生理術語太陌生，看不懂。 About Artificial Brain: AB is an implementation of two ingenious philosophic and neurophysiologic ideas of a Great Russian scientist P.K.Anokhin. The first one is “Outstripping reflection” theory that postulates an ability of any live form to predict future biologically important event in an environment by the signals extracted from the sequence of events that happens long before the bio-important event, based on time-space continuity and repeatability that allows learning by the trial/error method. The second idea is the “Functional system” (FS) theory that defines the universal architecture of live in terms of self-organizing, necessities driven, dynamically forming and result oriented organization of smallest units, that in turn also realize this architecture on any level down. I’ve translate and expand the original neurophysiological architecture into cybernetic one, expanding the feedback loop with interchangeable samples to compare with and associative memory apparatuses, that contain all learning and practical experience of FS. From the human stand point interchangeable samples are images of desire results that could satisfy dominant necessity. In terms of trading they are, for example, price levels to enter/exit market or combination of long/short term price direction that stimulate an execution of an appropriate trading action. The profit is the ultimate desire result of an entire system that directs learning and trading processes. Functional system Architecture Trader Station Architecture Tick data is being analyzed by the two receptor systems: “Market Forces” (MF) and “Market Spectrum” (MS). MF measures and encodes tick data in to code word. The idea is to calculate price reaction to the particular sequence of market forces. Because market is inertion system it allows a prognosis of a reaction before actual price moves and predicts expected price limits and short term price direction (trend). MS receptor analyses market dynamics in different “time frames” and generates second code word. That allows identifying long term trend. Both code words are combined in “Encoder” that also adds “Motivational” and “Internal State” code words in to machine word that is used by all associative memory apparatuses to identify market state. Market Forces code word is an input for Price predictor that is some sort of second level receptor system that generates expected prices and short trend direction. This system utilizes context-driven Markov chain. To improve short/long trend identification these directions go to Short/Long trend Learners that implement downgrades and simplified FS that leans to adjust original directions according their experience in different market situations, instead of just mathematical results. The resulting signals and limits are the basis for Receptor Functional system (RFS) that is the fully functional FS that learns to generate managing signals for Motor Functional system (MFS). The simple common sense hard coded trading recommendations direct initial cognitive rough-research reaction (CRRR) on the market changes. FS use CRRR when it does not know what to do… It performs hard coded recommendation and observes result, extracting during experiment more or less helpful information. In our case this is Maximum/Minimum price and appropriate market conditions to try Sell or Buy next time when these conditions will appears. Successful result will be associated with the performed action and market condition. Unsuccessful result will invert the initial trading action. The FS has two attempts to test its actions. If both attempts failed than given market condition is marked as worthless and system will do nothing if it happens. The resulting Buy/Sell/Close signals and Short/Long trend go to Motor Functional System for actual execution after evaluation in the context of real market dynamics. MFS learns execute signals using trading limitations and order execution features and strategies. Only this part is the place for developer to intervene and bring his subjective opinion in to the trading result. But, the biggest advantage of the Artificial Brain is that it wills CORRECT DEVELOPER’S MISTACES and produce positive result anyway! http://trader-assistant.com/ab.htm http://trader-assistant.com/how.htm http://en.wikipedia.org/wiki/Pyotr_Anokhin