Although you probably have already noticed it, the trading systems are evolving and becoming increasingly technical. As you can imagine, some professional traders recommend to make a qualitative leap and try to convince once and for all that the use of trend lines, price chart patterns and obsolete oscillators in the same way that comes in thousands of books it is not exactly the best approach to winning in trading. I know this is a very broad and intense debate and that many people are not quite agree, but I plan to present new ways of seeing trading that might be interesting for some.
That said, we will introduce a new way of looking at trading systems. And for this let us first reformulate their classification. Typically, systems classify roughly in the classical groups of trend followers, countertrend and pattern-based. But today I propose a completely different alternative to any other classification that you may have seen.
Considering how a system is developed, we can establish two groups of trading systems: trading systems based on models and trading systems based on exploitation (mining) of data. We will describe these groups in detail.
Model-based Systems
A system based on a model is based on the idea of modeling a market inefficiency, which may be based on trading psychology, macroeconomic data, market microstructure, or any other force affecting the price. The presence of an inefficiency causes an anomaly or a particular pattern in price action that may be exploitable by a trading algorithm when the detected anomaly generates predictable results.Within this group we have the trend followers systems, systems based on reversions to the mean, systems based in cycles, statistical arbitrage between markets and seasonal patterns. This group includes trading systems that apparently are different, but they actually have a basic development process quite similar.
What is the problem with this type of trading systems? Really the same that we have in many scientific fields: the model is not reality, but a simplified representation of reality, so that its validity can only be established by the effects that occur in the price curve. In other words, the usefulness of this method depends on the significance and long-term stability of the anomalies detected, requiring perform full backtests of such systems.
Trading systems based on exploitation (mining) of data
If in the previous case, we established a model a priori, in the case of systems based on data mining we work without restrictions of any kind, looking for patterns in price that can be adjusted to an algorithm. That is, here we are not interested to know what market forces cause the behavior, we simply consider that these patterns are there and will continue to be repeated in the future. It is in this group where are included most new methods that are currently used in trading and are virtually unknown, if not ignored, by the general public. Here we can find the candlestick patterns analysis, linear regression, correlation, methods of machine learning, neural networks, support vector machines, decision trees, clustering by K-means and even trial and error methods based ontechnical analysis.
The good news about using such methods is that they are not widespread (this situation will probably give us an advantage over a lot of other traders competing in the market), and require no prior assumptions about price action.
On the other hand, it is highly likely that the trader will find a rich collection of random patterns on which to build losing strategies, because it is very difficult to distinguish real patterns caused by market inefficiencies from patterns that are there by chance. Even using advanced verification test as White Reality Check is very difficult to obtain winning systems using such methods.
The key is market inefficiency
In both types of trading systems we mentioned at some point the word anomaly, pattern or inefficiency. And obviously, before analyzing in detail the two groups of systems, we must think about whether the inefficiencies really exist in the markets. We can even think that inefficiencies really do exist but there are also better informed traders with better technology and resources who already are exploiting market inefficiencies which makes impossible for other traders to take advantage of these phenomena (a clear example of this would be companies specialized in high frequency trading, where dependence on high tech is total, which is an insurmountable barrier entrance).
This brings to mind the classic Efficient Market Hypothesis, a theory from academia pointing, in its most restrictive form, that prices reflect full and instantly all available information, be it historical, public or private, so that no method of studying the information available will achieve extraordinary returns. In other words, if the market is efficient, it will only be possible to “beat the market” through luck or chance.
But if you look at the definition of efficient market it is implicit the idea of transmission rate of information. It is clear that everyone does not get the market information at the same pace and with the same quality so we can asume that the efficiency in the market will depend on the ability to process information by the investor . And here is where possibly breaks the hypothesis established by Eugene Fama in the seventies. The question is whether, today, with all the means available the retail traders can get ahead of other competitors in the field of trading.What is clear is that markets, today, are not efficient: we can simply perform some analysis of performance of different indices, currencies, bonds, etc. to conclude that markets do not follow a random walk process. And all this without taking into account the bubbles, the impact of unexpected news (the most recent example is the Brexit) or simple the irrationality of some price movements, which often seems like the real economy has decoupled from financial markets as these sometimes can behave in strange ways. Never question because the price does this or that, or seek rational explanations in financial theory, often it works do the opposite of what the theory have taught us.
In fact, if we analyze the movements of the financial markets we can determine some common guidelines:
Markets react immediately and violently to the news, with clear directional movements.
By contrast, the markets move very slowly and very little when the information handled is subtle. In these cases, it may take some time, even years, to detect new inefficiencies that can be exploited.
Generally simple trading methods work best. More complex strategies are very demanding in terms of capital and execution, which requires having very powerful technology that is usually only accessible to a small group of participants.
In return, systems based on very obvious inefficiencies can be very profitable but its life cycle is very short.
