The Scientific Method : MMLC (major move life cycle) v12 | Page 4 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Digital Indicators of Adaptive Trend & Cycle Following Method 4 replies

Scientific studies on technicals 4 replies

Looking at Exchange Rates in a Scientific way 92 replies

Let's measure the spread's fluctuation (scientific approach) 0 replies

Attachments: The Scientific Method : MMLC (major move life cycle) v12
Exit Attachments

The Scientific Method : MMLC (major move life cycle) v12

  • #61
  • Edited 9:40pm Nov 16, 2015 8:56pm | Edited 9:40pm
Quoting Sciurus
I tried reading the english translation for the russian books you cited, but I didn't get very far . I think my brain needs to be in a "special place" to understand Russian translated into english with google translate, hah hah. Maybe I need to drink vodka first. Anyway, I will try again sometime, but the gist of what I read reminded me a bit of Darkstar's book.
You mean MF-V's work? Well, I didn't require vodka, but his works really opened my eyes that the market, to a certain degree, is predictable. I didn't read Darkstar's book, but I've heard of it.


Quoting Sciurus
So far we have just been using descriptive statistics
I didn't do more 'advanced' statistics since I am no math mojo. I tried but wouldn't get my head fully around it. However, I understood a simple thing; Observe the frequency distribution 'spread' from the raw data, then decide the rest from there.

I understand that use of parametrics and transformation will help us have a better view of the data, but that's out indeed of my league (or perhaps I just require further practical exposure). Great to see that you weighing in this into the thread.


Quoting Sciurus
...Turns out that both leg length and percent retrace are log-normal (skewed to the high side)....
....Something interesting going on with leg length there, as its has a slight bimodal distribution going on...
Yes, its those 2 things that bothered me earlier on when I started to investigate swing analysis without swing type classification; the log normality with a slight bimodal distribution. That is indeed related my suggestion to you, in terms of possible further optimization of swing types that we discussed about earlier. Personally, I wouldn't touch it further despite of some attempts, since it looked 'pretty nice' already. Thats when I gave it up and decided look into other means of improvement (such as my latest post), but perhaps you would dare to.

IMO one would have to rerun the swing statistics again after optimizing the Swing Type criteria. To check if there is improvement or degraded odds, against the original baseline statistics.

Quoting Sciurus
...Here is the regression of retrace level against leg length. Note that there is a significant slope (p<0.0001). Generally, as leg length increases, percent retrace decreases...
As leg length increases, retrace percent decreases. Yes that's generally seems to be true but it didn't bother me too much. It is because I did swing statistics sorted by together with previous swing type (e.g. if the previous swing is UR, find all retracement lengths of DR type).

I found the results (as attached) for all possible combination types; 18 types - URDR/URDJ/URDT, UJDR/UJDJ/UJDT, UTDR/UTDJ/UTDT, DRUR/DRUJ/DRUT, DJUR/DJUJ/DJUT, DTUR/DTUJ/DTUT, with 2 lengths being investigated- extreme and wick lengths, making a total of 36 sets of data. It turned out, at least in my (limited) judgement, to be relatively the same to the singular swing type results.
Attached File(s)
File Type: csv Results.csv   68 KB | 445 downloads
  • #62
  • Edited 10:51pm Nov 16, 2015 10:24pm | Edited 10:51pm
Yes, I may need vodka to grapple with the MF-V work . For that, I might need to wait until this weekend.

When it comes to statistics, try to see the forest (the patterns), not the trees (all the numbers and weird names). I think I know why there is a small bimodal distribution in leg length, but I need more data to confirm it. And I do not think the relationship between leg length and retracement is insignificant. Actually, the funniest thing just occurred to me about it, and it is because percent retrace and leg length have the same type of distribution. They may well be one and the same. If leg length was Independent of the previous leg's length, then this may be the exact pattern we would get. i.e. if we have a very short leg, and then we get an average-length leg following it, it would result in a retracement greater than 100%. If we have an extremely long leg, and we get an average sized leg following it, then it would result in a shallow retracement, maybe less than 38%. Oh, how I love the market. Its like a snake that swallows it's tail until it disappears.

From what I am getting from the MF-V stuff which seems to preach wave theory a lot, and your insistence on sequencing the different swing types, I am guessing that you are leading us into a wave type model for market swings?
  • #63
  • Edited Nov 17, 2015 1:13am Nov 16, 2015 11:32pm | Edited Nov 17, 2015 1:13am
Quoting Sciurus
Yes, I may need vodka to grapple with the MF-V work . ......
From what I am getting from the MF-V stuff which seems to preach wave theory a lot, and your insistence on sequencing the different swing types, I am guessing that you are leading us into a wave type model for market swings?
MF-V does have acceptable english versions of his work, which could be found online.

I actually tried to model some kind of wave theory in my older works (Rel's Lab and Rel's Forge). Unfortunately that faded away because I figured on a practical level I could just trade from swing to swing, step by step. Sequencing does help a lot, but I felt that it cannot be overly done e.g. the logic falls when more than 3 swings are involved, at least based off what I had found so far. To anticipate anything further, it is indeed a step too far and I had accepted the current known limitations.

So no, I won't go there again in this thread, as I've already made mention of 2 swing and 3 swing sequences. Anything more than 3, I stay away and try something else.
  • #64
  • Edited 11:07pm Nov 17, 2015 6:44pm | Edited 11:07pm
Decided to post this because its quite fun to be able to anticipate in advanced, the important levels price tends to respect/break.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 14 KB


Update :
Attached Image
  • #65
  • Nov 18, 2015 7:39am Nov 18, 2015 7:39am
hi rel,
please give some explanation of this term ' LifeCycle' ?what does it means or how do you define it?
thanx
  • #66
  • Nov 18, 2015 9:19am Nov 18, 2015 9:19am
Quoting Relativity
Methodology :
1-Approximately 52 + 2 weeks * 2 worth of data will be used. That makes 2 years worth.
I've decided to perform some Out of Sample and Overlapping sample testing. Each shift of data would be 54 weeks worth. I did 4 sets, which meant data back to late 2010


Quoting Relativity
2-Sorted by gap size in pips, using Frequency Distribution of all gaps found. This is to determine what is considered as significant gap size, in order to create an acceptable technical workaround/compromise, to sort out gaps stated in #b
I am quite surprised that 15 pips is an optimal number, given all the Frequency Distribution results from the 4 sets.


Quoting Relativity
3-Exclude gaps that are smaller than the determined significant size in pips.
4-Include gaps that are larger than the determined significant size in pips ;
a-compare direction of gap vs direction of swing that contains the gap
An additional exclusion is if the direction of gap vs direction of swing is opposite of each other. From the 4 sets, this happens about 7% or less, which IMO is acceptable.


Quoting Relativity
b-sort by type of swing that contains the gap
c-calculate ratio size against swing that contains the gap
d-record swing types of the previous 2 and next 2 swings, before and after the swing that contains the gap
5-Review visual confirmation of results, but do not assume anything yet
6-Tabulate results for 4a,b,c,d. Again, do not assume anything yet
I've decided to expand the amount of variables being investigated, in order to look for edges plus get a more complete picture.

DateTime = Start DateTime of Gap. This way I could do review the visual results easily, by just seeking the date on the chart.

GapSwingDirVSgapDir = D(D) or U(U) means Swing Direction (Gap Direction); see #4a

Prev2SwingType & PrevSwingType = Swing Type of previous 2 swings, before swing that contains the gap
GapSwingType = Swing Type of swing that contains the gap
FwdSwingType & Fwd2SwingType = Swing Type of next 2 swings, after swing that contains the gap

GapRatioToContainSwing = (Size of Gap / Length of Swing containing gap) * 100

PostGapExtension & PostGapExtensionWick =
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 27 KB


StartRatioVSPreviousSwing & EndRatioVSPreviousSwing =
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot2.png Size: 31 KB


StartRatioInContainSwing & EndRatioInContainSwing =
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot3.png Size: 30 KB


FirstSwingFillVSGapRatio =
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot4.png Size: 24 KB


All right, I would need some time to review this and hopefully come back with something significant. I have uploaded the 4 sets of results as well.
Attached File(s)
File Type: zip WeekendGap.zip   10 KB | 458 downloads
  • #67
  • Nov 18, 2015 9:36am Nov 18, 2015 9:36am
Quoting manish9300
hi rel, please give some explanation of this term ' LifeCycle' ?what does it means or how do you define it? thanx
The name was based off pure observation over the years before I eventually improved and solidified the current version of MMLC.

Each independent swing, within its internals, go through certain phrases in my experience. For this example; Reversal->Trend->Accumulation->Continuation->Accumulation->Reversal. To me, thats a sort of a 'Life Cycle'.
Attached Image


Not all swings go through the full thing we seen above. Some just go Reversal->Trend->Accumulation->Reversal and so forth.

Note this is not new. e.g. one can look at the Frank Tubbs Model. Or even Wyckoff. What *might* be new (or perhaps theres someone who did it way better than I do), is to provide supportive descriptive statistics that are related to how price moves, relative to the model. This overall strengthens the model as it becomes less subjective/discretionary and more objective/mechanical/systems based.
  • #68
  • Nov 19, 2015 3:54am Nov 19, 2015 3:54am
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 25 KB


Please note that this is just the general model. The specifics... frankly... I don't feel like posting... haha.... because the general is really good enough for most to get some idea. To be able to obtain this general model from the research process is already a result itself, to say that this potential additional component just needs a little more work for successful integration with MMLC v12. The specifics... *rubs hands with glee*
  • #69
  • Edited 3:17pm Nov 19, 2015 12:25pm | Edited 3:17pm
Quoting Relativity
{quote} The name was based off pure observation over the years before I eventually improved and solidified the current version of MMLC. Each independent swing, within its internals, go through certain phrases in my experience. For this example; Reversal->Trend->Accumulation->Continuation->Accumulation->Reversal. To me, thats a sort of a 'Life Cycle'. {image} Not all swings go through the full thing we seen above. Some just go Reversal->Trend->Accumulation->Reversal and so forth. Note this is not new. e.g. one can look at the...
Hi Rel,
I'more passionate about this tread and I would ask you a thing about the quote...
could you explain if there is some method to identify the Accumulation zones of Life Cycle?

Thanks in advance.

W.
  • #70
  • Edited 4:45pm Nov 19, 2015 4:16pm | Edited 4:45pm
Quoting woww
{quote} Hi Rel, I'more passionate about this tread and I would ask you a thing about the quote... could you explain if there is some method to identify the Accumulation zones of Life Cycle? Thanks in advance. W.
I am sorry but I don't have to explain that. It's already right in front of us, that part of the swing formula does it. And I will not answer questions further that is not sufficiently related to the purpose of this thread. That's why the requirements exist. I have already stated the requirements and instructions very clearly. Try it and I am sure you will get your answers for the questions you have.

However if you really did try the steps provided and have questions after doing so, then it would be a lot easier for you and me to engage further. Simply because things became more practical and hands on. This is very important and very helpful to the overall learning process. The time of just talk and talk and questions has to end. Trading itself is practical. Try it and you will know. What I tried is to shorten this learning curve. Hope you can understand.
  • #71
  • Nov 19, 2015 5:19pm Nov 19, 2015 5:19pm
Quoting pete7
{quote} http://lmgtfy.com/?q=frank+tubbs+model#seen first graphics
Thank you Pete. Yes that's a good starting point but I have my problem with the tubbs model too. I mentioned this earlier, that the lack of practicality prevents others from ulitizing the model to the best possible. That being said, its still a start to look into his work.
  • #72
  • Nov 19, 2015 5:32pm Nov 19, 2015 5:32pm
Quoting Relativity
{quote} I am sorry but I don't have to explain that. It's already right in front of us, that part of the swing formula does it. And I will not answer questions further that is not sufficiently related to the purpose of this thread. That's why the requirements exist. I have already stated the requirements and instructions very clearly. Try it and I am sure you will get your answers for the questions you have. However if you really did try the steps provided and have questions after doing so, then it would be a lot easier for you and me to engage...
Ok . PM for you.

Thank you Pete for the answare but i dont' think is the same of Relativity.
  • #73
  • Edited 6:38pm Nov 19, 2015 5:46pm | Edited 6:38pm
I would explain a little further why I don't wish the indicator to be posted.

I did wish I myself can put out the lite version of the indicator out, but I learned the hard way that this isn't going to help anyone. They are not going to build up the tenacity required as a trader in the long run, if they got things 'the easy way out'. I know it sounds horrible but hear me out.

It's not about hiding secrets. There is no secret. The instructions for the indicator construction are out in the clear. It's about learning as much as possible what the indicator can offer. If you had a hand of its construction, you would be able to assess for yourself firsthand if MMLC v12 is for you, plus how does it line up with your current trading methods.

However, if you downloaded or got the indicator from someone else, part of the learning process is lost. Each step of the construction has a particular purpose and would contribute to the advanced versions of the work. Yes you got the indicator, but its difficult to advance it further because of the weaker foundation, as compared to someone who started from scratch.

A person who got the indicator from someone else becomes a 'user'. A 'user' is subjected to known strengths but also unknown weaknesses of a 'product'. No amount of 'study' or 'analysis' can change that. The 'user' is still limited to the pre-set power of the indicator, without updates, without bug fixes, without tuning.

A person who followed the instructions of construction becomes an 'owner'. It's rightfully theirs now and they hold a lot of power. They can tune it as much as they want to compensate for any weaknesses found, or to their liking add on strength to their existing works. They can do whatever they want, 'investigate' anything they see of interest.

Tell me then, which is more powerful?
  • #74
  • Nov 19, 2015 6:09pm Nov 19, 2015 6:09pm
Quoting woww
{quote} Ok . PM for you. Thank you Pete for the answare but i dont' think is the same of Relativity.
I am sorry but this thread is not about 'answer from Relativity is the best'. Please don't make that assumption.

In fact I respect Pete's work and answers because he has shown clearly that he did try the instructions, and did found 'something'. And hence is qualified to not just participate, but also critic in my work. That I appreciate, really.
  • #75
  • Nov 19, 2015 7:35pm Nov 19, 2015 7:35pm
All right, time for me to bow out of this thread.

Its been only two weeks, but the thread's purpose (and extended purpose) has been sufficiently fulfilled.

My target audience is just to have a few peeps who are able to fulfill the requirements stated, and bring something onto the table. That has been fulfilled as well. What is missing a little are further ideas to take this indicator forward, via additional components. But at least I tried to make up for it via an example that is still in beta stages, and hopefully into production, as it looks very promising.

Its understandable that this isn't achieved because much of the baseline indicator would require time to sink in. It took me many years to achieve this, and much of the time is actually about solidifying the swing formula. That it would capture the most important price points. Then later its more of seeking every single angle and measurement the indicator would offer. And then checking if the measurement provides some kind of edge. I had shortened this process by pointing out the essentials edges that should kick-start for anyone who has coded and tried the indicator.

Like all other threads I've started, its all right that if it is 'left for the dead'. The work here is a legacy, one that should survive the test of odds when applied practically. Its the best that I could come up with and I've decided to leave the foundation as it is. And in a sense, its my gift to the trading world (if there is such a term) out there. I had taken painstaking care to ensure the gift is well presented and of practical use, so long one actually tries it out.

Those of you who have my private email, just contact me when you like to. Its always nice to chat up. Focusing rather on the precious few rather than many helps me a lot and yes that's selfish haha. But that's also why the strict requirements exist. For those who did so to fulfill them, they clearly have my attention and respect. Its not like anyone can just contact me via email, please understand that I don't just allow anyone to do so.

So to sum it up, well done for those few who actually fulfilled the requirements. I would be available via PM (and private email later - just ask) should you have further questions down the road.

Additional Note : its only a demo account, but surely I had a lot of fun testing it out with MMLC v12.
Attached File(s)
File Type: zip Statement 6730882.zip   17 KB | 620 downloads
  • #76
  • Nov 21, 2015 12:44pm Nov 21, 2015 12:44pm
i hope i am not too late.
http://i68.tinypic.com/14kjn90.png
  • #77
  • Nov 21, 2015 4:34pm Nov 21, 2015 4:34pm
Thank you Rela,

I think this has been very useful. I am drawing out the frequency distributions as I comment and finding it immensely useful. I have more to work on this, but will crack on with it.

Its a beautiful insight indeed, thank you for the demo statement too.

Cheers
Silver
  • #78
  • Jun 16, 2016 7:08am Jun 16, 2016 7:08am
Quoting Relativity
{quote}
This belongs to the realm of additional components.
Without giving too much away;
1-I did consider news. I generally do not worry about news, because it is actually very surprising to many (even myself early on), that news respects the general swing levels and will break them when necessary. I had done my own statistics involving news, using data from Jan 2007 to Nov 2015. Its admittedly a lot of work (e.g. creating a separate indicator that sort swings with every piece of news), but well worth it to really tell me not to worry so much,...
Just to post something up since what happened today is a very good example :
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot2.png Size: 74 KB

One can just see statistical wise, its good enough to just follow the current line direction both just before and after news. This is especially true when price is in T or R state. The numbers on the left are just way too good edge wise e.g. DT to DT and UT to UT at 85%. UR to UR/UJ/UT is at 90%. DR to DR/DJ/DT is at 100%. FYI MMLC is now at v16.
  • #79
  • Edited 6:41am Jun 27, 2016 4:50am | Edited 6:41am
A poll has opened up on the 1st page of this thread.

MMLCv16 Core Components consists of the following :

1-M30 Swings as Lower Timeframe
2-Dynamic Custom Timeframe Swings as Higher Timeframe
3-Combined Lower and Higher Timeframe as Multiple Timeframe Swings view
4-Lower Timeframe interacting with Higher Timeframe Trend Channel levels on Multiple Timeframe Swings view
5-Lower Timeframe key swing ratios with Higher Timeframe context on Multiple Timeframe Swings view

In view of Brexit, a particular beta version Additional Component has shown not just great promise, but actual tangible results. It has great potential to be included as a Core Component in future, so long further research and further tuning could be done on it. This additional component is a modified version of B2i (iDoubleStoch's work and idea). It appears to work particularly well as an independent tool with #1 and #2.

While I am currently reluctant to disclose the full version of MMLCv16, I have a reasonable suggestion. I would like to open up the following here :
-M30 Swings as Lower Timeframe (MMLCv16 Core Component #1)
-Modified version of B2i (beta version Additional Component)

Time to vote
  • #80
  • Jun 27, 2016 7:22am Jun 27, 2016 7:22am
A box of 7 bars on M30 is used to represent price swing and further various types of boxes is used to represent life cycle of a trend swing.

It is very interested to see usage of scientific method to study swing trend in forex. If price can be qualified by statistics, statistical inference, etc scientific method theory, then, trading can be bit of easy.

Trading is based on speculation of price movement, greedy, fear, manipulation, etc. Speculation will often not follow statistics, etc scientific way.

Of course, in-sample data or macro format, etc. can be used to build the models, and out-sample ones can be used to test the model. Then, the model is used to forecast next swing cycle or each phase of swing cycle. Price move are all based on speculation, manipulation, etc, Probability etc concepts are not accuracy. Price move or trend has the cycle. It is hard to spot the cycle of a swing. In news, george soros long cable before UK vote, he lost huge due to fast drop of pound? Huge swing often seem somebody got huge loss? 1700 pips drop, killed all soros account if he has no hedge.
0 traders viewing now