Firstly what is a rule in a market. It is a regularity. The problem is that as a proposition it exists within a context of an underlying system, which is probably not rule based. This is like the way a chart looks like it has a directionality, after the event. Rules are generated by the market on an ad hoc basis, essentially.
Getting at the underlying process may have issues in information decay, but getting at the process of rule generation may indicate the appearance of regularities, or their dissipation. That is one would want to know whether a regularity such as that occurring when money flow fades rapidly, will be effective or not on the market. That is there is something more to the effectiveness of a structural trade such as this.
That is what is the relationship of money flow to structure in the market at a given time. One does not have to trade on time frames determined by the market that does not reference time frames except as they act like buffers for process in the sense that order flow acts as a kind of stop point for valuation.
So one asks what are the conditions for regularities to appear, based on real time analysis, not on past appearance as one can assume that these are essentially, over time, not regularities at all (as all who trade on them can attest, it is like chasing a will-o-the-wisp). One should look for what the conditions are for the appearance of those regularities as a staring point (one can only look at past data).
One can ask whether those regularities are not particularly time dependent, that is do they have a structural integrity that is a persistence or more than a persistence. One can expect a persistence to decay, but one might ask are there such regularities that they will not decay, but they will in fact be turned off by processes within the market.
That is, can one expect that those regularities one is looking for at a given time frames act to turn off such regularities. Indeed, as that is what they are doing. Thus to trade on them to turn on a process is the problem, as this is a random event, what is not random is the process of turning off another process. That is not deterministic, but it is biased, and in this suitable indicators could help, as they can show bias.
But what bias, that is the issue. One might say a bias to a process being turned off. This may give a justification for RSI, in that it can be seen as a binary indicator. It may be the case that indicators work well in forex as trading signals (that is for trying to trade on regularities) in binary mode. So if you are looking for binary signals, you are not given the directionality.
In forex, a valuation can quite easily move opposite (or not at all) to the apparent bias in the market. That is the problem of trying to use indicators in a non-binary way, because if you do this, you are tying to find directionality in an indicator. There may be an argument for using these indicators this way in equities. Whether there is an argument for using them in day trading equities is another matter.
I might suggest there is (and the utility of Stochastics there) but in forex the discrete nature of pattern formation, that is the practically random appearance of crystal clear regularities, is the problem. But there does seem to me to be a use of RSI to indicate when an appearance of a regularity may be happening, which is the equivalent of a directional use of indicators in equity day trading. The problem with this is time frames, that is there may not be a fractal structure to this process formation.
This may give a justification for multiple time frame analysis. The problem is that does process come from external source or is it an internal process. That is, what is the causality of the market itself. Its deep links into equities (in fact, the forex market can collapse the equity market into itself) suggests a way to find a warning signal of a process formation. This assumes that the market itself is adaptive to process itself.
However I do not believe that the market lacks continuity, there may be regularities in the market between sessions for example. That is, the appearance of process is a function of time frames, which may make multiple time frame analysis problematic, but of course that is the issue with all approaches in forex. It comes down to this: one must analyze the market as it is at the time one makes a trade, and assume that the past has little causality on the present.
I believe programs in forex may be a great area of application for AI. The underlying problem is that programs need to do the same thing. One might argue that a program which can take advantage of storage and reliable precision of reference might have some advantage. The mind has such advantages, but the problem is the reliability of biological data processing.
So what does one analyze, one analyzes, in my view, large scale structure to constrain valuation, and within this one tries to find a cohesion of the market at the time one makes the trade with this, as indicated by suitable indicators, along with the usual elements to keep an eye on.
This assumes the processes of the market are strong. But one needs to bear in mind that forex valuations are not inherently about increases or decreases, in fact the underlying process is one of stability. In forex, one is always fighting the market and the same issues apply as with this in equities.
© 2011 Guy Barry - All Rights Reserved.
