8
$\begingroup$

volatility clustering and mean reversion are very well known properties that one could use when trading. Traders, especially in options world, do take realized vol into account (e.g. by forecasting it or looking at which percentile does the current volatility correspond).

I am wondering if also intraday volumes have the same kind of properties that can be exploited somehow.

I see that some traders look at volume profiles and use indicators like VWAP (volume-weighted averaged price) and PVP (peak volume price, thus the price where the largest intraday volume was traded). In general they assume that intraday volumes tend to generate a symmetric distribution thus following this kind of rule to forecast the price direction:

if the PVP>VWAP then the volumes distribution is skewed upside and this generates a "pressure" to prices to move downwards, at least untile the VWAP. With PVP

There is an exception to this rule: when the price action is on one the extremes of the volume distribution (e.g. price>PVP>VWAP) then the previous logic doesn't apply (even if the PVP>VWAP).

Is there any statistical evidence that intraday volumes actually tend to generate symmetric distributions thus making it possible to exploit the temporary skewness that is generated intraday?

Is there any study on that or anyone willing to share her/his experience on that?

Thank you for your help.

Improve this question
$\endgroup$
10
  • $\begingroup$ Intraday volumes tend to generate a symmetric distribution - What do you mean by a symmetric distribution, do you mean one without skewness? What are the data points around which a PDF is being estimated - tick by tick volume or bar volume? Also why would PVP > VWAP skew the volume distribution? What are you defining as "volume" exactly, since under the standard definition of the number of contracts/shares traded, I see no theoretical reason why PVP>VWAP would cause skewness in the estimated density of this distribution. $\endgroup$
    – user2763361
    Mar 5, 2014 at 14:10
  • $\begingroup$ Hi, for each single price during the day, we have the total volume traded at that price. This will create a distribution of volumes. Each price is weighted by the corresponding volumes traded at that price (VWAP). This average price will change during the day and will be the average price to which one would compare the current price. If the current price is above the VWAP it means that you are buying a contract at a price that is higher than what has been considered (on average) as fair price up to that moment (VWAP). $\endgroup$
    – opt
    Mar 5, 2014 at 14:43
  • $\begingroup$ The PVP is the price at which the largest volume has been traded. If it is different than the VWAP this will create some skewness in the volume distribution. Indeed, the PVP is the mode of the distribution, while the VWAP is the mean. As long as the "true" volume distribution is supposed to be symmetric (thus implying mode and mean to be almost the same) then whenever the distance between the PVP and the VWAP increases (in absolute sense) this will pave the way to a mean-reversion in the price. As said, PVP>VWAP suggests that prices will move down. $\endgroup$
    – opt
    Mar 5, 2014 at 14:44
  • $\begingroup$ This "volume distribution" that you are talking about still has not been defined. Is this a PDF/density of traded volumes? What is this distribution and why does it change conditional upon the price at which trades were made? If the y-axis of this distribution is the density then what is the x-axis precisely? $\endgroup$
    – user2763361
    Mar 5, 2014 at 14:45
  • $\begingroup$ Suppose that during the day just 3 prices have been "touched": 9, 10 and 11. The numbers of contracts traded at 9 is 100, at 10 is 90 and at 11 is 87. This will be the "distribution" of the volumes for that day and you can calculate some statistics on it. Indeed, the mode (the peak volume price) is 9 because is the price at which the largest volume traded. The volume weighted average price will be: (9*100 + 10*90 + 11*87)/(100+90+87). So VWAP is 9.95. In this example, the mode is lower than the mean and (if the volume distribution is supposed to be symmetric) this should push prices up. $\endgroup$
    – opt
    Mar 5, 2014 at 14:58

1 Answer 1

Reset to default
1
$\begingroup$

The relation between volume and the price dynamics (via volatility and jumps), has been explored by various academic papers. Just cite this one and its contained references:

Wang, T., & Huang, Z. (2012). The relationship between volatility and trading Volume in the Chinese Stock Market: A volatility decomposition perspective. Annals of Economics and Finance, 13(1), 211-236.

If you have a look at it you will read

  • volatility and volumes are positively correlated at any time scale,
  • and if you zoom enough to see "jumps" in the price you will note a negative correlation between them.

It is usually explained by the "common" trading activity generating the continuous component of volatility, but information generates jumps and do not need a lot of volume. Moreover jumps are followed by low volumes.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.