The Scientific Method : MMLC (major move life cycle) v12 | Page 3 | 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

  • #41
  • Nov 11, 2015 9:25pm Nov 11, 2015 9:25pm
Quoting blueseahorse
{quote} in my understanding, the swing takes time or candle count to develop. there are truely a lot of retracements happened within the swing , even within one single candle, not only 0-10% retrace ,maybe 30%,50%... even more... in a candle stick chart, the swing formation is combined with price level plus time/candle count. just my humble opinion...
Yes this is true, and realistically we don't and can't watch all the retracements happened within the swing. This is the problem I have with classical MTF.

Yes, the swing formation is combined with price level plus time/candle count. You got this right, no longer an opinion but a fact you are pretty sure about =)
  • #42
  • Nov 11, 2015 10:25pm Nov 11, 2015 10:25pm
Quoting Relativity
{quote} Yes this is true, and realistically we don't and can't watch all the retracements happened within the swing. This is the problem I have with classical MTF. Yes, the swing formation is combined with price level plus time/candle count. You got this right, no longer an opinion but a fact you are pretty sure about =)
thank you for the reply, rel.
may i ask if i can post some pics using barros swing indicator(too difficult for me to code my own indicator) to join in the discussion, please?
patience is a virtue
  • #43
  • Nov 11, 2015 11:01pm Nov 11, 2015 11:01pm
Quoting Relativity
No of coz picking random spots for price to stop and then retrace minimally would be a bad idea. Yes, if you can project where the current swing end would be, that is the 1st thing to look for, then the minimal retrace comes 2nd. There is no other way around this, one needs to reduce risk first.

If the current swing is a retrace type, then one can more or less spot where it can stop or at least struggle to get through.
If the current swing is a reject type, then one can know but with less certainty, where it can struggle and reverse.
If the current swing is a trend type, then one has to use other methods to find that final swing end.

Price needs time, momentum and space to get through quite a number of 'barriers' before it can take off
Quoting Relativity
Not quite, but at least knowing that price needs to struggle through this area in some manner, smartly forces the trader to devote screen time when price reaches here. This is part of the edge I am talking about. One cannot do screen time all the time, but this is one key area I strongly encourage traders to watch and learn.
Wanted to post this earlier but decided to go for lunch 1st. FYI the future down swing has already printed at this time of writing.
Attached Image
  • #44
  • Nov 11, 2015 11:05pm Nov 11, 2015 11:05pm
Quoting blueseahorse
{quote} thank you for the reply, rel. may i ask if i can post some pics using barros swing indicator(too difficult for me to code my own indicator) to join in the discussion, please?
The barros swings indicator IMO will not do because of the swing formula differences. I had found that it tends to over filter out too much price action in my experience. Thank you for the offer but sorry...
  • #45
  • Nov 11, 2015 11:42pm Nov 11, 2015 11:42pm
Quoting Relativity
{quote} The barros swings indicator IMO will not do because of the swing formula differences. I had found that it tends to over filter out too much price action in my experience. Thank you for the offer but sorry...
ok,no worries. i will keep to myself.
you guys please continue, i will just sit on the bench and watch
patience is a virtue
  • #46
  • Nov 12, 2015 6:50pm Nov 12, 2015 6:50pm
Quoting pete7
Thank you for your reply. Hope you will not mind if I rephrase wick ratio formula into:
a) if the current swing is an upswing WR=(max(close, open) - low) of the last candle in the swing, divided by last swing's size,
b) if the current swing is a downswing WR=(high - min(close, open)) of the last candle in the swing, divided by last swing's size, I think such formulation is a little simpler.
upswing WR=(max(close, open) - start of current swing))
downswing WR=(start of current swing - min(close, open))
Yup I don't mind, it does work and I didn't know about max and min functions. Good thinking. Thank you.

Quoting pete7
1. Having said that, could you please specify what is the rationale behind using wick ratio ? One thing I can think of is checking for proper classification of reject swings - their extreme ratios > 100%, while wick ratios are < 100%. Apart from that I cannot see any usage of wick ratio.
Watch the difference between the bell curves of Extreme Ratios VS Wick Ratios. Its obvious that the classification rules will create the following :
a-the top of the curve is more centered for the Wick Ratios compared to Extreme Ratios
b-create a 'gap' at the bottom of the bell curve for the Wick Ratios
So yes, wicks will happen, but how big can they be so that we can anticipate them? What is not obvious is that a + b together serves a purpose to tell us this. The 85.1 to 100 zone can be seen as a 'last chance to form a wick here, or else there's no turning back'.


Quoting pete7
2. I do agree that measuring swings and finding repetitive phenomena which are statistically significant can provide an edge. But to translate such, theoretical edge, into something tradeable is another story. For example, swings which consist of one candle are included in the study but are not tradeable as I see it.
It is actually tradable. I can tell you straight off that one of the keys is to simply wait for the candle to close (certainly a true test of my patience!). We had only done a general study and found something significant as a start. Now its time to create more detailed and specific studies for these type of swings. Anything is tradable, so long one finds a proper way to exploit them. We now have the tools to deal with them data; sort them out and find those quirks out.

Here's another interesting exercise I recommend one to try, to attempt to answer questions #1 and #2 further ;
a-create a fib grid like this
Attached Image

b-and go around applying this grid from swing to swing.
c-if you are on mql, project the zones using some kind of shift function, from swing to swing.
d-observe how the next swing reacts from zone to zone

Quoting pete7
Furthermore, because of using window of length 7 we introduce a lag - current swing will be confirmed if in the next 7 candles no new low or high will be printed. How do you cope with that ?
I don't fully understand you by 'current swing will be confirmed if in the next 7 candles no new low or high will be printed'. Or did you meant by 'current swing will not be confirmed if in the next 7 candles no new low or high will be printed'? Either way, I hope I get you. I would explain it here a little longer.

Older versions of the swing did not have the price breakout filter (major move final candle must be the last candle of the box) and hence there's obvious lag, especially when used live. Adding the price breakout filter makes the swing 'follow the newest and most important prices, that carries momentum as soon as possible'.

The price breakout filter turned out to be the difference maker; it made the swing indicator usable for live trading and also improve the statistical edge further than expected.

Now that the concern about lag is addressed, because at least the indicator is optimally following price. Then thats where the zone projections come in to bring in the 'leading indicator' element.


Quoting pete7
Here are density plots green are produced using whole sample, blue ones are computed by using only swings with length of 7 and more: http://s000.tinyupload.com/?file_id=...90419875560112 By not much but the differ, especially in case of reject swings. Regards, P.
I am not sure whats the specific purpose of these plots, but they sure look very nice and organised.
  • #47
  • Nov 13, 2015 12:24pm Nov 13, 2015 12:24pm
Heres one of my early attempts at the Additional Components. The way I currently see the grand picture is that the swing model is correct, and the theory behind the zones is correct. From here my goal is to try to pin point actual trading areas a bit better. For example extensions beyond 1.20 are quite hard to project (for us mere mortals) and the ideal retracement range between 40% and 85% is a bit soft and at times too large. Retracement trading provides a clear hard stop (beyond 100%) so its the area Im focused on first.

Even in this specific area, there are two kinds of assistants: projection vs reaction (I think price will top out here in the future vs I dont think price will take out the current extreme). Reaction produces a later, but slightly stronger signal, and in my opinion its slightly easier to find knowledge about the market using this method.

So, Im only looking at the areas between 40% and 85% of the previous swing for potential trade entry in the original direction (up, retrace 40-85% down, trade signal, long). The idea is to filter out as many bars in this area as possible and just leave the ones that give a good signal. For this I used a single post-extreme TCD reversal bar. That is, I wait for a bar that lands somewhere in the price range, wait one bar for a TCD reversal, and see how many of these appear in all the swing types.

For those who are unaware of what a TCD looks like:
TCD Long= H-pL, where p stands for previous
TCD Short= pH-L
TCD signal= TCD Long-TCD Short. So if Im looking to buy, Im looking for a positive TCD signal out of the two TCD components.
example of a TCD short signal
Attached Image


Note that I'm only using TCDs at extremes and the follow up bar (singular) so there are instances where there is no signal at all, which would look something like this:
Attached Image


And the results are below:
Attached Image (click to enlarge)
Click to Enlarge Name: results.png Size: 25 KB

Before I talk about them, I will say that I like these results considerably more than using a single post-extreme close bar. I have not tried a traditional 3 bar pattern (pre-extreme, extreme, post-extreme) or 5 bar fractal, though I will definitely try them sometime this weekend. Moving on..

A 0 count to me means either one of two things is happening: either price isn't stopping through the zone (it's got things to do) or it pauses/reverses in a sort of "choppy" fashion, such as shown by the screen shot above. There a couple of ways of "untangling" this and now that I'm writing this I think sorting this issue might be worthwhile. Anyhow, if one is willing to wait for just 1 bar (the TCD reversal close), the odds become quite stacked in favor of either a rejection or retrace swing. The "Net" column on the right shows an updated probability of either of those swings occurring as the counts grow. The only "catch" is that without further analysis, one can't quite set the stop to the extreme so fast. It's possible (in a long scenario) that price makes a low extreme, prints a TCD reversal bar, makes a new low, does not make a TCD reversal but stops and reverses long anyway. In this case everything has played out in a very predictable way, but the stop has been eaten.

I don't have the indicator because all my work is in excel (heh), but maybe someone else can take a look at it, try it out, or come up with a better idea.
"Mrrraggglhlhghghlgh"
  • #48
  • Nov 14, 2015 11:11am Nov 14, 2015 11:11am
Hi guys,
i'm studing very hardly for many days this approach to trading and I'm thinking how can this approach makes profittable (and maybe one day I will) making many demo test on the chart..
Unfortunatly i'm not a coder and so I cannot partecipate with activity that I would on this discussion.
Now I have a question for Relativity. How do you apply the scientific method in relation with red news? Do you not consider? Because the statistich of swing don't consider.

Plese consider that english is not my mother language and forgive any grammathical errors.

Thank in advance.

W
  • #49
  • Nov 15, 2015 1:55am Nov 15, 2015 1:55am
Quoting LiquidGenius
Heres one of my early attempts at the Additional Components. The way I currently see the grand picture is that the swing model is correct, and the theory behind the zones is correct. From here my goal is to try to pin point actual trading areas a bit better. For example extensions beyond 1.20 are quite hard to project (for us mere mortals) and the ideal retracement range between 40% and 85% is a bit soft and at times too large. Retracement trading provides a clear hard stop (beyond 100%) so its the area Im focused on first. Even in this specific...

*sits back and watch*

No seriously lol, this thread is open now for you to test out stuff like this, to find your own additional components. Just do your own thing, and post results and conclusions if any.
  • #50
  • Edited 2:51am Nov 15, 2015 2:03am | Edited 2:51am
Quoting woww
Now I have a question for Relativity. How do you apply the scientific method in relation with red news? Do you not consider? Because the statistich of swing don't consider.W
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, good or bad news.
2-In fact there are too many instances when I am trading, I didn't know or bother about news but just the levels that are being respected or broken, then I respond accordingly as usual, as if there is no news at all!
3-One can also look into MF-V's work regarding this. He also mentions the same thing :
http://masterforex-v.su/forex001001001.htm
http://masterforex-v.su/forex001001002.htm (most important)
http://www.masterforex-v.su/002_017.htm

Frankly, this question is close to a holy grail type of question, but also with really surprisingly answers (especially #2)!
  • #51
  • Nov 15, 2015 2:16pm Nov 15, 2015 2:16pm
Ok.
this is my 2 attemps.

1)
I make this consideration. When do swing appear?
Before a new swing appear there is a rule that has to respect -> for example down swing:
- or there is the minimum of the 7 group of M30 candels in the last candel,
- or have to incercept and pass the signature down,
wathever type swing will be.
So..at what percentage of previous swing this happens?
And so:
- after 25% -> probably current swing must close in 45%- 85% zone (and retrace?)
- after 85-> probably current swing must close over 100% zone (maybe over 120%??)
- near 100% -> probably downtrend and current swing must close over 120% zone.

2)
I have used the table of the percentage of swing provided by mr. LG and I have put in to one unique printout all down swing (down rejwect + down retrace + down trend) and Up swing.
Unless errors, I calculate percentage of all down swing that touch at least at (about the same for Up swing):
50% zone -> 80%;
85% zone -> 50%;
120% zone -> 30%;
155% zone ->20%;
200% zone -> 10%.
and so i think maybe profittable build a piramid of order with decrescent lot size.

I'm studing market and so I hope don't say too many rubbish.

W.,
  • #52
  • Nov 15, 2015 4:56pm Nov 15, 2015 4:56pm
Quoting woww
Ok. this is my 2 attemps. 1) I make this consideration. When do swing appear? Before a new swing appear there is a rule that has to respect -> for example down swing: - or there is the minimum of the 7 group of M30 candels in the last candel, - or have to incercept and pass the signature down, wathever type swing will be. So..at what percentage of previous swing this happens? And so: - after 25% -> probably current swing must close in 45%- 85% zone (and retrace?) - after 85-> probably current swing must close over 100% zone (maybe over 120%??) -...
Generally what you posted is true but... there are parts of it based off my experience, are not practical.

Not trying to pour cold water over your efforts, but I felt you would need 'something else', as it appears you are at a very early stage of learning. What I am about to say is generally off topic : Most of the things in this thread will be eventually out of reach on a practical level for most, not because they don't have the programming skills to pull it off, but largely due to other reasons. e.g. I have known LG over the years, hes not a programmer by trade. But I am very impressed that he learned how to use Excel so much more better than I do and pull off all sorts of research.

As a start, even myself, I read this post over and over again : http://www.forexfactory.com/showthre...73#post5673773

Hope this helps you. It is a long walk, but worth it.
  • #53
  • Nov 15, 2015 6:41pm Nov 15, 2015 6:41pm
I have placed a poll to decide further additional progression of this thread.
  • #54
  • Nov 15, 2015 11:09pm Nov 15, 2015 11:09pm
Rela its fantastic approach, I have been stunned how I did not pay attention to this earlier. I have looked at the swings and moved ealier but never kind of got to which time frame and how many candles as otherwise the swings can never be used as a decision to trade. Please keep going.

I have another question, I was taking a look at some of the results you had posted and they had timeframes of only m1, m5, ,m15 mostly and sometimes largerTF based on how long you held the trades for. Did you actually use these M30x7 as the base there or is it applying the same approach to a lower TF or was it using a completely different approach.

Thank you

S
  • #55
  • Nov 16, 2015 12:42am Nov 16, 2015 12:42am
Quoting silver1surfe
Rela its fantastic approach, I have been stunned how I did not pay attention to this earlier. I have looked at the swings and moved ealier but never kind of got to which time frame and how many candles as otherwise the swings can never be used as a decision to trade. Please keep going.
Thanks Silver =)

Quoting silver1surfe
I have another question, I was taking a look at some of the results you had posted and they had timeframes of only m1, m5, ,m15 mostly and sometimes largerTF based on how long you held the trades for. Did you actually use these M30x7 as the base there or is it applying the same approach to a lower TF or was it using a completely different approach. Thank you S
I actually don't understand your question... the results I posted and they had timeframes of only m1, m5, ,m15? Sometimes largerTF based on how long I held the trades for?

Are you actually referring to the results in Trade Explorer, that I have in my Forex Factory profile?
  • #56
  • Nov 16, 2015 1:13am Nov 16, 2015 1:13am
Quoting Relativity
{quote} Thanks Silver =) {quote} I actually don't understand your question... the results I posted and they had timeframes of only m1, m5, ,m15? Sometimes largerTF based on how long I held the trades for? Are you actually referring to the results in Trade Explorer, that I have in my Forex Factory profile?
Yes Rel, I am referring to the Trade Explorer outputs you had posted. That is a bit new for me, but I see much smaller TF's in most cases. But that is super encouraging for everyone out there to understand that a micro account can still be traded will small deposits.

S
  • #57
  • Nov 16, 2015 1:36am Nov 16, 2015 1:36am
Quoting silver1surfe
{quote} Yes Rel, I am referring to the Trade Explorer outputs you had posted. That is a bit new for me, but I see much smaller TF's in most cases. But that is super encouraging for everyone out there to understand that a micro account can still be traded will small deposits. S
Ohh... I think to properly term what you really meant is 'per trade duration'.

Frankly, I can't remember where I read some advice (in FF forums, yes, here) from a pretty good trader. He said that if one is undercapitalized (most of us are), he/she needs to learn how to scalp in order to succeed in the long run. As it is the only way to get out and climb up the 'forex ladder'. I wholeheartedly agree.
  • #58
  • Edited Nov 17, 2015 8:21am Nov 16, 2015 9:02am | Edited Nov 17, 2015 8:21am
Poll to Further Advance this thread :

Poll is closed. Results for the poll;
Example of Research Process for a potential Additional Component?
-Yes = 9 Votes (blueseahorse, LiquidGenius, manish9300, Neio, pete7, Rparm, Wanderer272, woww, xux99) 90.00%
-No = 1 Vote (TiouK) 10.00%
-Not sure = 0 Votes 0%
-Need more time to digest the basics first = 1 Vote (Rparm) 10.00%
FYI 2 more 'late' Votes for Yes from Sciurus + silver1surfe


All right, lets begin.

------------------------------------------------------------------

Example of Research Process for a potential Additional Component

Background :
There are 3 types of news to my knowledge / understanding;
-Scheduled Economic Figures e.g NFP, GDP, CPI...etc
-Bank Interventions e.g. SNB, BOJ, MAS (Monetary Authority of Singapore), PBOC (People's Bank of China)
-Significant Events over Weekends / Off Market Hours e.g. Terrorist Attacks (RIP Victims of the recent Paris Attacks), Greece Crisis

We shall be looking at the third.

a-Through observation, these Significant Events over Weekends / Off Market Hours manifest into gaps in the chart. By definition, it is the difference between the final market close VS first market open.

b-Based off my memory years ago: If one is using Oanda, you can see the actual earliest market open for Wellington, New Zealand. I think majority of brokers, along with MT4 doesn't cover this, hence there are gaps created this way. Frankly, I don't have this data, so I don't know to what extend these gaps are plus how big they tend to be.

c-Because of #b, this means there are 2 types of gaps; those that created by actual Significant Events, and those that aren't but are created merely because of chart source / broker differences.

d-LG and I did some work regarding gaps. https://lgtrades.wordpress.com/2014/...e-weekend-gap/. Unfortuntely, I later realised these results are skewed :
d1-We used all weekend gaps we could find and failed to consider #c
d2-Lack of proper data, historical records of what happened over weekends. However, realistically speaking, can this be really done?
d3-There must be other acceptable ways to classify these gaps other than by pip size.
d4-The study used data from 1999 to present. Over the years, market volatility is different. 10 pips today is different from 10 pips in 1999. Perhaps its much better to accept a smaller sample size.

e-Despite of these issues (d1,2,3,4), it still warrants further investigation as some kind of edge can still be seen. e.g. Gaps bigger than 20 pips getting filled reasonably consistently.

f-The other problem is that the study does not specify the best possible entry and exit points. Although to be frank, one can just almost blindly trade the opposite direction of the gap, since the chance of the gap being filled looks high.

g-It is also possible to anticipate the gap by placing a position on Friday, in the direction of the anticipated gap on Monday and profiting from it. Unfortunately I had yet see any studies that investigates anticipated gaps before it happens, but only studies regarding how price responses after a gap happens.

h-MF-V has his own work regarding gaps. http://www.masterforex-v.su/002_011.htm. He also posed valid questions of his own.


Methodology :
1-Approximately 52 + 2 weeks * 2 worth of data will be used. That makes 2 years worth.

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

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
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
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 127 KB


6-Tabulate results for 4a,b,c,d. Again, do not assume anything yet
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot2.png Size: 70 KB
  • #59
  • Edited 8:02pm Nov 16, 2015 3:57pm | Edited 8:02pm
Quoting Relativity
Poll to Further Advance this thread : Poll is closed. Results for the poll; Example of Research Process for a potential Additional Component? -Yes = 9 Votes (blueseahorse, LiquidGenius, manish9300, Neio, pete7, Rparm, Wanderer272, woww, xux99) 90.00% -No = 1 Vote (TiouK) 10.00% -Not sure = 0 Votes 0% -Need more time to digest the basics first = 1 Vote (Rparm) 10.00%
Oops, apparently I missed the poll. I am still following the thread and I am interested to see it continue. 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.

Here is some stuff I have to contribute. I could go further, but I am limited in the time it takes to adjust code in between every test I'l like to make. Ultimately, liquidity moves the market, not statistics, but if enough people follow statistics, then liquidity is created .

I decided to so some "real" statistics on this data, but I believe I am just scratching the surface of what is possible. So far we have just been using descriptive statistics. Personally, I was curious about the relationship between previous leg length and percent retracement.

First, I extracted leg length and subsequent retrace data. Since I wanted to do parametric stats, I first needed to test for normality. Turns out that both leg length and percent retrace are log-normal (skewed to the high side).
Attached Image (click to enlarge)
Click to Enlarge Name: Screen Shot 2015-11-16 at 4.36.04 PM.jpeg Size: 147 KB

So I log transformed both variables and tested for normality again. Something interesting going on with leg length there, as its has a slight bimodal distribution going on. I will have to look at the later, but for now, I am satisfied that I have dealt with the skew. The bimodal stuff won't affect what I want to do next.
Attached Image (click to enlarge)
Click to Enlarge Name: Screen Shot 2015-11-16 at 4.38.48 PM.jpeg Size: 143 KB

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. That makes some sense. I will have to look at this in more detail sometime later. Right now I'm hungry and its dinner time. And for anyone wanting to interpret the following regression, you'll need to back transform the values on the axes in order for it to make any real world sense. Its Log base 10.
Attached Image (click to enlarge)
Click to Enlarge Name: Screen Shot 2015-11-16 at 4.40.32 PM.jpeg Size: 93 KB
  • #60
  • Nov 16, 2015 6:15pm Nov 16, 2015 6:15pm
So the average retracement over 677 samples was 1.9999. Back transformed, that is 10^1.9999 = 99.98 %. Wow, never underestimate to power of the market to be efficient, folks. the average leg length was 57.5 pips long.

Attached Image
0 traders viewing now