Rel's Lab | Page 3 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Rel's Philosophy of Trading Discussion 24 replies

Rel's Forge 53 replies

Rel's Evaluation 13 replies

Nikola's Trading Lab 247 replies

Spud's Lab Accident 22 replies

Attachments: Rel's Lab
Exit Attachments

Rel's Lab

  • #41
  • Aug 15, 2014 11:19am Aug 15, 2014 11:19am
Quoting Relativity
{quote} You are right. But thats their job to figure it out plus also see if this style fits them. I can only give 1 example and that is the one related to iDoubleStoch's work. It is exhausting typing all these out, I got to admit, because it describes my thought process in detail.
I got to admit what you have done so far is simply amazing, awesome and very much appreciated. Thanks for sharing. Truth be told, I'm already using a form of this concept using the more traditional indicators and also with good results(not considering my personal psych,emotional and impatient issues of course!), the major difference here is, you being a statistician, have been able to back up the probabilities with hard figures. And as they say 'numbers don't lie'. Well done my friend!
  • #42
  • Aug 15, 2014 11:26am Aug 15, 2014 11:26am
Quoting Relativity
{quote} You are right. But thats their job to figure it out plus also see if this style fits them.
+1 The goal isn't to bottle feed people with magical orb indicators but to share ideas. From these ideas one can build his own trading method.
I showed that the range bar charts are nicely Normally distributed and were therefore the frame of choice to run a kalman filter. Guess what the first question I got was: "how do you trade this?".
No greed. No fear. Just maths.
  • #43
  • Aug 15, 2014 11:44am Aug 15, 2014 11:44am
Quoting PipMeUp
{quote} +1 The goal isn't to bottle feed people with magical orb indicators but to share ideas. From these ideas one can build his own trading method. I showed that the range bar charts are nicely Normally distributed and were therefore the frame of choice to run a kalman filter. Guess what the first question I got was: "how do you trade this?".
  • #44
  • Aug 15, 2014 11:56am Aug 15, 2014 11:56am
Quoting LITEchild
{quote} I got to admit what you have done so far is simply amazing, awesome and very much appreciated. Thanks for sharing. Truth be told, I'm already using a form of this concept using the more traditional indicators and also with good results(not considering my personal psych,emotional and impatient issues of course!), the major difference here is, you being a statistician, have been able to back up the probabilities with hard figures. And as they say 'numbers don't lie'. Well done my friend!
Yes thank you. My statistics aren't so complicated thou. Its mainly Frequency Distribution and Outliers.
  • #45
  • Aug 15, 2014 12:03pm Aug 15, 2014 12:03pm
Quoting PipMeUp
{quote} +1 The goal isn't to bottle feed people with magical orb indicators but to share ideas. From these ideas one can build his own trading method. I showed that the range bar charts are nicely Normally distributed and were therefore the frame of choice to run a kalman filter. Guess what the first question I got was: "how do you trade this?".
To be fair to them, its a valid question given what (little or much, more probably little) they know. We can only ask based off is a combination of what we know plus admit how much that we don't know. Thats why for your case I didn't have to worry when I saw your post =) Because I knew what you are talking is substantial and dared to go into the details straight on. I am no mathematician, but I am basically aware of the terms you use due to being exposed to statistics.
  • #46
  • Aug 15, 2014 2:43pm Aug 15, 2014 2:43pm
You said "1-The bars are created using a "significant" value, to determine the best range bar value to use."
How do you determine this value? I also have my own version of the zigzag. Here are the number of swings depending on the ranges and on the pairs:
Inserted Code
       AUD/USD   GBP/JPY  EUR/GBP
RG40    2145      6818       711
RG20    8082     26753      2717
RG10   30935     97855     11477
Obviously the volatility is in the mix. I would say the more swings the better; both for the quantity of samples to gather stats and for a higher frequency of setups. So the smaller range the better but the spread limits how low one can go for the setups to be profitable. I thought of a multiple of the spread but I find it too much artificial.
No greed. No fear. Just maths.
  • #47
  • Aug 16, 2014 5:21am Aug 16, 2014 5:21am
Quoting PipMeUp
You said "1-The bars are created using a "significant" value, to determine the best range bar value to use." How do you determine this value?

http://www.forexfactory.com/showthre...58#post7675258
Quoting Relativity
As for what range number to pick, I chose not to go into the details as it can confuse people. But basically it adjusts every 5 days on a rolling basis.
However I would explain why I chose 5 days (which is a trading week = W1);

I had some exchanges with this guy called DeltaSkelta, back in the BabyPip's days. He took my Pip/Time Efficiency idea further and expounded on it very well. Basically we determined 2 things;
1-H1 is the most efficient timeframe
2-W1 is the best timeframe to look for context
http://www.traderbrains.com/average-...of-timeframes/

I really like his work, because his calculations are obviously more in depth than mine, so I would quote him here.


Quoting PipMeUp
I also have my own version of the zigzag. Here are the number of swings depending on the ranges and on the pairs: AUD/USD GBP/JPY EUR/GBP RG40 2145 6818 711 RG20 8082 26753 2717 RG10 30935 97855 11477 Obviously the volatility is in the mix. I would say the more swings the better; both for the quantity of samples to gather stats and for a higher frequency of setups. So the smaller range the better but the spread limits how low one can go for the setups to be profitable.
I would say that I am not worried about number of swings. I am way more worried about range bar sizes, as it would determine the quality of the swings. I had this problem when I was using time-based candles e.g. I would get Trend State pattern in Asian session hours, which is bad, as the pattern would be extremely small and reflects poorly on the overall sample results.

This is also a 'push' factor that led me to use Range Bars. Then I figured that if I am using fixed value for the Range Bars, I would not be accurately reflecting the true state of current market volatility when the bar is printed. I recalled seeing theres some method out there (can't remember source) of Renko bricks using a non-fixed value that is based off ATR, which eventually led to the 5 day rolling idea for my range bars.


Quoting PipMeUp
I thought of a multiple of the spread but I find it too much artificial.
I actually agree with you, because I think there is no 'true' spread from what I understand. It differs at which level of liquiity you are trading on. e.g. spread of retail VS MM and spread of bank VS bank is obviously different. I understood the spread as some kind of a 'mark up' of price whenever buying and selling is taking place. Hey, someone's got to eat, right?
  • #48
  • Aug 16, 2014 6:22am Aug 16, 2014 6:22am
Thanks for the answer about the size of the range. It looks like 10 pips is a good rule of thumb for the majors.

I was a bit curious and decided to do some data mining with the swings. I extended your definition to include the case where two swings in sequence are almost the same length. I consider them of equal size if their heights don't differ by more than one range. I motivate this choice by saying that in a constant range bar chart one range is also the measure of volatility. I also distinguished the sub-cases in the state #4 in order to be thorough.

I then built the table of the frequencies of the states given the previous state. Fascinating! This 11x11 table tells you that if you buy the support of a range the statistically best exit method is to wait for the price to reach the resistance (i.e. not taking profit at the middle equilibrum price) then trail the SL one range size below the price. If price breaks out take profit when the end of the swing is detected (trail you SL). This is all in the frequencies table but not easy to read. Will the broken resistance turn support? This is also evidenced in the table.
No greed. No fear. Just maths.
  • #49
  • Edited 8:04am Aug 16, 2014 7:32am | Edited 8:04am
Quoting PipMeUp
Thanks for the answer about the size of the range. It looks like 10 pips is a good rule of thumb for the majors. I was a bit curious and decided to do some data mining with the swings. I extended your definition to include the case where two swings in sequence are almost the same length. I consider them of equal size if their heights don't differ by more than one range. I motivate this choice by saying that in a constant range bar chart one range is also the measure of volatility. I also distinguished the sub-cases in the state #4 in order to be...
there you go! Have fun finding those hidden swing VS swing relationships, that now you have all the data on hand.

Quoting PipMeUp
I then built the table of the frequencies of the states given the previous state. Fascinating! This 11x11 table tells you that if you buy the support of a range the statistically best exit method is to wait for the price to reach the resistance (i.e. not taking profit at the middle equilibrum price) then trail the SL one range size below the price. If price breaks out take profit when the end of the swing is detected (trail you SL). This is all in the frequencies table but not easy to read. Will the broken resistance turn support? This is also evidenced...
Yes I tried frequencies of the states given the previous state before and it somehow does work. I call this the 'Sequence Checker'; in an attempt to anticipate the next state given the previous state. I even tried anticipating....
the next state given the previous state x 2 or x3 or x4....
OR
the next state x2 or x3 or x4 given the previous state
OR
the next state x2 or x3 or x4 given the previous state x2 or x3 or x4

I've learned a lot just by data mining all these possibilities, and kinda understood the flow of the market much better. Unfortunately I had yet to make this finding more easier to use on a mechanical basis. Why do I say that?

There are 3 known problems;
1-In reality, states can go beyond 3 swings as you know;
http://i59.tinypic.com/1hbion.png
Squeeze and Expand states are categorized using 3 or more.
Trend and Reverse/Resume are categorized using odd numbers; 3,5,7,9...etc
Does the difference matter?

2-If we stick every state strictly to 3, we are being blind to the overall picture
e.g. a 9 swing trend; 3 swing trend pattern in the middle of the overall trend is slightly different from a 3 swing trend pattern near the end of the overall trend

3-Inversely, if I use the N count strictly as a state, then the states with higher N counts will suffer.

How do we reconcile this? One more thing to reattempt in my to do list. I know something is there but I wish not to let it go totally.

Thank you =)
  • #50
  • Aug 16, 2014 10:18am Aug 16, 2014 10:18am
Scientific Rigor is key.
Inserted Video
  • #51
  • Edited 12:27pm Aug 18, 2014 10:56am | Edited 12:27pm
Research for TCD to Range Bar Swings relationships has began. For those who actually followed the Babypips thread will know what I am trying to do. I don't mind guesses anyway. Floor's open! =)

Note : updated picture. Got the numbers in. A particular person I know will be super happy to see that I finally produced this =)
http://i60.tinypic.com/anmu6h.png
  • #52
  • Aug 18, 2014 11:59am Aug 18, 2014 11:59am
Quoting Relativity
Research for TCD to Range Bar Swings relationships has began. For those who actually followed the Babypips thread will know what I am trying to do. I don't mind guesses anyway. Floor's open! =) {image}
Is this a range bar chart? I ask because there are wicks both sides of the candles and this isn't possible with range bars.
No greed. No fear. Just maths.
  • #53
  • Aug 18, 2014 12:16pm Aug 18, 2014 12:16pm
Quoting PipMeUp
{quote} Is this a range bar chart? I ask because there are wicks both sides of the candles and this isn't possible with range bars.
Ok, modified range bars =D
Quoting Relativity
.....This leaves me with Range Bars and here are the standard rules of construction;
A-Each range bar must have a high/low range that equals the specified range.
B-Each range bar must open outside the high/low range of the previous bar.
C-Each range bar must close at either its high or its low.
..which I want to keep things really simple, so I only use A. That’s all.
  • #54
  • Aug 18, 2014 1:14pm Aug 18, 2014 1:14pm
Quoting Relativity
{quote} Ok, modified range bars =D {quote}
A- implies both B- and C-
No greed. No fear. Just maths.
  • #55
  • Aug 18, 2014 5:09pm Aug 18, 2014 5:09pm
Quoting PipMeUp
{quote} A- implies both B- and C-
OK, then I shall shamelessly call them Relative Bars. Still 'RB'
  • #56
  • Aug 18, 2014 6:29pm Aug 18, 2014 6:29pm
Quoting Relativity
Research for TCD to Range Bar Swings relationships has began. For those who actually followed the Babypips thread will know what I am trying to do. I don't mind guesses anyway. Floor's open! =) Note : updated picture. Got the numbers in. A particular person I know will be super happy to see that I finally produced this =) {image}
Pls pardon my ignorance, but being neither a statistician, nor a mathematician, I have no choice but to view your work from a 'layman's perspective. So to me it looks like you worked out a different approach to draw ZigZags using your modified range bar algorithm, with every instance of change in any given direction creating it's own leg instead of the usual 'repainting' we get with the conventional ZZs?
  • #57
  • Edited 7:44pm Aug 18, 2014 7:30pm | Edited 7:44pm
Quoting LITEchild
{quote} Pls pardon my ignorance, but being neither a statistician, nor a mathematician, I have no choice but to view your work from a 'layman's perspective. So to me it looks like you worked out a different approach to draw ZigZags using your modified range bar algorithm, with every instance of change in any given direction creating it's own leg instead of the usual 'repainting' we get with the conventional ZZs?
Partially yes. That was how I eventually created stuff like this years ago;
http://www.forexfactory.com/showthre...65#post6320065
http://kreslik.com/forums/viewtopic....r=asc&start=10

....using TCDs as a means to draw ZZs, that eventually led me to what I have today. I no longer use those works today, but I acknowledge their contribution to the overall evolution and research. Sometimes I take a part of code that's useful, sometimes I use a part of the methodology; nothing is wasted. Very often, I revisit ideas I thought that were dead ends. A good number of them turned out to show 'edges of the market' eventually, aiding to my overall work and understanding of repeatable price movement of low variance.

It also marks, 2 years since I had my last bump onto Mr SignalBender / iDouble / iDoubleStoch / 7thSignalTrader / WaveTop / (whatever, insert his known SN here). Remarkably, I never looked back. I do love to give credit where and when it is due. I don't care who he is (even though, yes, I have my personal suspicions and my own leads, not that I am not aware... but seriously, get a life? there are other more important things to do than doing 'internet police work' on someone living 123121312 miles away; let the appropriate authorities and people perform their duty if necessary)

Do note that the indicators I posted on Kreslik and Babypips are from far the same version it is today, especially for Modified Barros Swings. Please don't say I didn't warn ya! =d

I will post more later, including a little fix I've noticed that's needed plus a roadmap for development.
  • #58
  • Aug 18, 2014 8:06pm Aug 18, 2014 8:06pm
Quoting Relativity
{quote} Partially yes. That was how I eventually created stuff like this years ago; http://www.forexfactory.com/showthre...65#post6320065 http://kreslik.com/forums/viewtopic....r=asc&start=10 ....using TCDs as a means to draw ZZs, that eventually led me to what I have today. I no longer use those works today, but I acknowledge their contribution to the overall evolution and research. Sometimes I take a part of code that's useful, sometimes I use a part of the methodology; nothing is wasted....
Thanks. Clears things up for me a bit. I've just recently started paying more attention to ZZs recently myself, so I find your work very interesting, even if way over my head technically. Due to these (my) technical shortcomings, I try to 'compensate' for the inherent limitations of these ZZs, in addition to figuring out the optimal settings with respect to timeframe, by using a few other simple filters. A kind of crude workaround of what you are trying to achieve with a statistical approach. One things for sure. 1. The patterns do exist and are tradable 2. They are measurable. 3. They can even be visually identified(sometimes with the help of indicators).

For me though, when all is said and done, it's more important to make money than to be right. For me that means even when i'm wrong, what impact does that have on my equity? As long as that is well under control, my bank manager doesn't care!
  • #59
  • Aug 18, 2014 8:18pm Aug 18, 2014 8:18pm
Quoting LITEchild
{quote} Thanks. Clears things up for me a bit. I've just recently started paying more attention to ZZs recently myself, so I find your work very interesting, even if way over my head technically. Due to these (my) technical shortcomings, I try to 'compensate' for the inherent limitations of these ZZs, in addition to figuring out the optimal settings with respect to timeframe, by using a few other simple filters. A kind of crude workaround of what you are trying to achieve with a statistical approach. One things for sure. 1. The patterns do exist...
Yes I understand how you attempt to compensate, I used to do that until I figured that I needed more accuracy. I already know programming so therefore I figured might as well use it. Not that I am not making money; I am. I just do not want to leave too many things to chance, if I actually have options and choice on hand.
  • #60
  • Edited 9:28pm Aug 18, 2014 9:12pm | Edited 9:28pm
Quoting Relativity
{quote} Yes I understand how you attempt to compensate, I used to do that until I figured that I needed more accuracy. I already know programming so therefore I figured might as well use it. Not that I am not making money; I am. I just do not want to leave too many things to chance, if I actually have options and choice on hand.
Don't get me wrong, I'm convinced you are definitely making money!!! I was writing more for the sake of less experienced traders(whom I tend to empathise with a lot), and who usually get the impression that it's even more difficult making money trading without similar indepth technical research, statistical and mathematical analysis. Every traders needs are different, and I am convinced every trader needs to discover what his' or hers' is. That's why I highlighted that phrase, and you also clarified with..."I already know programming so therefore I figured might as well use it''. I believe a lot of traders need to be reassured that you can still comfortably trade well and make money without any excessive degree of accuracy in terms of the usual gauging parameters, et al, win rate, R:R, trend trading, countertrend, scalping, swing trading, etc,etc,etc....

Also, I have discovered from personal experience, that there are basically three types of rewards in trading.
1. Money (naturally)
2. The emotional rush. (especially when you are 'right')
3. The Ego Feed. (similar to 2 above, but still different)

But strictly speaking, I am of the opinion that any extra effort, extra mile and/or extra time invested into refining any already profitable pattern or methodoly, has to be sincerely and objectively weighed against the reward/results expectancy. Questions like is 'a 70% increase in effort, time, research,etc, justifiable just to increase my reward expectancy by 5 - 10%'? Of course these are personal baselines that each and every trader must define and come to terms with on a personal level, but still clarity and perspective needs to be well and fully disclosed.

As a further illustration, imagine two traders at the top of the trading leaderboards on FF live accounts feed. Both registering >400% returns for the month, and consistently topping these charts intermittently for the past 4 months. On closer investigation, I found out one uses exhaustive analysis and research equivalent to a Yale Professor of Quantum Physics to enter and exit trades for an average of 7pips per trade. He admitted it took him an approximate 200,000 dollars in time, staff and materials to fully develop and customise the tools over a period of about 10 years! The other bloke is your average joe and just started trading live 2 years ago, and according to him uses only, and I'm quoting him directly "...S&R, pivots and price action."! I still have the correspondence in my inbox. Now it's all good since both are are making money, but like I said earlier, it begs the question of effort vs reward! Again bearing in mind that of course, each trader could actually be after a different kind of reward.
0 traders viewing now