That in the "black box"?


Thus, any program can find a pattern of a certain, more or less restricted class. If in fact there are other laws of nature, the program or do not find them, or its conclusions may be wrong.

An example of a monkey

Here is an example. Monkey sitting in a cage, opened the fundamental law of nature that if you push the red button, you get a banana, and if you press the green, you get an electric shock. However, it is possible that there is a pattern of a different order, by the plan of the experiment, designed to switch the buttons just as when the monkey was finally convinced of the correctness of its law. One day we will get a monkey a banana instead of an electric shock, and it does not mean the collapse of the whole of science.

Therefore, to use reasonable prediction program, you need to understand, first, what kind of patterns can detect the program, and secondly, what the laws are in the market - that is, again, to know stuffing <black box>.

Analysis of the data shows that at sufficiently long periods of time, there are simple relationships between some parameters of the market. But at some point in the nature of these relationships can qualitatively change. This does not mean that the operation of the market is not subject to any laws. Just these laws is more complex than it seems at first glance.

But let's say, a certain pattern is found. What's next? Classical science responds unequivocally: further it is necessary to understand the economic meaning of this law, that is to construct a model to realize the place was found in a series of similar laws. Without this it is impossible to establish the limits of applicability of the found correlations and ultimately can not make informed decisions. Max Born wrote about this: <The experiment does not mean anything until it is interpreted by the theory>. But he was referring to physical phenomena, where the possibilities of the experiment is much wider than in the economy and the empirical regularities can be much better grounded. More so not too much into the empirical regularities in economics. Otherwise we risk as the very ape and get a telling blow.

Point of view of the supporters of the method <Black Box> opposite. At one of the workshops were presented the following results. The dynamics of indicators characterizing the state of the Russian economy as a whole. Was considered a score of these indicators, and it turned out that the fractal dimension of the set, which describes the values of these indicators is four. In essence, this means that there are 16 non-trivial and independent relations between the indices. On the question of economic sense these relations was the answer that the mere posing questions to the author do not recognize. Relationship exists, and all.

But it is not clear in advance that they reflect: the objective economic law or a specific method of calculating the State Statistics Committee. In the first case found dependencies can be used without fear, and in the second they are valid only as long as the procedure remains the same.