Hey lj, I see what you're looking at. Generally you'd expect to see less overlap or more trendiness with higher volume. So it has to used in some relation or comparison to the bar's volatility. Looking at the chart is the best way to see how the overlap and volatility compares to the matix. In the EUR/USD chart I used 1 ATR for volatility. The Volume/Volatility matrix charts show a tendency, but doesn't account for the real distance traveled in the bar. A 10 point bar vs a 10 point bar traversed several times. And then at the ends of the matrix you have the bars that go against what the average bar is expected to do. I mistakenly called this an anomaly at the time, but it turned out to be a very important observation - I was looking at it differently, turned out to be change. The overall point is expect a certain bar volatility when you have a certain volume. Using overlap as a percentage might kill the signal or indication. Then there are the inside and outside bars whose overlap can be identical to a bar in a completely different context. Again looking at the chart, high overlap can mean your changing direction, or going sideways. In a strong move you expect higher volatility with less overlap so I can see how a % volatility to overlap would help there. But I'm not sure how that's going to get us a volume proxy. Combining the overlap with volatility might account for the extra traversing , hence the activity inside the bar and help out the volatility to be more accurate volume representation. Regarding indicator comments, so many have the same information just displayed differently. And they all have smoothing, so lag, built into them. But it's all in how you look at them, what you use them for. Then lag isn't a problem or a consideration. There's quit a bit of useful information that can seen in them - a topic for another thread. Here's the daily volume volatility matrix for the ES for reference: http://www.elitetrader.com/vb/attachment.php?s=&postid=1430247
Hi Ezzy, Good points. Agree 100% with your interpretation of the 'anomalous' behaviour. I don't think it matters volatility-wise whether or not there are intrabar (IB) traverses because of how volatility is defined, either as simply H-L/unit time or as you have done with the ATR (which also has a time constraint). I see your idea of how if there is minimal intrabar traversing, then the volatility is more 'meaningful', but there are IB traverses which are definitely meaningful and the classic example would be the IB Gaussian shift. The question is how to deal with something other than this and of course we know that there are experienced traders who would just 'trade' the IB traverses; others would hold their position with or without a fractal change. Inside bars are special cases of lateral price movement and outside bars might be dealt with by looking at what follows immediately after the bar ('anomalous' behaviour may be important here). However I think what we are both asking is whether there is some way to reasonably anticipate that if there is such and such an amount of overlap and if there such and such an amount of volatility, then what can be anticipated to happen next? The question of how best to use volatility in the FOREX market as a proxy for volume is non-trivial but one idea would be to contrast # of ticks/unit time to tick range movement/unit time and then look at some sort of acceleration parameter as the anticipatory trigger. It could be quite Boolean. It's great to exchange ideas and thanks for your thoughts. lj FWIW, for those who might think this stuff is OT, IMO this stuff looks at ways to figure out when your trend(line) is changing.
This is brief comment that will, hpefully clarify a few things. I will do the details later. Sample size for lookup tables has to be adequate and at the same time contemporary. 1600 bars a month on the trading fractal is reliable. Markets move with a rhythm that is remarkable. To trade to make money, a person adapts to that rhythm in several ways. There are many ways to look at the markets from a tooling point of view. I feel that at about 100 milliseconds, the human has reached a limit and, there, the computer takes over and processes faster signals (and faster data) from the market. To make such possible to understand and utilitze, there has to be a way to feed this fast information to the trader and in a manner that is compatible with the trader's routine. I compare two routines ordinarily simply because both are based on Science and they are very high in contrast. OODA and MADA. OODA the fighter pilot "reaction model" is not a good one to use for making money simply because it "tests" the market with capital. MADA deals with a binary vector orientation which make it non-subjunctive; there are no "IF's except as Boolean statements which are set comparisons which always have the same correspondence. So for the very fast paced flow, an animal/neural type filtering works best. Anything that is unchanging/static is understood as is and anything that is dynamic gets attention. For MADA the Analysis has specific conclusions and the conclusion is a statement that always includes a set of factors. The conclusion, s a statement is comprehensive and complete (as measured using a sufficiency test). Forex markets have one twist that has to be woven into their fabric. In one activity range (volume) they tape very stridently. This just means that things are simple for part ot the trading time. Pace overrides all other factors in forex and the tapping seam goes down the middle of pace dividing pace considerations into three categories (the seam is one of them). This was why I felt a general approach to markets could be modified to get to the nub of making money more quickly. Trending/reversion is measured quickly and repeatedly and is used in the Analysis process which leads to the complete conclusion statement. Using volatility and overlap to define reversion's presence simply meant that is was easy to define the mutually exclusive time-price regions of the market's operating point into either trending or reversion. The case distribution is: 2 trending; 1 restart; and 7 reversions. Trades take place in none of these cases. All of them demand a "hold" for trading Forex. Trading may only take place when case change occurs. This means that analysis is all done with respect in reference to the prior analysis conclusion. Naturally, this is built into the analysis process by taking data sets that are vector in nature which assures this practise. The key log notation that I make to record the Analysis conclusion is the MODE (X for change and C for continue) WWT and it's successor, WTF, are the informative things that explain the degredation of data set sufficiency; this beginning incompleteness signals very near term case change. Trend volatility exceeds reversion volatility; therefore, holding through reversions is likely. This is the lateral aspect of the lookup tables. Pace increases are indicative of increased money velocity for making money. Pace decreases lead to reversion ordinarily. Both of these are vertical aspects of the distributions. Trading, essentially, comes down to vector analysis and it resembles animal/neural functioning. If a constant signal is precent, then it is not a significant signal; only a changing signal is significant. The small core of boxes from the few raw signals is improved in two directions for Forex. One is the Global indicator that gives a preflight analysis for the wolrd as it rotates. The other improvement is the one that provides sensitivity by drilling down into the details. On any given pace, we take what the market offers during the time when cases change. Adding in the RTL and LTL as a box set, provides more context which is at the left of the boxes overall lay out. The vaste space in the middle is for developing sensitivities to signals that have a sequential relationship. WWT fits here. On the right, is the gating that funnels everything down to the heartbeat of doing the turns. You may visualize the Forex as only having so much raw data. By processing data, you get information of quality and this information is tempered by it's importance at various times. You will find that you have everything you need. Quality is dictated by the markets in terms of "knowing that you know"; importance is created by the trader (in terms of optimizing, efficiency and effectiveness).
Regarding "one" causal factor: Limited movement by hitting support or resistance, a shift in sentiment, the balance of buyers/sellers changes or one group can't overcome the standing orders. All is a factor of volume, or insufficient volume to push price past that point. The intraday pace was divided in to deciles for pace. The daily was 5 groups. I would assume that 5 paces for overlap is sufficient. Because Forex goes 24 hours there is the question is of whether the pace and overlap should be based on all the entire week's session data or just the more active sessions during the week. Even though there could be some skewing during low volume hours, I think using volatility and overlap would take that into account. A full week of 5 min bars would give a 2,016 sample. Catenary is a more apt description of the curve (pace distribution). - EZ
There's a few other ideas on volume, though now that CD23 has explained why to use overlap, they might not add value. Thought quite a bit about just using the raw ticks vs range to see divergences (range vs volume/tick discrepancies). Like in the EUR/USD chart. The drawback is a large move on few large trades vs. many small - you don't know if it's size moving in. But still could be useful. Another idea I've heard is using lower fractal bars, say 1 min and stacking/adding their volatility to substitute for volume. Taken to the extreme, at some point it would become the same as tick volume. Another one is taking volume from the various futures exchanges which would require some very creative coding, if it's possible. Instead you could just have the exchanges chart up and watch volume there - and hope it tracks. Which leads into the issue of spot vs. futures contracts, is one leading the other? It would be nice to have a volume chart to draw trendlines (of gaussians) on, but have to use what's available. There was some talk of the major banks putting together a quote system to display several levels of orders, like the DOM. Haven't heard anything more about it but that would be great. I'll have to work on getting a platform where I can code the overlap. Qcharts doesn't let you code. - EZ
It's called the Price-Volume Relationship for a reason. Ever see the drill where people have Volume and are asked to draw in the Price bars? It's been posted a few times. If you know one part of the relationship, you already know the other part as well. - Spydertrader
Spyder, We all here know that it is possible to construct an approximate picture of the V when only having the P, the problem is to determine what is the nature of the V. Jack's got three proxies but for me at least, it's not yet apparent what they are. That may change when I understand more clearly what he has said so far but as yet with this particular aspect of things I'm not in a position of "knowing that I know". I do however know what it feels like to "know that you know" and as MAK once said it has nothing to do with "thinking that you know". lj
I've done that drill, more than once, aaaaannd you probably knew that . Are you suggesting that because of the fact that by knowing one part, we already know the other - that what we have is sufficient and forget the other stuff? As in trying to fine tune a volume type indicator or come up with new one is a waste of time because the volatility/volume is so highly correlated, volatility is all we need. I know you're not suggesting volume isn't important, and all we need is price. So point taken. Thanks - EZ
You know this as well. Hence the name: Price - Volume Relationship. Remember, the ultimate goal does not involve knowing the exact Volume level. The ultimate goal reflects the trader's desire to understand the signals provided by the market. As traders, we 'see' these signals reflected in the market mode - continuation or change. - Spydertrader