Auction Market Value Theory & Analytics

Hello Sinqua!
Glad to hear from you after so many years. I can't describe how happy I am to see your post, as it's clearly a sign of old members still pursuing AMVT and are interested in further development of ODC.
Although, I am very excited and I really want to ask much more from you, for now, I'll focus on answering your concerns about my theory.

In the scientific world, what you've done is called a "Peer-Review". Where the specialized members of that field, critique an idea, not for the intention of offending the discoverer/inventor, rather to get to the ultimate objective truth. Now this process has begin by you taking interest in my post and replying to it.

As my bias will inevitably, be attached to my own findings, however, I will try to keep myself as objective as possible. This is to warn you, that don't agree with me on everything and don't get offended by tone/answers, as I am not thinking about your personal feelings, rather we're all trying to understand/critique this "Chronological Distribution Theory". My intentions are not to hurt you/offend you in any way or form, rather, to get to the ultimate truth, which is, is my method valid or not, or can we modify this idea to either improve upon it/take inspiration from.

Now, that some things have been made clear, I will respond to your questions, line by line.

"First off, your examples are unclear, mainly because you mention [3rd April] which the 20,10,5,3,1 day ODC stats you see below in your chart do not reflect in any way (note the ODC start and end dates). You should note that the composite profiles are in no way tied to the candlestick prices you see on the chart, ideally they should be removed, leaving only the current price line."

You're stating that my dates do not align with the days typically used in the ODC framework, i.e. 20, 15, -- 1 day.
I see why you have been confused by this. Let's clear this up.
The date for 10 days is 03-31 till 04-11. If you were to manually count this, it would be 11 days, not 10 days. However, the extra day is added so that Fridays and Sundays can be calculated as 1 day. In other words, Friday's session is typically around 00:00 till 21:30 or 20:30 etc. It's not a full 24 hour trading session. Mainly because Friday's are half-days. To mitigate this "anomaly" Sunday's provide the rest of the session, hence, together (Fridays & Sundays) make a complete day. Hence, the reason why the dates/days don't add up to the typical ODC framework.

Your second statement: "You should note that the composite profiles are in no way tied to the candlestick prices you see on the chart, ideally they should be removed, leaving only the current price line."

I am very much aware of this fact. There had been a long battle with the members of the forex factory and Bandung the original creator of this indicator, among many others. For context, people wanted the profiles to align/represent the candle-stick prices. Which from a programming stand-point, was impossible. Mainly because the indicator was coded not in mt4 language, and only rented "space" on the chart.

HOWEVER! The Break-Out Prices align with the chart/candlestick prices. No matter what the breakout price is, it'll always align on the chart as well, why? Because the profile is plotted on the chart by using what? PRICES! Hence, the breakout price is the same, regardless of the "position" of the profile, because the position is done by via aligning prices. It will not mimic the candlestick prices "exactly", mainly because the timeframe is too small, but technically, the breakout prices are the same.
"ideally they should be removed, leaving only the current price line."
I do remove them personally, but for the sake of a visual example, I had to place the ODC profiles, for better understanding.


"Now AMVT ALREADY chronologically computes the ODCs, hence you see MZVegas screenshots showing day to day changes. We have, 1,3,5,10,15,20-day [time range] ODCs (the importance of each one is left to the user)."

Chronologically computing "ODC's" & "Distributions" are two very different things. To be very clear, I am not talking about ODC's themselves, rather, the distributions, which are not calculated chronologically.

"If you mean calculating distributions chronologically by the hour, then that is a whole new path separate from existing works which is by the day, and you should describe it a such. (I have such intentions myself). Here, new characteristic interpretations of the ODC would need to be defined. For instance, UL and LL Alerts will be acted upon differently."

Yes, I do mean calculating DISTRIBUTIONS by the HOUR. I'm not sure why this wasn't clear, as I had very long explanations regrading this. However, this is a better description.

"Your answer in the above statement from this post is right, but your interpretation is confusing, you see, if a distribution is forming outside a balanced market it is testing to trend, if it fails, the balanced market has undergone a "Range Extension" BUT it is STILL ONE DISTRIBUTION, i.e., the close price still fell back into the previous 3TPO range."

You're correct in a general sense, but the example clearly states there were two distributions, not one. Let me clarify once again.

The example states that the market closed above the bracket limits, but stayed there for 5 hours before returning back to its original distribution. By saying the market had spent 5 hours, I'm assuming the market has formed another distribution. WHY? Because we need 3 TPO's to start and end a distribution (it's a bit more complicated then that because, we will not ALWAYS find exactly 3 TPO's, so some adjustment has to be made) and 5 hours is more then enough time to have at least more than 3 TPO's. Hence, the reason we're assuming, in this example, that there are more then 2 distributions.

You are right, the market has indeed undergone a "Range Extension", however, that is not enough to justify that it's one distribution. Mainly because we find distributions by TPO's and TPO's are formed via TIME (5 HOURS in our example), not price or price extension etc. In other words, there was a failed-breakout, leaving us with 2 distributions.

"Now onto what you call "two distributions", you need to clarify what you call a distribution, as a bracketing market has only one distribution (bound by 3TPO's, no less than 3tpos is within the upper 3tpos and lower 3tpos), any other "distribution" is not bracketing but distributing. So you cannot have a "balanced market" with two distributions, it is inherently unbalanced. These are the fundamentals."

Firstly, let's clear up my definition of a distribution. For timeframes lower than 10, assuming we already have a market profile built, I will look for price points where the TPO threshold is greater or equal to 3. For instance, price A has 3 TPO's and Price E has 3 TPO's, hence, my distribution is starting from Price A till E. For timeframes higher than or equal to 10, I use 5 TPO's, all of which are standard practice in A.M.V.T.

Now, onto your second statement. You say, a balanced market has to have 1 distribution to be balanced. Which is a very standard view. However, my argument is for saying the EXACT OPPOSITE. I believe that markets having two distributions is not "inherently" a sign of imbalance. I have supported this argument via Steidlmayer's 4-Step Process.

I won't spend time repeating my same statements (for respect of your time), although, I will encourage to read the post again, this time, reading every line twice.
However, I will explain, this time in short.
I believe, Markets are Imbalanced if the activity is not present in the 1st distribution (Hence, the need to chronologically sort the distributions). It doesn't matter how many distributions are present. If price activity is contained within the first distribution, then I deem it a balanced market. Why? You have to read the post again to find answers. But, I assure you, everything is present.

"To count for what I assume you call two-distributions, the shape factor was introduced, but it does not negate the balance/un-balanced rule and only tries to quantitatively measure a balanced markets' similarity to an ideal gaussian distribution."

No, I haven't used shape factor for calculating distributions. It serves a very different purpose (to measure the quality of the distribution).

Some other things I wanted to point out from your post:
"On your discussions and theory of ODC, distributions and definitions of bracketing. I think you have over-complicated very simple rules. It is okay to come up with your own idea's but also important to separate them from the original ideas and rules."

I might've overcomplicated it, it's your own opinion. However, my effort was to base my argument via resources I could find that can hint at this potential.

Secondly, I have done the best effort to clearly state that this is my own idea, and I'm not "mixing" it with the original rules. Hence, the reason you see a bright red disclaimer, and other disclaimers I have poured throughout the post to clearly state to the reader, that this is something new and needs to be critiqued/tested before we are to rely on it.

Anyways, I hope that this post has made some things clear on my part, if not, I would strongly suggest a re-read of my post, as I can tell from your writing, you're missing very key points. Regardless of what you think about my idea or "theory", it's important to post and contribute to this forum as much as possible. I hope that I see your posts in the near future. I hope that whatever you may be developing will be created soon, it's just a matter of persistence. If not, I am here to help you in your ventures, if they fall under my expertise. Hope to see you soon!

PER BADUNG Indicator:

HOWEVER! The Break-Out Prices align with the chart/candlestick prices. No matter what the breakout price is, it'll always align on the chart as well, why? Because the profile is plotted on the chart by using what? PRICES! Hence, the breakout price is the same, regardless of the "position" of the profile, because the position is done by via aligning prices.

I have since built my own analysis due to this confusion if I remember correctly, the data aligns, but the X-time and Y-price axis visually do not, it is an optical illusion, I used to think they did too, change the MT4 TFs and see how/if the ODCs react, if they do then good for you and everyone.

I think you are mixing up AMVT and MP, it is well known that they have similarities but they're very different, your 5-hour example is not an AMVT concept but an MP concept, why do I say this? AMVT (ODC and its characteristics) calculates based on EOD (End of Day) data, so data after 5-hours of the next day is inconsequential until its EOD, then the ODCs are recalculated. Now onto distributions. Take TPO example below,:

3-DAY ODC
A x
B xx
C xxx
D xxxx --------> Close Price
E xxxxx
F xxxxxxx
G xxxx
H xxx
I xx
J xx
K xxx
L xxxx
M xxxxx
N xxxx
O xxxx
P xxx
Q xx
R x

By your definition there are TWO distributions here. This is not a balanced market, AMVT is not vague about it. CISCO was clear about it, some of Steidlmayers works were added to the CISCO AMVT but his works should not be misconstrued as the final AMVT as we have it on this thread by MzVega.

If Price I and J did not exist, it will be one distribution (balanced), TPO's forming by the hour are not valid for AMVT application until EOD, only in MP concepts are they valid by the hour where initial balance and the likes come into play. Also, PRICE defines a distribution not TIME, by definition there must be a minimum of 4 different price levels for it to be considered a distribution whether it happens in 1 min or 1 day is irrelevant.

Like I said it seems you wish to apply AMVT concepts by the hour, TPOs calculated by the hour as in Mind over Markets (is Market Profile Concept). AMVT is EOD, if I were to do AMVT by the hour, I would be doing a rolling 24,72,60 hour etc, to keep in accordance with AMVT.

Chronologically computing "ODC's" & "Distributions" are two very different things. To be very clear, I am not talking about ODC's themselves, rather, the distributions, which are not calculated chronologically

This is where you lost me and probably others. Distributions and ODCs are inter-changeable, TPO (unique price levels) prints by the hour/min or whatever timeframe are simply TPO prints, it becomes a distribution when it is collapsed (where you now have Multiple TPO print per price level) AN ODC IS A DISTRIBUTION with defined parameters. Hence why I had to define everything in my previous post. If you do not wish to apply the ODC parameters to the "distribution" then we are not talking AMVT.

Perhaps you need to post a visual of the differences between AMVT ODC and these distributions which are not calculated chronologically (how are/were they calculated??).

If a balanced market has two distributions as defined by AMVT (see my above example) - then by your implication it has two Fair Values, two Upper Limits, two lower Limits, Internal trends are two, TPOs A/B are two... you see how this veers off AMVT completely ???

However, I only make these comments for posterity so as not to confuse other new comers when going through the thread and to make sure you are sure of what you are doing. Else I would have been a silent observer. Since you mentioned your DISCLAIMER I will not discuss this further. Cheers.

Guys one question, did anyone use API KEY from some data feed and can suggest one that is good and kind of free. I was using IC Market data but if i want to go through my anaylsis faster i need some API from somewhere. Any suggestions?

You didn't have to quote reply since this question has nothing to do with the discussion quoted, it doesn't help when searching the thread.

Anyway that aside, I used tiingo for a while, they had the best free tier I could find. Cheers.

Thanks a lot, could you send me private message. I would like to discuss with you smth related to that, but dont want to spam this thread with silly questions. Thanks

I'm glad you weren't a silent observer and demanded questions, as that is how progress is made.

As you are very new to this concept of "chronologically sorting distributions", it's valid that you are confused about some fundamentals of it. I would also make an assumption that you aren't fully refreshed with AMVT as you clearly have left-over development to do, and also, are not trading the AMVT concepts (correct my assumption if wrong).

I was initially thinking of collecting and putting up examples, and spend a lot of time explaining this concept. HOWEVER, your questions indicate a lack of knowledge from my previous posts. Hence, I would not repeat my self, and waste yours and my time talking about the same thing. I will just clear your doubts. I will do this for the sole purpose of explaining it better to "you". Otherwise, the original post contains the relevant information, if anyone else including sinqua is interested in diving deeper.

Also, in the very near future, I will be posting back-test results. Explaining the analysis, the process and everything relevant to it. That back-test is also done via my method of sorting distributions, which is different from others/the traditional method. Hence, it'll not only be evident that this concept works and is fully in the boundaries of AMVT, but it'll cause some misunderstandings among the uneducated who aren't aware of AMVT, because I am doing things a bit differently. Hence, disclaimers will be made for the intention of not misinforming the members and to be fully authentic with the results. I'll explain briefly in my next post of my plan to back-test, and the results I have gotten so far. Let's get back to sinqua.

"I have since built my own analysis due to this confusion if I remember correctly, the data aligns, but the X-time and Y-price axis visually do not, it is an optical illusion, I used to think they did too, change the MT4 TFs and see how/if the ODCs react, if they do then good for you and everyone."

I am not aware of this "illusion" as nobody mentioned it, or perhaps I have mis-read or even missed some crucial posts. If that is so, it's not a problem I have to deal with, because like you, I have built my own ODC indicator within excel. To be clear, it was used only for visual representation of chronologically sorting distributions, i.e. in the example, you can see 2 distributions on the 10 day profile (it's 12 days on the chart and this confusion as been since cleared upin my previous reply to sinqua) I was claiming the 1st distribution to be a sign of balance, because the market is trading within the bounds of the 1st distribution.

"I think you are mixing up AMVT and MP, it is well known that they have similarities but they're very different, your 5-hour example is not an AMVT concept but an MP concept, why do I say this? AMVT (ODC and its characteristics) calculates based on EOD (End of Day) data, so data after 5-hours of the next day is inconsequential until its EOD, then the ODCs are recalculated."

I'm very puzzled as to how to respond to this comment. Because it's a valid statement, that reference points are calculated based on EOD, but it's a complete mis-interpretation of the example. The example is admittedly not concrete, but it serves the purpose of laying the foundations for this concept and that was the intention for it. The example was made to introduce the concept of chronological distribution sorting, it was not intended to assume that the reader has incomplete knowledge of AMVT concepts itself! In other words, yes you are very right that data after 5 hours of the next day is inconsequential un till its EOD, but the 5 hours is a random amount of time I choose, because it's sufficient enough to start a new distribution. It's a real-time example, but that's the intention. It's real-time so the reader can have an idea of what it means to code and understand chronological distributions. If we are to critique the example very literally, like you are, there are other things wrong with it as well , for instance, why are we assuming 5 hours? The markets don't normally spend 5 hours at one place. You see how that can cause issues, when taking examples literally. It's an attempt to clarify the concept. That's why it's an example. Otherwise, I have given an actual example, taken from the market, which further concretes the understanding, but nobody seems to be asking questions about the actual "evidence" of it.

"By your definition there are TWO distributions here. This is not a balanced market, AMVT is not vague about it. CISCO was clear about it, some of Steidlmayers works were added to the CISCO AMVT but his works should not be misconstrued as the final AMVT as we have it on this thread by MzVega."

Tell me this. If the ODC is skewed to the bottom, or the right, is that a balanced market? Even though it has one distribution.
What even is a truly balanced market? Is it a gaussian-distribution? YES! So, let's say you are to find a gaussian distribution, however, the market has 2 distribution within that gaussian distribution, now what? Is it balanced? Even though, technically it's a gaussian-distribution. Some might argue that you can't get 2 distributions in a gaussian-distribution because its statistically "recognized" as a balanced regime. And yes, that is correct. However, the spirit of the concept is not related to how many distributions there are!! It's related to the time of their occurrence, in other words, their (the distributions) chronological order.

In other words, nobody really knows exactly WHAT quantifies as a balanced market. The market can be balanced in 5 day timeframe, but unbalanced in all other timeframes. Does that mean we should trade the market? Does that mean the market is now balanced? No! That just means the market is taking a breather. However, to quantify a market, to even say a market is balanced is wrong. We can classify these assumptions via market units, which helps in understanding these different regimes better. If I were to go deeper into this discussion it's going to be a long post, but I hope you get the "idea" not the literal meaning, because we're dealing with a concept.

"If Price I and J did not exist, it will be one distribution (balanced), TPO's forming by the hour are not valid for AMVT application until EOD, only in MP concepts are they valid by the hour where initial balance and the likes come into play. Also, PRICE defines a distribution not TIME, by definition there must be a minimum of 4 different price levels for it to be considered a distribution whether it happens in 1 min or 1 day is irrelevant."

I find it funny, when you say TPO's forming by the hour are not valid, because TPO's themselves are a construct of time, but I get what you're saying. The idea is new and fresh, hence, there are a lot of assumptions being made either about my personal knowledge of AMVT or the concept itself. However, you said you had a similar idea sorting distributions this way, hence, are you also not applying AMVT if you were to construct that? Your statement: "If you mean calculating distributions chronologically by the hour, then that is a whole new path separate from existing works which is by the day, and you should describe it a such. (I have such intentions myself)."

I'll answer it. No, it's still a concept of AMVT. But it's tuned. It's different from the norm. However, it's ingrained within AMVT, why? Because we're still classifying different market conditions, using the same analysis and logic, but with a unique approach. That's it.

Also, MP concepts like IB are used in AMVT, I don't know why you are mistaken about that. They're not useable for a person trading forex, but they still apply to AMVT.

Price defining distributions, is a rather poor way to counter my argument (sorry for my tone). It's TPO's, not prices.

"Like I said it seems you wish to apply AMVT concepts by the hour, TPOs calculated by the hour as in Mind over Markets (is Market Profile Concept). AMVT is EOD, if I were to do AMVT by the hour, I would be doing a rolling 24,72,60 hour etc, to keep in accordance with AMVT."

How do you imagine I trade AMVT? I also do EOD analysis, the reference points be it from an ODC or Market Profile metrics, all are calculated EOD. You're missing the point of distributions sorting, and reverted to base your arguments based off of a misinterpretation of the example.

"This is where you lost me and probably others. Distributions and ODCs are inter-changeable, TPO (unique price levels) prints by the hour/min or whatever timeframe are simply TPO prints, it becomes a distribution when it is collapsed (where you now have Multiple TPO print per price level) AN ODC IS A DISTRIBUTION with defined parameters. Hence why I had to define everything in my previous post. If you do not wish to apply the ODC parameters to the "distribution" then we are not talking AMVT."

An ODC is a composite profile that consists of different profiles aggregated together, hence a composite. The distribution is part of that, including reference points like UL, Octants etc.
For context, my response to you for this was clear, but you still misunderstood it. Let's go back to it.

I replied to your statement, when you said the following: "Now AMVT ALREADY chronologically computes the ODCs, hence you see MZVegas screenshots showing day to day changes. We have, 1,3,5,10,15,20-day [time range] ODCs (the importance of each one is left to the user)."
I cleared up the misconception you had about ODC's already chronologically computing ODC's and chronologically sorting distributions. I also chronologically compute the ODC's. From one day, to the very next and so on. However, I cleared this doubt. Via this following statement:
"Chronologically computing "ODC's" & "Distributions" are two very different things. To be very clear, I am not talking about ODC's themselves, rather, the distributions, which are not calculated chronologically"

You were talking about a different type of chronological sorting, that we already do. A rolling window in other words. However, distributions are different from ODC, in that, they are the tool for discovering market condition. If it wasn't clear then, then I hope my "definitions" will provide some guidance. However, I don't really need to spend time defining the already known aspects of AMVT with educated members of this forum, as it feels like I am disrespecting them. But ye ask and ye shall receive !

"If a balanced market has two distributions as defined by AMVT (see my above example) - then by your implication it has two Fair Values, two Upper Limits, two lower Limits, Internal trends are two, TPOs A/B are two... you see how this veers off AMVT completely ???"

No, it doesn't veer off AMVT, it's an extension of AMVT itself. The concept of multi-valued market is clear Don Jones early works, mainly his book. That is where I got this idea. Indeed it has 2 fair values and 2 upper limits. Here's what I said in my original post: "What is true, is that the market is now multi-valued. That second distribution, now serves, in this case, a rejection. However, both distributions, are in a sense breakout zones. If that doesn't make sense, don't worry, it's an extension of this topic."

This extension of the topic has not been covered with the same amount of depth. Mainly because the idea needs some digestion, before I can talk about it more. Just for you sake, I don't use the second distribution's reference points, because it's an area of imbalance. Why? Because if we are to chronologically sort the distributions, then, only the first distribution is where balance exists. Hence, second distribution, or any other apart from the first one, is a breakout zone.

I will go deeper into the implications of this, for now, it's best that I can answer as many questions regrading the fundamentals as possible. I will encourage to at-least check the resources I have mentioned in my original post, if you are not convinced enough to read my post again. That will also help clear up some doubts or create more doubts, which I am happy to answer.

BACK-TESTING & OPTIMZING PLANS

THE READER SHOULD BE AWARE THAT THE BACK-TESTING IS DONE USING MY OWN METHOD OF CLASSIFYING MARKET CONDITIONS. In other words, I AM NOT USING THE ODC THAT OTHER MEMEBERS ON THIS FORUM ARE USING. WHY IS THIS DISCLAIMER IMPORTANT, YOU MAY ASK? IT'S IMPORTANT TO SAY THIS, BECAUSE YOU CAN'T REPLICATE MY RESULTS USING THE TRADITIONAL APPROACH TOWARDS CONSTRUCTING AN ODC, YOU WILL NEED TO BUILD THE ODC IN THE WAY I HAVE DESCRIBED, WHICH IS SOMETHING THAT'S MAJORITY OF THE MEMEBERS AREN'T WILLING TO DO. HENCE, DON'T BE FOOLED THAT THESE RESULTS WILL BE THE SAME FOR YOU ASWELL, BECAUSE IT WON'T BE.

Okay, new that the disclaimer has been made, I would like to share my plans and the progress I have made as of yet to the back-testing.

General Guidelines:

1. The trades being executed in the back-test includes NO SUBJECTIVE BIAS. In other words, NO ANALYSIS HAS BEEN DONE/ADDED TO THE EXECUTION OF TRADES. THE TRADES ARE DONE PURELY ON THE REFERENCE POINTS OF UPPER LIMIT AND LOWER LIMIT.

2. The criteria for trading is the following: The market needs to be bracketing in 5 days TF, when bracketing, a pending order is placed, on both the Upper and Lower Limit. If a market hits any limit, a pending order is simulated, and, is tracked on the basis of following RISK MANAGEMENT Assumptions: A Fixed Stop-Loss is used if $2.5. A Fixed Profit of $5. The Pip Value is 10 cents (meaning every point is worth 10 cents), and we're trading a micro lot of 0.01.
ALL OF THESE RULES ARE APPLIED TO EVERY TRADE.

3. If any target is reached (SP or TP), the trade is closed within the day. However, if no target is reached for the that day, then the closing price is used as the exit price, and the next day is monitored for potential losses or profits.

4. MAX TRADES IN A DAY = 2

5. Timeframe Used = 30-Minutes

6. STRATEGY = BREAKOUT FROM 5 DAY LIMIT.

Before I would discuss the plans I have for how to achieve this back-test as efficiently as possible, I would like to share the progress I have made as of yet:

Attached Image (click to enlarge)

This looks complicated but is very easy to understand. Many of the metrics I am tracking are already well understood, i.e. entry time, exit time etc. However, I would like to define some: T in T (Bars) means, Time in Trade by counting the Bars of 30 minute timeframe. MFE means, Max Favorable Excursion, in other words, what was the highest amount of profit. MAE means, Max Adverse Excursion, which is the complete opposite, what was the highest amount of loss. There are other metrics not visible in this screenshot, but it at least gives you the general gist of my approach.

If you are to look at the bottom left corner, you'll find some stats for the P/L.

Total Bracketing Count= 78
Total Trades Executed = 39
Average = $0.87
Sum = $33.99
Win Rate = Roughly 60%

What's even more fascinating about these stats is that this result is based on PURE MECHANICAL EXECUTION. Meaning, there's a potential of win rate to be way above 60%, if we were to introduce flow analysis into the mix.

Also, the results seems to be better when using different risk management technique, one which I have developed personally has the following results:

Attached Image (click to enlarge)

This one, uses VTY SL's but a different type of Profit Target. The results seems to be better by simply changing the risk management on the same dataset:

AVG = $1.25
SUM = $48.63

Anyways, these numbers are "some" evidence that the ODC approach I seem to applying is not invalid, rather, it's an improvement, using the same AMT principles.

Please bear in mind, these profits and losses are a result of mechanical trades. No analysis was done. It was all based on the Upper and Lower Limits of 5 Day TF Brackets. It does show potential of Algo-Trading, but that is something way beyond my expertise, and a topic of future that regards a lot of data, not just 39 trades.

So, this is the progress as of current period. I am actively working on adding other risk management techniques and back-testing the results for 2.5 years.
Here comes the part, where I can explain the plans ahead:

I plan to test 2.5 years of data, which will be used as a training data-set. I will apply the flow analysis, RI-QC, and other calculations to see patterns in the data, or to derive any useful information that can used either for Algo's or Discretionary Trading. The end goal is to test the robustness of this method.

The training data will include periods of 2024, 2023 and 2019. The testing will be done on periods 2020-2022.
The testing periods are very specific, mainly because it allows us to stress-test the models during very volatile and unpredictable periods of the market, mainly covid, the Ukraine war, elections, all of which are highly volatile periods. This tests the robustness of our system, and allows for optimization during such periods.

The goal is also to avoid Overfitting, which is a common issue when back-testing. Hence, the reason we're separating data sets by classifying them as "Back-Test Periods", which are used for pure mechanical trades, and afterwards, refining and optimizing them. Then comes "Test Periods", which are simply periods where we then test the results we get from Back-Test periods, and compare them together to judge whether an edge exists or not. If so, how much? ETC ETC.

There's so much to do, and many exciting projects awaiting in the very near future. I hope you all stick around, and be able to witness the progress, also, sometimes helping me out in the process if needed. Your opinions and help is always appreciated . See you soon!

Just to refresh our memory of the past

I wonder if our use of our old excel calculation is of concern. Hopefully AMVT new findings will lead us into a new exploration. I can't wait to see the new venture. I am sure RI and Quadrant, including the many reference points, are to be considered for breakouts. This chart was taken from Friday May 16, 2025

Just to refresh our memories of the past

Now today is Monday, May 19, 2025. We have seen false breakouts down on Sunday evening. But most reference points bias are long and upward, especially if we see the 5- and 3-day alerts. Let's see what it looks like the whole day Monday. Now I can see AMVT's concerns about Friday-to-Sunday timing, as it seems not a full 24 hours, as he said. Am I right, AMV?

Hello JP.

Delighted to see you post again!

Regarding your question, YES, You're Correct the Friday's Session is not a 24 hour session, hence Sunday is added to mitigate this.

I don't know if you're using the indicator's export for this, or you have a personal built indicator. Either way, you have to include Fridays with Sundays otherwise, you're not trading a 3 or 5 day profile, instead it'll be 2 or 4 days (Even graphically, the ODC is showing 2 or 4 colored Profiles, not 3 and 5). I built mine considering this issue, hence, I don't have a problem with it. If you also have a personal indicator, then it would be wise to tune/modify this setting, or manually change the number of days in the MT4 Indicator.

However, you should know, the indicator does not export correct market profile data. At-least not for me. Hence, it's even wiser to leave it alone, if the calcualtions you do by hand don't meet the indicator's exported data.

"But most reference points bias are long and upward, especially if we see the 5- and 3-day alerts"

Judging by the screenshot of analysis you have provided, it doesn't "look" as if the data is pointing towards a long bias, rather the opposite. I don't know if it's my personal misinterpretation of the data, but from what I can see, it doesn't look like a buy situation (mainly judging by the Flow Nets)

False Breakouts are something that's inevitable. Although, they can't be avoided, there frequency can be lessened, if you are do the analysis a bit more. By bit more, I mean judging by the activity reference points, or including other data calculations not done traditionally in AMVT. That is up to you. I have a method for this, but I can't say now. When the full back-test report is uploaded, I'll provide the necessary information.

Anyways, great to see your post. Keep us updated. Also, so far, what do you think about the back-test results and the "Chronological Sorting Distributions" Technique? Love to know your thoughts on it. Happy Trading

Hi Rokica, Did you get my PM in your inbox?

AMVT,
1. Thank you for your remarks; I really appreciate that. Your backtest test is remarkable with a win rate of 60%. I can imagine what it would be if trading manually with all consideration of all variables, reference points, and fundamentals were applied. AMVT doesn't consider fundamentals, as we are here dealing purely with mathematics. You put in a lot of effort as you are creating your own mechanical data imports and organizing your analysis.

2. As for me, I am just depending on the MP MultiComposite Profiles we were using in this forum. I put my workbook in Metaquote file folder and link all csv files into it so I don't have to worry about any mechanical imports except using my macro to update and sort it. I appreciate the many forum members who freely gave us the necessary tools during our discussions. The unforgettable Iron Lady, Claudia (mzvega) was here day in and day out, giving us all the information and instructing us tirelessly for a long time. She was ready to type long paragraphs and put together chapters. She was upfront to challenge us when we needed spoon-feeding and shortcuts. I am thankful for her, and I am still learning in this vast and wide array of subject of AMVT. May God bless her wherever she is.

3. My color code is of a preference; otherwise, I am applying 1, 3, 5, 10, 15, and 20 days in my chart. You said, "However, you should know, the indicator does not export correct market profile data." It is scary, but I am willing to learn new findings. I hope this vibrant forum and subject will revive and go forward so that we may tackle this worldwide manipulated market and do exploits.

4. Today is Tueday May 20, 2025. USDJPY is just an example, food for thought, and to refresh our old memory of this forum. Today's market could be short from daily UL as the bias is downward. And also buy back from the bottom of the breakouts, as it is all bracketing except the 10-day, and seems like false breakouts. It is a 135-pip range and a lot of opportunity for responsive trading as price rotates up and down.

I am hoping that the back-test leads to new findings and uses for the flow analysis/RI & QC methodology. What I think will happen is, I will find patterns within the activity or direction variables of the flow analysis, or patterns among the market condition. Either way, be it patterns, anomalies etc, there'll be useful information when considering other variables of the market. That part is a bit tricky in its practical application, however, I am committed towards delivering my promise.

In the end, I want to be able to give empirical evidence about my findings and motivate others to use AMVT.

There's no need to be fearful if the indicator is not working. I built mine within 1 month of heavy development. With the tools and amount of information available today, anybody can start and end a venture in a short amount of time. But, that's not to say it's easy. I had to learn a lot along the way and it was a pretty frustrating time. However, when market's provide the "evidence" that the AMVT methodology works, then it's a no brainer to invest time and effort into building the tools associated with AMVT as well.

Responsive Trades are the topic of my new back-test, after fully completing the 5-Day Breakout Strategy. Not only 3-Day Responsive Strategy, but also 3-Day Breakouts. Responsive trades are a bit tricky, in that, we can't completely do responsive trades mechanically. Flow Analysis Direction Variables are needed to make the decision. Which makes it tricky. For instance, what timeframe's Flow would provide the most optimized results? If we were to test all Flow Timeframes, then it would take a long time. Hence, I would like your thoughts on it. Should I perform the back-test for responsive trades not considering Flow Variables at all? And afterwards optimizing the results, which is less time-consuming and a better option. Or do you think there's a need for direction variables? If so, which timeframe?
Your suggestions will be crucial, as I don't have experience with Responsive Trades, I only do Breakout Trading. However, I would like to dive deeper into responsive strategy.

I don't follow the USD/JPY Pair, but from the analysis, it seems as if the market is taking a breather or testing for trend end. All timeframes coming in to balance and repeated Alerts signify, the market is testing for trend end. The market after 12th may, has went down quite a bit. If the activity and direction variables alongside VTY remains stable, it's a good chance there'll be a breakout to the downside, assuming there'll be a push for this (A major upcoming news event etc). Otherwise, it'll stay at the lows for sometime, balancing, and breaking out in the opposite direction. It's tough to say for me, as I don't follow this pair, but from the available data, it seems the market is slowing down and bias for now is short sided. And yes, this is a market where responsive trades will be working best, as the market is currently transitioning from trend to bracket.

Why not? You did break out trades mechanically right? I've revisited the thread and noticed a whole gang of activity here. Pretty cool.

I record Flow but don't use it in my trading yet. I'm planning to redo the Flow variables as intended, using RTH only.

Glad to have you back

The reason I'm being careful with responsive trades, is because 1. I don't have much experience with it and 2. Responsive Trades use direction variables to guide the decision of buy/sell. That's not to say that, breakout strategy does not use other flow variables, it does aswell. However, with breakouts, the direction is not as much important as other variables. Hence, you see MzVega and Evak going sometimes against the flow direction. However. responsive trades mean, the market is going up, but you have to go the opposite way. There's no way of knowing if the opposite way is the best possible route/bias to take, unless, we are to take into consideration direction of the flow.

I have to be considerate before I take on a time consuming task. Otherwise, a small decision can lead to the whole project/back-test as worthless. Hence, allow me to be a bit careful, as I won't like to waste the hard-work I'll put into this.

Anyways, welcome back. Hopefully you can share your own findings and contribute to the forum like before. It's very refreshing to see old members still taking interest in AMVT, and I hope we can continue this thread with further refinement and contribution to the AMVT Methodology. KEEP POSTING !

For responsive trades normally you would use the 10, 15, or 20 day ODCs as a base. 3 day and 5 day ODCs are going to be too narrow in range. You'll get picked off by the volatility. Look for large TFF, declining volume, declining VTY, etc., at the edge of the bracket. (you need to define the brackets properly though)

Humans can make better trade decisions than an algorithm. I believe that 100%. Regardless, I've been backtesting for at least a few months now on my own model because I am interested in true odds. I took years of screenshots on multiple commodities to form a beginning of my understanding of how Trade Facilitation, Volume, Volatility, etc., all interplay. Then I had some of these interactions hardcoded. After backtesting, I can tell you with full certainty that the proof of concept is there. It is possible to create a winning formula.

My trade for today
Eur/Usd also has a buy signal but I'm not trading it

Thank you for the suggestion. This helps me to understand, not only 3-Day Responsive Strategy is of worthy back-testing, rather, bigger timeframes as well. Your argument is valid, as lower timeframes generally have narrow range, hence, there's too much risk. However, it's worth testing anyway, to see the difference. However, you have convinced me to look at this in another light. By testing different timeframes. Very Interesting indeed.

However, the question still stands, mechanical or inclusion of flow variables. I believe mechanical is the most efficient way to proceed (from what I have gathered), as testing afterwards with different flow variables will allow for better optimization. Seems like I have to take the same route as I did with my previous back-test.

Also, your mention on algorithms is fascinating and motivating. Although, my intentions do not involve a fully automatic system, however, I'm also not opposed to its possibility. With you mentioning a winning formula, it leads me to believe if I can execute this right, then I am able to create that. Whatever the end goal may be, the amount of information which will provide assistance to discretionary trading, is immense and worthful. Hence, either way, it's a win-win. Hoping for the best

AMVT,
"Note! Markets are time-frame related. You can have a four day trend inside a ten day balance. You can have a ten day breakout while the twenty day Overlay remains in balance. The Balances database has nominal five, ten, fifteen and twenty day balances. You can play the Initiative/Responsive Game on short, intermediate or long timeframes." Mezvega

I am not sure of your intention, whether you are persuing into automation or manual trading. You have to incorporate into your system on holistic maner. There are four types of trading actors as we had discussed in this forum many times. Mevega had difficulty to convice people in addressing the issue of automation and manual trading. There are many variables into play, and Mezvega had chosen the second one and her aruguements were convincing. The market is not monlotic, "many currency pairs affects the one and the one affects the many." Beside that our analysis shoud be based on 1,3,5,10,15 and 20 days aproach. Long term time frame affects the short term and vice versa the short affects the long. I wish you the best. JP

Thanks JP for pointing this out,

That Cisco Article is indeed a goldmine of information when it comes to testing for my system. To be clear, I understand the limitations of an Algo System on a Discretionary Method (AMVT). I'm not denying the fact. However, her argument was mainly based on the limitations MT4/MT5 Presents, as they are linear environments. Hence, the lengths we have to go, to create a successful automated system, does not only require its own separate platform, rather an immense amount of work, that is impossible to do by one person.

Though, My intentions are clear, I want to find out what causes loosing trades, and what causes profitable trades. Simple. If I happen to come across any patterns among the timeframe's variables and their associated interplay, I will share that information. The intention is not focused on an automated system, rather to gain any useful information that can guide the decision making process, be it for an Algo or Discretionary Trader.

And yes, you are correct. A Holistic Manner is needed for finding such patterns. The main plan is to see how each timeframe's variable "communicate" with each other. For instance, if there was a breakout on the 5th of May, what was happening in the Volume of 1D? As Compared to 3D? Each component of market profile will be tested/observed independent of one another. After we've done that, only then, can we have a holistic approach.

Your comments are much appreciated, and I plan to incorporate this approach when optimizing the back-test results.