Here are some notes on a technique for mapping the relationship between spread and volume using a simple 3-D graphical representation, designed primarily as a tool for a "right brain" approach to reading a stock chart. I am experimenting with using this as a first-pass filter for identifying potential trade setups, for subsequent confirmation with VSA. I'll add more information over time.
As a quick introduction (more will come later), look at the figures below. The center graphic shows a V/S map of a number of key stocks in FTSE over a 5-year period. Try to visualize this as a map with contours - you are seeing an aerial view of a somewhat oval shaped mountain with a peek at the center. The left hand graphic shows this mountain in 3d (though with some loss of detail).
To create this map, I have taken all the end-of-day price and spread figures for the stocks in the FTSE100 over a 5 year period, and placed them in the relevant possition on the map. One axis is volume, and the other is spread. The figures are normalized using mathematical formulae to place a score on relative volume and spread before being plotted. It is obviously important that the scores have the same significance for all the stocks included in the analysis, and focus on "relative" scores. So we define two normalization functions..
- FnS(spread) -- normalizes the spread values.
- FnV(volume) -- normalizes the volume values.
We then add a dot to the map for each bar, using these cooked values. This means that each EOD price/volume bar becomes a single point on the map, and the density of the dots indicates the frequency with which bars with a particular of FnS/FnV score pair occur. The next step is to plot these densities on the third (Z) axis. We again "normalize" these to generate a visually meaningful map - without this step the center of the map swamps the fringes and it becomes hard to read. Note that this is done using logarithm maths, NOT averages (so there is no loss of detail).
Now we have the 3-d figure shown in the first graphic below. The next step is to turn this into an aerial view and define a set of contours that allow the shape of the graph to be interpreted visually. Now we have the second (middle) graphic below. The circular rings have been added to visually highlight the off-circular nature of the base of the "mountain".
A few things immediately "jump out" from this representation..
- The chart is not circular. If there was no correlation between volume and spread, then a largely circular set of contours would be expected (indeed this can be demonstrated by "disassociating" the volume and spread before plotting). This shows clearly that spread and volume are correlated in a way that supports some of the key tenets of VSA.
- The figure is "squashed" on the upper left - showing that bars with wide spread but low volume are unusual. There is a similar but lesser flattening on the lower right - showing that low spread, high volume bars are also unusual - but not as unusual as wide spread bars on low volume.
- The figure is elongated along the diagonal from bottom left to top right, showing a favoring of volume/spread correlation. Ie: high volume is normally accompanied by wide spreads, and low volume by narrow spreads. This supports the idea that volume represents "effort" and spread represents "result", as per one of the laws of Wyckoff and VSA.
- The high point of the map is not at the center. This means that the most commonly occurring relative volume/spread bars are not also the mean values, but are skewed towards a combination of high volume and wide spread.
Now to use this information on a chart.. The third graphic shows a colour-coded V/S mapping histogram under the price bars for a FTSE stock. The bar height is an inverse of the "height" at which the bar appears on the map - and hence is roughly equivalent to the distance from the center (if the map were circular, which it isn't , and shows us how "significant" the bar is. The colour is used to indicate the radial position in which the bar appears, and hence classifies it in terms of high/average/low spread and volume.
This technique is very much a work in progress, but you may observe the following early indications in the right hand graphic.
- The bars in the lower part of the right-hand graphic represent the location of the price bar on the map (the left hand and center graphics). The bar height represents the degree of significance or "unusualness" of the price bar (the Z-axis height of the bar's position on the map). The colour of the bar denotes it's rotational position on the map, as follows..
- Red: volume and spread are both higher than average
- Pink: wide spread on average volume
- Orange: wide spread on low volume
- Yellow: average spread on low volume
- Green: volume and spread are both lower than average
- Cyan: narrow spread, average volume
- Blue: narrow spread, high volume.
- Purple: average spread, high volume
- The gathering of power (effort to rise) on narrow spreads where a cluster of rising VSM bars are cyan (light blue) resulting in an upward move as the bars transition to purple. The cyan bars are seen to take the momentum out of the preceding downward swing.
- The dwindling of enthusiasm for the upward move where a group of green and yellow bars appears, resulting in a fall in prices soon after. Tall green VSM bars suggest stagnation of the move.
- The subsequent downward move is maintained while the VSM bars are red (Red is a sign of commitment to a move in the direction of least resistance).
- The downward move falters as the VSM bars start to take on a more blue/green-ish hue (Green is a sign of stagnation - a lack of commitment to the move).
- The "constrained explosion" at the tall dark blue bar - where volume is ultra high, and the spread ultra narrow, stopping the downward move dead in it's tracks, resulting in a somewhat cautious upward swing.
Please note that VSM is far from ready as a trading tool and ignores a number of VSA inputs (such as the relative possition of the close within the bar spread). This is simply the result of a set of statistical experiments, showing up some interesting patterns - but the development and proving of the reliability of the method is not yet complete.