Rel's Forge | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Rel's Philosophy of Trading Discussion 24 replies

Rel's Evaluation 13 replies

Rel's Lab 103 replies

Attachments: Rel's Forge
Exit Attachments

Rel's Forge

  • #21
  • Oct 15, 2014 12:45am Oct 15, 2014 12:45am
Quoting MAM.Trader
if you are still using fixed periods for determing siwings that wil fail for sure. we can not fit market into fixed periods. more importantly it is not possible when the case is zig zag & swing points.
Yes we cannot fit the market into fixed periods, but to clarify... IMO we still need an acceptable dynamic range of fixed periods to measure the market on a quantitative basis. This dynamic range must self learn and adaptive to adjust to the market conditions, to allow the volatility to 'stretch as far as it wants to'. Its abit of a catch 22 situation. This is why I talked about context in the 1st few posts.

Quoting MAM.Trader
in my thread some posts earlier I explained how a swing point can be coded. if you ask here at FF or your trader friends many will show you on chart. but when its time to create algorithm (logical explanation that can be coded) for the swings that is the point where those guys/gals stuck. I am working on zig-zag issue for a loooooooooooooong time. its really difficoult to handle & find an edge. I think at last I found one. still forward testing. may be we can dicuss later what we have in our hands. suncribed to thread keep up the good work GL {quote}...
Yes thank you. Its not easy to solve the issues, create and test algorithms, all at the same time. The good thing is that if you read my lab thread, a lot of it talks about the overall process and thought put into it. Unfortunately, I decided not to discuss about the specifics of my swing algorithm too much, as the algorithm by itself is an edge as far as I had observed.

Again basically the formula is actually very simple and has its roots from the PH-L / PL-H TCD, combined with a small part from Barros Swing's formula. That is as far as I can describe. There some minor revisions to improve the flexibility the last 6 months, but the most important core basics have not changed.
  • #22
  • Oct 15, 2014 12:54am Oct 15, 2014 12:54am
hi rel
I did not asked for the algorithm u created but its good to hear your approach before I put years of reserch results and thousands of code lines on the table.

anyways GL

Quoting Relativity
{quote} Yes we cannot fit the market into fixed periods, but to clarify... IMO we still need an acceptable dynamic range of fixed periods to measure the market on a quantitative basis. This dynamic range must self learn and adaptive to adjust to the market conditions, to allow the volatility to 'stretch as far as it wants to'. Its abit of a catch 22 situation. This is why I talked about context in the 1st few posts. {quote} Yes thank you. Its not easy to solve the issues, create and test algorithms, all at the same time. The good thing is that if...
My Best Trade is Next Trade!
  • #23
  • Oct 15, 2014 1:04am Oct 15, 2014 1:04am
Quoting MAM.Trader
hi rel I did not asked for the algorithm u created but its good to hear your approach before I put years of reserch results and thousands of code lines on the table. anyways GL {quote}
Yes you didn't ask for the algorithm. Its just me considering how much I should show the public, that's all.

Well... surprisingly, the more I improve the principles of my work, the less number of lines of code that showed up, progressively for each version.
  • #24
  • Oct 15, 2014 8:01pm Oct 15, 2014 8:01pm
Important post. Just marking it down here as well for reference;
http://www.forexfactory.com/showthre...86#post7804186
  • #25
  • Edited 5:44pm Oct 18, 2014 12:23pm | Edited 5:44pm
I used to had a very hard time accepting the use of averages, until I saw this plus thought about my formula for range bars;
Quoting TradeVector
.....
You will need to pick a time-frame (a bar of data) and then determine the price delta between two (2) clearly identifiable points within that bar of data. Then measure the average delta between those two points over X number of bars. Use that as your baseline. Study that baseline until you know when the delta is expanding, or contracting.
.....
Forget conventional TA, everybody and their dog uses it. You are going to have to build your own TA from raw data. So, start with raw Forex data and begin the process of studying it and analyzing its...
The keywords is 'baseline', 'expanding' and 'contracting'. I suppose normalization of OHLC data regardless of chart being used is important to find and understand patterns correctly. Often, I see volatility as the problem to be solved, as it shows up as directionless increased magitude. Volatility expands and contracts. Such differences of scale can make certain trading rules unstable e.g. trade breakouts of 20 pips. Hence this is why I prefer to use ratios.

But this is also where I realized that of me using range bars, there is also a baseline set in place; the values to generate the range bars themselves. The range bar values are also used to project the next day's volatility on a micro scale. Therefore I went back to the range bar generation indicator, did some fixes and did this;
http://i57.tinypic.com/igvold.png
http://i61.tinypic.com/2u4u3d1.png
Using the values of range bars, I created some bands;
Middle line ; Middle price of the 'grey boxes'
Upper bands ; + x2.5 and + x5 value of projected range bar value for next day
Lower bands ; - x2.5 and - x5 value of projected range bar value for next day
This looks way way better than using moving averages and these bands seem to have confluence with the actual swing turn break levels?
  • #26
  • Oct 18, 2014 5:39pm Oct 18, 2014 5:39pm
Another important post about creating true trading systems.
http://www.forexfactory.com/showthre...96#post7809896
  • #27
  • Oct 19, 2014 3:18am Oct 19, 2014 3:18am
Another overdue work. Another attempt at Allied Currencies. Gathering all relevant pairs D1 swing data.
http://i59.tinypic.com/sqnr48.png
http://i61.tinypic.com/357ovfc.png

http://www.forexfactory.com/showthre...41#post7810141
  • #28
  • Edited 4:50am Oct 19, 2014 4:37am | Edited 4:50am
How to read all this (at least manually).

EuroZone Crisis and SNB intervention. Money is escaping from EURxxx and becomes 'lead driver' of the market.
http://i61.tinypic.com/qoj23m.png

Things look calm for xxxJPY during the same period.
http://i62.tinypic.com/wbdkax.png

Now EURxxx takes a rest.
http://i57.tinypic.com/dpbtro.png

But xxxJPY takes over the role as 'lead driver'
http://i61.tinypic.com/4g4scw.png

My eventual goal is to real and anticipate all this on a almost mechanical basis. I used to create my own currency indexes before, but IMO theres still a lot of work needed to be done to tune them correctly. Back then, its also interesting that someone did mentioned that my USDxxx index almost seemed to correlate like a particular US stock index (Dow Jones if I am correct).

Correlation is nice and is good to know. I am not exactly looking at correlation but beyond it. This is a very important study IMO because it correctly identifies the true nature of trends for currencies. Regardless if you are trading a single pair or multiple pairs, knowledge of this gives you another edge and reason why certain market states happen for the pair you are focused on.
  • #29
  • Oct 19, 2014 10:40am Oct 19, 2014 10:40am
Quoting Relativity
{quote} You got it right. The percentages/stats are just the against N layers, depending on the type of level. I've already revealed the one (1) type of level in the last few posts, which is the swing turn breakout level. I call it 'CurrSwingPostBreakoutTurnLevels' {quote} No. {quote} Logically 1 pip is needed to break any line. The 70% edge is price to go beyond more than 1 pip at the breakout. Inversely, you only play the reversal 30% of time. These percentages are of a single MTF/MPF layer. These percentages will change when we do it on multiple...
Thank you very much for the clarification !
I am late, I have to catch up with what you posted next ^^'

Your post with the 'averages' and also the last one about the driving currencies are very interesting. Thank you : )
All this is really becoming a very informative and complete framework about the market if I may say. It is really nice to observe.

Edit : and thank you for the Al Pacino's speech ! Inch by inch...

030985
  • #30
  • Edited Oct 22, 2014 12:34am Oct 21, 2014 8:41am | Edited Oct 22, 2014 12:34am
From
http://www.forexfactory.com/showthre...31#post7801231
(with updates)
http://i58.tinypic.com/6ga06o.jpg


Dream come true =) This deserves to be called MMLCv4. Did some tested on minimal retracement lengths.... and the minimal 30% retracement 'law' still holds around 99% of time.
http://i60.tinypic.com/1p9pgx.png


http://i61.tinypic.com/dewtok.png
Attached File(s)
File Type: xls 2nd swing vs 1st swing length.xls   171 KB | 523 downloads
  • #31
  • Edited 9:49pm Oct 21, 2014 9:35pm | Edited 9:49pm
Done! Customized Swings + Transient/Recurrent Zones + MTF/MPF analysis + Market States
http://i60.tinypic.com/racgll.png

Seems to work for Range Bar charts too;
http://i60.tinypic.com/2cyk1ex.png
Attached File(s)
File Type: tpl EURUSD H1.tpl   825 KB | 586 downloads
  • #32
  • Oct 22, 2014 12:31am Oct 22, 2014 12:31am
Quoting 030985
{quote} Thank you very much for the clarification ! I am late, I have to catch up with what you posted next ^^' Your post with the 'averages' and also the last one about the driving currencies are very interesting. Thank you : ) All this is really becoming a very informative and complete framework about the market if I may say. It is really nice to observe. Edit : and thank you for the Al Pacino's speech ! Inch by inch... 030985
Thank you =)

I posted that because any most mathematical std dev and outier calculation will involve some form of average, hence I decided not to deceive myself and go straight on the record; that yes, my range bar formula does use some kind of average. I do think averages should not be used 'for the sake of it'.
  • #33
  • Oct 22, 2014 2:12am Oct 22, 2014 2:12am
Quoting Relativity
Done! Customized Swings + Transient/Recurrent Zones + MTF/MPF analysis + Market States {image} Seems to work for Range Bar charts too; {image} {file}
Very nice setup!
k
  • #34
  • Oct 22, 2014 7:58am Oct 22, 2014 7:58am
Quoting kprsa
{quote} Very nice setup! k
Thank you =)
  • #35
  • Oct 22, 2014 9:16am Oct 22, 2014 9:16am
IMO my take on finding an optimized H value.
http://i57.tinypic.com/11b458i.png
Attached File(s)
File Type: xls TZstats.xls   1.4 MB | 562 downloads
  • #36
  • Edited 10:45pm Oct 22, 2014 9:17pm | Edited 10:45pm
I want to laugh. Really... not offending anyone but the 2 threads regarding TZ are both starting to become like a circus.

Not hard I guess. Just take the concept and make your own tools out of it, see it for yourself. Everything else is commentary.

Admittedly, there are problems with TZ... 4 main problems as far as I know.
1-The optimal H value is going to be different across for each timeframe. The statistics I've posted is for range bars, which eliminates this problem.

2-Sometimes an optimal H value isn't the answer, because MTF/MPF analysis makes the H value fall apart. e.g. Optimal H value for lower timeframe forced into outier value due to higher timeframe 'stretching it. This is especially true for a ranging market. Again, range bars avoid this problem.

3-I am not saying an optimal H value for each timeframe doesn't exist. It does, but a combination of #1 + #2 calls for situational H values to be used. This leads up to market states analysis.

4-H value has 2 sides; the left and right side. The left side is what I call the 'reference'. The right side is what I call the 'projection'. The left and right side need not necessarily always be 1:1, although seems to me, successful transient zones tend to be 1:1 or greater. For practical value, its best to leave it at 1:1, as there are 2 types of prices after the zone has formed; 'Price passed after zone expires' (like a lid being lifted to allow price to return and breakout) and 'Price did not pass after zone expires' (price did not return but went further).

http://www.forexfactory.com/showthre...19#post7818619

#1 + #2 + #3 + 4 are not easy to see without some help from an 'indicator', hence my earlier post for the H1 EURUSD template. What I attempted is a brute force approach, to test 100 periods. Surprisingly, the code is less than 600 lines and did not produce any lag. So IMO such analysis can be done, but it does take an experienced hand to logically code it correctly.
http://i57.tinypic.com/2ecf4ab.png
  • #37
  • Oct 23, 2014 8:41am Oct 23, 2014 8:41am
I am done. I know there are at least 2 things that I want to look at ; Allied Currencies and Market profile, but at this point I want to spend more time on these new tools. Time to close the forge. Questions are still welcomed =)

Combining v3 and v4 together;

Macro Analysis
http://i62.tinypic.com/oiszy9.png

Micro Analysis
http://i58.tinypic.com/2nlecnc.png
  • #38
  • Edited 11:20pm Oct 24, 2014 11:02pm | Edited 11:20pm
Revisiting an old thread after I realized I could apply what I have in v4 to this; http://kreslik.com/forums/viewtopic.php?t=2503
Having some fun trying to combine Macro and Micro together using PH-L and PL-H TCDs on a MTF and full vector analysis.

http://i59.tinypic.com/jaa0p3.png
http://i57.tinypic.com/5x1nnq.png
http://i61.tinypic.com/105edlg.png
  • #39
  • Oct 24, 2014 11:53pm Oct 24, 2014 11:53pm
All right enough I think... Time to head out for a good weekend =) I don't have stats to check if the PH-L/PL-H TCD break will make price turn and continue running more than 90% of time (like the gray boxes for swings), but no harm observing this as well.
http://i61.tinypic.com/256ba7t.png

Thats it... no more... Its very tempting to browse the forums and post up stuff; this I got to admit. I do love to share and discuss, but at some point, it has to end.

Signing out! =) See you at 2016. Trade well.
  • #40
  • Nov 3, 2014 2:50am Nov 3, 2014 2:50am
FYI; IMO Excel is still the best tool for indicator development. Its good that I've came to full circle regarding this.

Its really not hard to stream MT4/MT5 OHLC data to Excel. Hence relegating MT4/MT5 to charting only and delegating analysis to Excel seems to be the best fit.

Graduation time! =)

http://i57.tinypic.com/2h7e1c4.png
0 traders viewing now