Nathan asks:
Does volume play any role in your analysis? Do you look at increased volatility in conjunction with increased volume? Also, since the spot FX market does not have any volume, does the volume in the futures FX carry any significance?
Good question, Nathan. Short answer is no, I don’t use volume in my analysis or trading in any way. The reason is simply that I have been unable to verify a statistical edge to volume, with a few exceptions. (For instance, it appears that equity indexes making new highs on dramatically lower volume is a slightly bearish sign, even when adjusted for seasonality of volume.) I am obsessive about only using things that I can show have an actual edge, and this means that many of the traditional tools of technical analysis… well, I don’t use ’em. Moving averages? Nope. Statistically speaking, there is no edge to a market touching any moving average. Fibonacci ratios? Covered that one here and here. Volume? Negative. Candlestick patterns? Pretty names. I could go on (and, in my trading course, I have gone on and on) about what doesn’t work, but I think there are more important points.
The first, and, truly, probably the most important thing I have to say to anyone who has anything at risk in financial markets: be sure that what you are doing actually works. Everything is testable and quantifiable, and there is no good reason to not understand your edge. Don’t accept anything at face value because it is in a book, a trading course, a curriculum for a test, etc. It is your money, your time, your emotional capital, and a piece of your life at risk–is it not worth your time–no, is it not your responsibility–to make sure that what you are doing actually works? (And how much more so if you manage client money and risk!)
How would you test volume? Well, one simple way is to take a pattern that has a positive expectancy, and then separate them by those that have good volume support and those that don’t. Compare stats for the two, and does the set with volume support show that it is better in some way? Higher return, lower variance, higher win rate, or some other statistic? It is so easy to find books that explain why volume is critical (and it is completely logical that volume would be critical), but what do the numbers say? (As an interesting aside, Bulkowski actually did this work in one of the later editions of his book, and was surprised to find that volume did not add an edge to the patterns he was examining.) Volume is pretty tightly correlated to range and some measures of volatility, so you risk adding a highly correlated input to a trading system if it is already based on chart patterns, and highly correlated inputs are never a good thing.
I wonder if the volume studies were done on one minute charts or daily charts or what.
For instance, on a breakout on a daily chart that showed high volume, we don’t really know what the volume was associated with. But on a one minute chart breakout, you can tell with some precision that the volume came right with the break.
I wonder if that makes a difference.
Good points. Without combing through a bunch of old data I can’t give you a firm answer, but I know I did look at intraday and daily both… probably much heavier on the daily. Again, I’d assume that if volume is truly significant it should scale and should be apparent on any timeframe, but perhaps I’m wrong.
(There are quite likely some intraday distortions that may have volume significant attached. (poorly worded, sorry). But they also have range and volatility moving too… so there’s again a high correlation between those inputs.
Hello Adam, did you try Market Profile / Volume at Price? To some Traders it seems to be the
“philosphers Stone”. While for others like Al Brooks there is enough Information on a simple Chart, even single time frame
I have looked at market profile but certainly there is a lot I don’t know about it. My experience was limited to most of a year using it intraday and I haven’t really done much if any testing on the concepts. Some people do find it useful.
You said: “How would you test volume? Well, one simple way is to take a pattern that has a positive expectancy, and then separate them by those that have good volume support and those that don’t.”
The problem with finding an edge in volume is that you’re using it just like everyone else does. “Good volume support” in your world means that a breakout on higher volume is “confirmation,” for example. This is how everyone views it. So of course there is no quantifiable edge with this.
“Everything is testable and quantifiable,” only if you test and quantify the same data points everyone else does. Patterns are quantifiable, but not all data can be so easily compartmentalized into a “pattern.” Sometimes original, case-specific thought is necessary, without the reliance on a “pattern.”
There’s some truth to that but there’s also the point that if something is significant and truly useful you shouldn’t have to look too hard to find some quantitative proof. Absolutely possible that I’ve missed some way I should be looking at volume… no doubt, but I do ok without it, too… so there’s the other side of that argument. 🙂
Good points. Thank you for the thought provoking comment.