating'" realizing you paid beaucoup dollars to have something unproven coded that you now no longer really care about... grail seeking with an entourage

 

“Most Frustrating” = ... you fill in the blanks

 

4. What descriptive know-how do I have to possess in order be able to properly explain the programmer what I want from him.

 

You should be able to write the context and the COMPLETE trading logic in pseudo code. You should leave zero areas even remotely open to ‘assumptions’. You should be able to thoroughly clarify any areas the programmer asks about - first verbally then by intricate amplification of the pseudo code for any and all sections in question – and, btw, it’s a huge sign of impending trouble if the programmer programs instead of asking many many clarifying questions…

 

Just saw your posts on kreslik

There are several rarely discussed but absolutely crucial points made about programming / automation in this link http://kreslik.com/forums/viewtopic.php?t=1086 … ignore any of them at your own peril…

Edited June 9, 2008 by zdo
cn't spel

[Guest Tresor]

You should be able to write the context and the COMPLETE trading logic in pseudo code. You should leave zero areas even remotely open to ‘assumptions’. You should be able to thoroughly clarify any areas the programmer asks about - first verbally then by intricate amplification of the pseudo code for any and all sections in question – and, btw, it’s a huge sign of impending trouble if the programmer programs instead of asking many many clarifying questions…

 

 

Hello zdo

 

I have already structured in my mind the next thread. I will post it during the weekend in the Beginners Forum. It will mostly focus on pre-pseudo code stage and on pseudo code stage

 

the stages are :

1. ideas in newbie's mind - newbie's part of the job

2. pre-pseudo code stage: writing the ideas on paper (very difficult: I am trying to define at the moment what should trigger divergence detection in my strategy and already arrived at 4 contradicting conditions :crap: ) - I will ask in the thread for recommendations on books on logic for non-mathematicians - newbie's part of the job

3. writing a pseudocode - newbie's part of the job? maybe yes, maybe no - I will ask for help on answering this question

4. writing the code - programmers job

5. debugging

 

Regards

[Sledge 13]

I am happy with this discussion for two reasons:

(i) my trader IQ rose from 5% to as much as 10%

(ii) you partially confirmed my assumption that a profitable system must be complex (100 / 1000 / more hours of coding).

 

Could you please elaborate more on a complexity of such an edge system? 1000 / 10000 lines of code? 10 / 100 conditions? 20 / 50 indicators?

 

Foregive me my being so inquisitive. Just need to asses whether it will take 5 years of my life to devise and code such a system or 15 years.

 

Regards

 

Toss the Indicator coding nightmare- this is said in all sincerity- take your thousands of hours to automate a system- and use that time to read the bars, learn to read a chart- the chart will tell you exactly what you need know in order to succeed. Trying to "beat the market" or "outsmart it" will lead to frustration and loss.

[zdo 20]

re "what should trigger divergence detection in my strategy and already arrived at 4 contradicting conditions"

Indicators work!

Both ways!

 

... "use that time to read the bars, learn to read a chart"

[Hlm 10]

Toss the Indicator coding nightmare- this is said in all sincerity- take your thousands of hours to automate a system- and use that time to read the bars, learn to read a chart- the chart will tell you exactly what you need know in order to succeed. Trying to "beat the market" or "outsmart it" will lead to frustration and loss.

In my opinion one should only program to automate parts (if not all) of an already successful discretionary system. It is very dangerous for an individual to try and program a new system by plugging indicators together which creates a curve fitting nightmare. My guess is that this is what Sledge was referring to. Many times when individuals disagree with automation they don't specify which route they are talking about...and there is a huge difference. I disagree against the idea that there is no benefit in automating aspects of an already successful concept. If you use anything other than the raw tick data (1 tick chart) than you have in fact already simplified one aspect via automation. Many traders get the wrong idea when the word "automation" is used. However, this is not surprising since the large majority don't make it anyways. Again, trying to build a house without first knowing how everything should fit together, how to use the tools, and what the end result should look like is a long dark road that will lead to endless headaches.

 

Added in edit: Automate to simplify, not to find success.

Edited June 9, 2008 by Hlm
Added last little note.