There is an obvious problem in trying to cross-correlate price change and volume: price change can be positive or negative, but volume is only positive. Actually volume is only REPORTED positive. It can be either positive or negative, a fact that never seems to disturb those who talk about the relationship of volume to price. But all volume in a bar is aggregated. In fact positive volume is very close to negative volume, at least for NQ, so this is not a big issue. (Believe me, if it were, and it had any value, I wouldn't say anything about it.) So the right way to cross correlate price change with volume is to make all price changes positive by taking the absolute value. As Infolode hinted, this amounts to frequency doubling because it is full-wave rectification. The scattergram for rectified price change versus volume is attached. Is there a relationship there? If so, it is pretty weak. We'll see when we do the cross-correlation that the correlation coefficient is so low as to make an hypothesis of causality between price change and volume suspect.
Joe Doaks' Data Analysis
- Thread starter Joe Doaks
- Start date
For clear lack of interest I will wind up this misbegotten thread quickly. The cross correlation function of pure volume and the first difference of price has values so low that clearly they are so nearly random as to be of no predictive value, as was evident from their scattergram. The first +/- 20 shifts in the cross-correlation revealed no peaking behavior suggestive of causation, only near-randomness. Taking the analysis one step further I analyzed the first difference in price and the first difference in volume, solely for the sake of compleness. The scattergram is attached. There is nothing there. The cross-correlation degenerated into randomness on the first few shifts. So in conclusion (and don't say you didn't see it coming):
THERE IS ABSOLUTELY NO FUCKING MATHEMATICAL JUSTIFICATION WHATSOEVER FOR THE ASSERTION THAT VOLUME LEADS PRICE! CRAYOLAS ARE NOT MATH!
I'm done. Thank you for your inattention.
THERE IS ABSOLUTELY NO FUCKING MATHEMATICAL JUSTIFICATION WHATSOEVER FOR THE ASSERTION THAT VOLUME LEADS PRICE! CRAYOLAS ARE NOT MATH!
I'm done. Thank you for your inattention.
Thanks for Volume/Price data analysis!
(not sure how wacky cult beliefs get started, but they die very slowly if at all...)
I know very little about Crayola, but I do know a thing or two about Volume leading Price. If Volume X causes Price volatility of Y distance, then volume X + W creates Price Volatility of Y + Z distance. I describe such an environment as a period of continuation. More importantly than continuation, are the sequences which occur during change. I'll leave those sequences for you to discover on your own (or in person over a bottle of vodka).
One need only run Volume vs. Volatility (complete with a standard deviation or two) across the last 1600 5 minute ES price bars (it helps to break the total into components subdivided by market 'pace' as well) to 'see' the mathematical proof you claim fails to exist.
Good Trading to you.
- Spydertrader
Thank you for putting finis to the thread. I have an interesting little study I run on NQ called Volume Per Move that I think might shake your confidence in your position. But, hey, if you make money with it, who am I to argue? I, too, look for continuation versus change. Mine is a little simple. I just ask of every new bar "Are you still green (or red)?" And if that bar says "No", isn't, I ask it on the next bar "Have you made your mind up yet?" Very difficult TA to master, haha!
Actually, the best way to "see" the "P,V boolean relation" is by getting drunk on grape kool-aid.
At the link below is Jack Hershey's landmark paper, "Catch Up with Tomorrowâs Paper Today," (adobe llamingos publishing) in which he describes the "P,V boolean relation," and how to score it.
http://www.elitetrader.com/vb/showthread.php?s=&postid=1278985&highlight=buying+boolean#post1278985
On page 8 of "Catch Up with Tomorrowâs Paper Today," Jack says to buy at the turn from 0 to 7.... he mentions no filters or further requirements. So I tested that on 1000 S&P stocks over five years, exiting 5 days later* and got the attached equity curve. So much for having tomorrow's paper today!
* 5 days worked better than 1,2,3, or 4 days.
http://www.elitetrader.com/vb/attachment.php?s=&postid=1278986
At the link below is Jack Hershey's landmark paper, "Catch Up with Tomorrowâs Paper Today," (adobe llamingos publishing) in which he describes the "P,V boolean relation," and how to score it.
http://www.elitetrader.com/vb/showthread.php?s=&postid=1278985&highlight=buying+boolean#post1278985
On page 8 of "Catch Up with Tomorrowâs Paper Today," Jack says to buy at the turn from 0 to 7.... he mentions no filters or further requirements. So I tested that on 1000 S&P stocks over five years, exiting 5 days later* and got the attached equity curve. So much for having tomorrow's paper today!
* 5 days worked better than 1,2,3, or 4 days.
http://www.elitetrader.com/vb/attachment.php?s=&postid=1278986
Or, one could simply follow the instructions in the previous post.
- Spydertrader
'Volume per move' isn't quite what I described, now is it?
- Spydertrader
LOL, you're a Jenius!
Shssh, stop giving out the dayumn secretz.
Mr. Doaks failed to... "simply follow the instructions".....
he he he...LOL....