Unambiguous Description ≠ Unambiguous DecisionBitcoin / U.S. dollarBITSTAMP:BTCUSDBorbotLately, I have been thinking less about predicting the market’s next move and more about a fundamental question: What makes sense to automate in trading? To answer it, I had to go back to how the market itself is described. My first requirement became: Market movement should be described completely and unambiguously. Any market segment should be representable as a sequence of clearly defined elements, with no “gray areas.” Given the same rules, the description should not depend on the observer. But description alone is not enough. The next question is: When can an event be considered realized? Each element needs an unambiguous realization criterion: the event either satisfies it or it does not. Realization belongs to the market. Objectives and results belong to the operator. Once realization is defined, the next level is exploitation. A realization may continue, develop into a sequence, be lost, and later restored. So the object of analysis becomes the organization of realizations and losses over time. This gives us: market description → realization criterion → realization → continuation or loss → exploitation chain → organization of chains → exploitation regime. A regime remains valid while its defining sequential structure reproduces itself. When that structure breaks, the regime ends. This does not necessarily predict what comes next. But it tells us that the previous method of exploitation is no longer supported by the observed sequence. This leads to another distinction: An unambiguous market description does not imply an unambiguous trading decision. Two operators may see the same market state and still make different — yet internally consistent — decisions. The difference comes from operator choice: objectives, resources, acceptable risk, performance requirements, horizon, and participation criteria. Changing these parameters does not change the market or its description. It changes how the same market state can be exploited. This is why I am increasingly skeptical not about automated trading, but about a universal trading machine producing one “correct” decision regardless of the operator. I find another architecture more interesting — a control panel. The automated layer identifies market elements, realizations and losses, exploitation chains, their organization, and regime changes. The operator defines: objectives, resources, risk, performance norms, and participation criteria. The algorithm can then determine an admissible action for that configuration. So: the market is the same; its description is the same; its realization state is the same; but exploitation decisions may differ. Perhaps the most interesting boundary of trading automation is not asking a machine to find one universal “correct decision,” but separating the observable market state from the operator’s formalized objectives and constraints — and automating their interaction.