Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Auction Market Theory and Market Profile 25,556 replies

MarketProfile/Volume Profile/Auction Market Theory indi? 17 replies

ECN && STP && Scalping && Hedging - Broker Review - Oct 2009 24 replies

Auction Market Value Analytics Charts & Data 14 replies

Analytics, recommendations - S&P 500, Euro and other 11 replies

Attachments: Auction Market Value Theory & Analytics
Exit Attachments

Auction Market Value Theory & Analytics

2) The Tick-TPO or "Meta-Profile"

Requiring audited (cleared) data for "Market Profile" methodology excluded most of the auction market world. This problem was solved in 1987 when the "Tick-TPO Profile" was announced. Tick-TPO Profile, later called "Meta-Profile", is a methodology that generates the profile graphic and value area from tick data, either within a day or at end-of-day.

The Meta-Profile Market Graphic
Meta-Profiles generated by ticks create both a profile graphic and a value area. Profile graphics from the LDB audit data and from ticks are usually quite similar and often give virtually identical profile charts. Value area, too, is quite close between the two methods, except in directional markets. Unlike the "Market Profile", Meta-Profile construction and analysis does not rely in any way on the tick data having a gaussian distribution. Other features of "Market Profile" analysis, such as day types and reliance on the gaussian distribution of the profile graphic are not used in Meta-Profile based analyses. Meta-Profile value comes from TPO counts, with no reference to the shape of the Market Profile type graphic. TPO counts play the same role for Meta-Profile value that volume does for the "Market Profile" value area.
Markets are not efficient, rather they are effective - Jones
A Tick - TPO Profile With Half-Hour Timeframes

A Meta-Profile representation of the data for the Dow Index of figure 1 is in figure 3. Price bar is on the left, then the TPO count (#), next are the Profile TPOs, followed by all the half-hour vertical bars, stated in TPOs. Value at any point in time (indicated by the TPO half-hour bar columns dashed lines) is the middle seventy percent of the TPO counts up to that time. The vertical bars on the TPO columns, the value area to that point, change as value changes throughout the day.

For instance, at the end of 'A' period there is one TPO at 105700, two at 105650, two at 105600, three at 105550, three at 105500, two at 105450 and one at 105400. The center is at 105550, indicated by the arrow '>', and the value area is 105600 to 105450 (the vertical bars).

At end of day, the value area lies between 106100 and 105530. The center of trading for the day is at 105750, with 9 TPOs (BDEGHIJKL). By comparison, the Liquidity Data Bank of figure 1 reports a value area of 105980 to 105330, with the peak volume of 2076 at 105550. There is good agreement between the two, considering that two different "data sets" are used and there is rounding from the large price compression.
Attached Image (click to enlarge)
Click to Enlarge Name: meta-profile vs MP.jpg Size: 337 KB
Markets are not efficient, rather they are effective - Jones
Meta-Profile Value
Meta-Profile recognizes that the profiles are only approximately bell shaped, (clustering is a better description), but in balanced, non-directional markets trading does indeed cluster about a 'center of value'. The 70 percent rule used by Steidlmayer applied to the Tick-TPO's finds essentially the same value area as does the cleared volume, end-of-day "Market Profile" value area (70% VOLUME Summary).

Examples of five Meta-Profiles in figure 4 illustrates profile structure in a balanced market. It is clear that the five days of trading are bounded by 110800 and 108200. This is a 'balanced' market. Even in this well-behaved environment the profiles are at best only quasi-bell shaped. Two of the days, October 31 and November 6 have double distributions. It does not appear that there is any analytical value is associating the gaussian distribution with profiles. Furthermore, it is clear that value varies from day to day.

Meta-Profile value is based on the profile (clumping) structure. For October 31 there are actually two values, one centered at 109200 for early in the day and a later one at 110200. This would be called a double distribution up day in "Market Profile parlance" and could presage directional movement. The longer term view of profiles, does not support the day types of "Market Profile" theory, as illustrated in this example.
Attached Image
Markets are not efficient, rather they are effective - Jones
Meta-Profile value

The agreement between "Market Profile" and Meta-Profile value in balanced markets validates Meta-Profile methodology.

Releasing the constraint of audited data is the basis for profile analysis of equities, options, forex, other futures exchanges and tick producing markets in general.

Meta-Profile also permits the value analysis to be performed intra-day. An aspect emphasized by "Market Profile", the structuring of a day's analysis around tails (excess) and range extension, as in figure 2,is little used in Meta-Profile analysis.

The reason for this is that in the Meta-Profile view markets need not be gaussian and so the appearance of a quasi-bell shape early in a trading day does not lead to the conclusion that the quasi-bell's peak will continue to be the market's focal point the rest of the day. This is obviously true in double distribution days (examples: Oct 31 and Nov 6 of figure 4). It is true as well for Nov 2 and 3 in figure 4, where the 'yzA' beginning is nowhere near the ultimate center of trading for the day.
Attached Image (click to enlarge)
Click to Enlarge Name: 2 & 4.JPG Size: 150 KB
Markets are not efficient, rather they are effective - Jones
Meta-Profile value

Meta-Profile focuses on value. A consequence of the data base (live ticks) is that value can be examined as it develops throughout the day, as in figure 3. There, by the end of period D (9:30 to 10) the excursion of period C (9 to 9:30) can be seen to have little effect on value (the vertical dashed lines). Although Nov 28 showed a good bit of volatility, the center of value as indicated by the arrow, '>', symbol for each half-hour, moved very little. By the end of the day the clustering is evident in the 105750 region. This day is unexceptional except for the C period run-up and retreat. The clustering is evident, a gaussian distribution is not.

Meta-Profile is the second major auction market discovery. It releases profile analysis from the dependence on a gaussian distribution function, and on audited data. It extends the reach of value analysis to most auction markets. The flexibility of Meta-Profile makes it a useful tool in further studies.
Attached Image (click to enlarge)
Click to Enlarge Name: figure 3..jpg Size: 148 KB
Markets are not efficient, rather they are effective - Jones
3) The Overlay Demand Curve (Composite/Merged profile)

Multi-Day Value in Auction Markets, the Overlay Demand Curve Graphic

From the inception of "Market Profile" the emphasis has been on intra-day trading. Steidlmayer was a floor trader (local) who developed the "Market Profile" for a time horizon mostly limited to the current day, based on value location from the previous trading day. While profiles reliably find day value in balanced markets, even in such balanced markets day-to-day profile value may fluctuate substantially.

The Overlay Demand Curve was developed to extend the one-day range of the Meta-Profile to longer time-frames. An Overlay is a linear aggregation of profiles. Whereas the profile finds one day value (value area), the Overlay locates value for the time period desired (days or parts of days). Overlay's are especially valuable in identifying balanced markets. They show upper and lower limits, estimate risk and provide a distribution that can be analyzed (the shape can show "how demand" comes into the market).
Markets are not efficient, rather they are effective - Jones
3) The Overlay Demand Curve (Composite/Merged profile)

Markets are known to be not serially correlated on a day-to-day basis. Simply comparing one day's profile to the next offers no firm ground for analysis if the market condition has changed. The goal is to find market condition, where the market is in it's evolution from balance to trend. Markets go through a four step process, which can start with balance and end back at balance. A balance will move into the trend phase over a period of time, a part of a day or more. A trend's end and the beginning of a new balance also traverses a transition period governed solely by the behavior of market participants. An orderly four step progression often occurs, but it is not a path the market always takes. There may be a break out from a balance that is quickly followed by a return to balance (false breakout). Or a trend may pause and retrace before continuing on. Any of the four steps may happen in a time as short as minutes or as long as days.
Markets are not efficient, rather they are effective - Jones
3) The Overlay Demand Curve (Composite/Merged profile)

A Balanced Overlay Demand Curve
A way to examine time frames longer than a day is to linearly combine the requisite number of Meta-Profiles, forming an Overlay Demand Curve . The composite telescopes temporal differences into a master, multi-day time frame. As an example, the five days of Meta-Profiles in figure 4 are combined into the Overlay Demand Curve of figure 6.

Recognizing Market Condition in a Balance
Figure 4 is a balanced market. It is observed to be bounded over the five days shown, but the proof is in figure 6, where the physical combination of the five days, the Overlay, forms a single distribution. Also, the mixing of the days (1 through 5) at various prices is a result of price rotating up and down over the five day period, a characteristic of balanced markets.
Attached Image (click to enlarge)
Click to Enlarge Name: overlay1.JPG Size: 151 KB
Markets are not efficient, rather they are effective - Jones
3) The Overlay Demand Curve (Composite/Merged profile)

Overlay Demand Curve Value in a Balance
Value in "Meta-Profiles" is the middle seventy percent of the cluster. For Overlays the a similar formula is not applied because the profiles are not gaussian curves and volatility spreads them even more. Summing several profiles emphasizes the lack of a gaussian process, instead producing a more squared distribution. For the five day Overlay of figure 6, value is defined from the three TPO level, 110600 to 108300.

Balanced activity is distributed in a single quasi-bell shape, much like the Meta-Profiles from which it is made. The single distribution and price rotation identifies the condition as balance. Center of the distribution is around 109600. Balance limits are obviously near 110800 and 108200; the rules developed for Overlays require a minimum of three TPOs to identify a distribution. For this case the upper and lower limits are 110600 and 108300. The limits are called resistance and support by traders who seek either breakout points for trend starts or directional indicators for trading within the balance (responsive trading). However, just how traders use the Overlay data is not the point; the fact that they have market generated information about the state of the market is.
Attached Image (click to enlarge)
Click to Enlarge Name: overlay1.JPG Size: 151 KB
Markets are not efficient, rather they are effective - Jones
3) The Overlay Demand Curve (Composite/Merged profile)

Recognizing Market Condition in a Directional (Trending) Market

The five days of figure 5 constitute a directional phase. Again, the imbalance is clear in the Overlay of figure 7, which consists of several distributions. Unlike the balanced figure 5, the prices in the trading days of figure 7 are segregated by time; earlier times are at higher prices, the latest day forms a distribution all it's own at the lowest prices. The activity is distributed in four quasi-bell shapes. There is no central focus for this distribution, it is made of days with quite disparate value centers. There is no balance, there are a number of distributions; the market (condition) is directional.

A Directional Overlay Demand Curve
Figure 5 is a directional market. Applying the Overlay process to it results in figure 7.
Attached Image (click to enlarge)
Click to Enlarge Name: 5 & 7.JPG Size: 91 KB
Markets are not efficient, rather they are effective - Jones
3) The Overlay Demand Curve (Composite/Merged profile)

Overlay Demand Curve Value in a Directional Market

Value was found for the balanced market of figure 6. The concept of value depends on a stable, balanced market, one where upper and lower bounds are apparent. In directional markets no such bounds exist. Price moves and pauses, then moves some more. The pauses are short time frame congestions. A pause may last minutes to hours, sometimes even days. A general rule is that value takes time to develop (value is price over time). Practically, value is not apparent in pauses and so, practically, value is not known in a directional market.
Attached Image
Markets are not efficient, rather they are effective - Jones
3) The Overlay Demand Curve (Composite/Merged profile)

Profiles can Only be Understood in Context (Condition)

Comparing profiles from a balanced and a directional day will illustrate the contention that a single day's profile in an unknown context (condition) is lacking in trading information. Take Nov 2 from figure 4: it ended trading near it's lows (N period). In figure 5, the Nov 8 profile likewise ended near it's lows. There is little to differentiate the two profiles, they both started near their highs, balanced and ended lower. Strictly reading profiles, it would be hard to know that the Nov 2 day was part of a balance while Nov 8 was an initiating day for a downside breakout.

The Overlays of figures 6 and 7 are perfectly clear. Within the context of market condition, a trader can interpret the profiles of Nov 2 and Nov 8 appropriately.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: 4 & 5.JPG Size: 104 KB
Click to Enlarge Name: 6 & 7.JPG Size: 132 KB
Markets are not efficient, rather they are effective - Jones
Quality of the data.

When dealing with experimental data an appropriate but ofttimes hard to answer question is 'how well does the data being used address the question (s) being asked'. An important question is "how well does the data fit the theory"? The theory, of course, is that the profile data fits the a bell shaped distribution curve, with the close inside the distribution.

Profiles assume a bell shaped (Normal or gaussian) distribution (TPOs for Meta-Profiles). One condition, the close within the distribution is easy to check. But what of the assumption of a 'bell shape with the maximum (TPO count) at the center and a symmetrical distribution about the maximum?. Such a check can be done mathematically, which the college statistics courses do: Students take (or are given) a series of measurements, plot their data, locate the mean value on their plot and then find the central 70% (actually 66 percent). With the school data it is generally easy to see if the mean is at the center of the distribution and if the central 70 percent of observations is centered around the mean. Symmetry is pretty easy to estimate by eye.

Most traders who use profiles blindly take the calculated numbers and apply them. But if the data do not fit the theory (if in fact the data for that market for that day or days does not fit a Normal distribution) then trading decisions based on the data BEING normally distributed fit into the category of guesswork.
Markets are not efficient, rather they are effective - Jones
What can be done to ameliorate the guesswork, the trading on questionable data? Possibly the first step is to find the market's condition. If balance exists, the data can likely be trusted. Without balance value cannot be found and the profile concept collapses.

First, you, the trader, must realize that a problem may exist. Most traders are not there yet.

Second, you must understand that the industry, your data sources, are as blind to the problem as you are. They consistently compress profiles to get them on a page. This can remove much of the detail from a profile, making it much more symmetrical appearing than is the case. Compression causes problems at the basic level, the definition of a distribution. That is:If we define a distribution cut-off point as three TPOs then we must be able to SEE where in the profile the distribution goes from 3 TPOs to 2 TPOs. Take a double distribution situation where the TPO count between the distribution goes 4, 3, 2, 1, 2, 3, 4. Now compress the plot by a factor of 10 (not unusual with long profiles). The compression can easily cut out the 2, 1, 2 points, ending with a single distribution after compression
Markets are not efficient, rather they are effective - Jones
The new TPO Chart 2.1 indicator has recently been made available for the new mt4 build. The new merge feature is a work of art!
the indicator can be found at http://www.mql5.com/en/market/product/3033. you can see a video of how to use it in the comments section.
So far it's amazing!
Attached Image (click to enlarge)
Click to Enlarge Name: TPOChart2.1.JPG Size: 157 KB
Markets are not efficient, rather they are effective - Jones
At the outset Thank you for putting such an effort in bringing such an educative subject on AMVT,

After reading all of your posts for few day ( sometimes twice,Thrice), have just made an attempt to bring the subject in to practice.(On Paper by now).
Posting some of my views on the same.
As Iam using NT7 for plotting the Meta-Profiles with one min Data, as the tick increment is of 0.05 (20 Increments for a Point) and the range of the instrument Iam tracking is almost of 80-100 Points Its bit difficult to plot. Hence I have made the tick increment of 1 Round Point. My worry is am I distorting the market generated information or It should not be much of a problem apart from shifting the reference points by few points.? (A sample work sheet attached for reference)
Also, Iam bit confused on quadrant and Octant's issue. As mentioned in most of the documents it says "Octants 1/8th of the Bracket limit and Quadrant 1/4th of the Bracket Limits". But while seen some charts on Cisco-Futures site, Iam unable to see the figures coinciding (OR I may be making some silly mistake).
With my understanding have made some work on a 5day ODC, would like to see and comment on same,Please.

I would also request you to post more charts (If Possible) so as by seeing that I can understand the subject more in detail.

Regards,

Santosh
Attached Image(s) (click to enlarge)
Click to Enlarge Name: 5day ODC 28.7.PNG Size: 21 KB
Click to Enlarge Name: AMVT sheet.PNG Size: 34 KB
If it was easy, everyone would do it.
Quoting santoshdts
At the outset Thank you for putting such an effort in bringing such an educative subject on AMVT, After reading all of your posts for few day ( sometimes twice,Thrice), have just made an attempt to bring the subject in to practice.(On Paper by now). Posting some of my views on the same. As Iam using NT7 for plotting the Meta-Profiles with one min Data, as the tick increment is of 0.05 (20 Increments for a Point) and the range of the instrument Iam tracking is almost of 80-100 Points Its bit difficult to plot. Hence I have made the tick increment...
Welcome, I appreciate the post.
First, its the data being used. What you're describing by using "1 min data", you're calculating ticks (tick volume) not TPO's. What I think you're tracking a tick distribution, rather than TPO distribution. A TPO uses 30m data. If you can post an fx pair I will post a chart so we can compare and I might be able to explain what the differences are. Also do you know "how" a TPO is calculated?

Perhaps, I'm wrong, but what is the 1min data used for? The only thing that comes to mind, is many tick volume profile indicators use 1min as a way to calculate Ticks. TPO's are calculated using 30min OHLC data. Have you tried to use 30 min data?

As for the octants & bracket limits are determine by TPO count. I addressed the "data" issue. If you're calculating ticks at price, and not TPO's of course they wont coincide, two different sets of data. Tick data is "at price" data, TPO's is price through time. Two different kinds of distributions that use two different sets of data.
Markets are not efficient, rather they are effective - Jones
Quoting santoshdts
At the outset Thank you for putting such an effort in bringing such an educative subject on AMVT, After reading all of your posts for few day ( sometimes twice,Thrice), have just made an attempt to bring the subject in to practice.(On Paper by now). Posting some of my views on the same. As Iam using NT7 for plotting the Meta-Profiles with one min Data, as the tick increment is of 0.05 (20 Increments for a Point) and the range of the instrument Iam tracking is almost of 80-100 Points Its bit difficult to plot. Hence I have made the tick increment...
Are we looking at the same pair?
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Capture.JPG Size: 210 KB
Click to Enlarge Name: Capture 1.JPG Size: 246 KB
Markets are not efficient, rather they are effective - Jones
Thanks for the reply,

Now most of my basic query's on data and reference are sorted out, Thanks.
Now I get the point, I have initially tried plotting based on 1min OLHC, as The Day range is big and min. Price increment is of .05. It was getting a bit heavy on the CPU,now Plotting the profile based on 30min OHLC.
Initially my worry was about visibility of profile, as with .05 price step the profile was getting too compressed( due to a big range of 75-100 Points( around 5000 TPO's/Day), then i increased the Price step to 1.0 it increased the visibility (Around 250-275 TPO's/Day), but was missing some TPO's (That Price Occurred )in the process.

By looking at your 2nd & 3rd chart. the doubt on Octant and Quadrant is also resolved.
If it was easy, everyone would do it.
Quoting santoshdts
Thanks for the reply, Now most of my basic query's on data and reference are sorted out, Thanks. Now I get the point, I have initially tried plotting based on 1min OLHC, as The Day range is big and min. Price increment is of .05. It was getting a bit heavy on the CPU,now Plotting the profile based on 30min OHLC. Initially my worry was about visibility of profile, as with .05 price step the profile was getting too compressed( due to a big range of 75-100 Points( around 5000 TPO's/Day), then i increased the Price step to 1.0 it increased the visibility...
Sounds like you were dealing with way too much info to manage.
I'ts good to hear you're getting it worked out.
Markets are not efficient, rather they are effective - Jones
0 traders viewing now