Data Dictionary
Input formats, output columns, and field definitions for the Decision Machine platform.
Input Time-Series CSV
SML-App accepts time-series data as CSV files uploaded through the Time-Series Views panel. Each file represents one or more time-series to be analysed.
| Rule | Detail |
|---|---|
| First column | Date in YYYY-MM-DD format — required on every data row |
| Remaining columns | Numeric time-series values — at least one column required, maximum of 25 |
| Header row | Required — first row must contain column names |
| Minimum rows | At least fifty rows beyond the header |
| File type | .csv only — .xlsx and other formats are rejected |
| Encoding | UTF-8 or UTF-8 BOM |
SML-App validates the following on upload:
| Check | Detail |
|---|---|
| Date column present | All values in column 1 must match YYYY-MM-DD |
| Numeric columns present | At least one column beyond the date column is required, with a maximum of 25 columns |
| Minimum length | At least 50 data rows required |
| File type | Only .csv accepted at upload |
Binary Output
Triggered by submitting a Binary job from the SML Measurements panel. Uses the config_ prefix file. Machine learned measurement dynamics in a generally non-equilibrium setting — any time-series of values can be equally well processed.
Time-series identifier derived from the input filename.
Date of the measurement, corresponding to the input date column.
The raw measured value from the input time-series at this timestamp.
Computed, machine learned probability that the value will go up in the next time step. Computed exactly up to machine error.
The probability is a function of the system energy and process extent (Lagrange multiplier).
Computed, machine learned probability that the value will go down or stay the same in the next time step. Computed exactly up to machine error.
The probability is a function of the system energy and process extent (Lagrange multiplier).
Computed, machine learned system energy measured from equilibrium as a zero offset. Positive: Bull. Negative: Bear. Internally generated energy could include emotions (plural) — human-generated displacement energy from equilibrium.
At equilibrium, the probability that an asset price will go up equals the probability that it won't (an "unbiased coin"). Markets are driven by emotions which throw participants into non-equilibrium almost all of the time. The energy/emotions measurement quantifies how far away the data is from statistical equilibrium.
Rate of energy flow per time step. Combines Emotion and Resistance. At equilibrium, Power = Emotion² / R (V²/2R). Power is the energy available to perform work — "to move the needle".
Wherever there is energy available to move the needle, there is also resistance. The more resistance, the harder it is to move the needle. Resistance is not constant in systems that are not in equilibrium.
Computed, machine learned market (Nyquist) noise that dissipates system energy so that it cannot be used for price movement. Power can be wasted through dissipation (strain or viscosity), making it unavailable to do work. As noise increases, the amount of wasted power increases.
The entropic temperature of the system. Temperature is the reciprocal of the derivative of entropy with respect to energy — the general definition of entropic temperature.
In non-equilibrium dynamics, free energy and temperature are coupled to produce a heat engine. By observing free energy and temperature together, price entry and exit points can be identified. Recommended to plot as a double-sided plot alongside free energy.
Helmholtz free energy (F = E − TS). The total energy/emotion available to do useful work. Minimum Free Energy is a more convenient form of maximum entropy in non-equilibrium problems.
When free energy decreases it does work in the dominant emotion (bull or bear). Local maxima indicate when energy is available for price movement — an entry point. After the price movement a local minimum develops, equivalent to maximum entropy. The stable minimum signifies an exit point for the dominant emotion trade.
Computed, machine learned probability that the value will go up in the next time step when in a thermal bath of temperature T_R. Where thermal probabilities dominate over p+ and p-, thermal probabilities drive price movement.
The thermal probabilities depend on the temperature difference between the system temperature and the reservoir temperature T_R. Decision Machine uses a default value for T_R at statistical equilibrium (T_R = e/4) but it can and should be set by the customer in the configuration file.
A dissipative system is a thermodynamically open system operating out of, and often far from, statistical or thermal equilibrium, exchanging energy and information with its environment.
Computed, machine learned probability that the value will not go up in the next time step when in a thermal bath of temperature T_R. See therm_p+ for full description of thermal probabilities and dissipative systems.
Units Output
Triggered by submitting a Units job from the SML Measurements panel. Uses the measure_ prefix file. Scientific measurements of time-series dynamics based on the science of counting — counting units of measure with constraints on maximum entropy.
Time-series identifier derived from the input filename.
Date of the measurement, corresponding to the input date column.
The raw measured value from the input time-series at this timestamp.
Momentum measures the inertia of the time-series. As momentum increases, greater force is needed to produce a given deviation from the current direction.
When displacement velocity vanishes (v = 0), momentum equals the mass (p = m). When E ≠ 0, p = m·exp(v). Note: momentum is non-linear and not equal to mv.
Dispersion relations define the algebraic relationships between mass, momentum, and displacement energy measured from equilibrium.
The energy at equilibrium is subtracted from the total system energy to give the displacement energy. When E = 0, the system is in equilibrium — all states are equally likely, with no energy entering or exiting the system. When E ≠ 0, energy enters (E > 0) or exits (E < 0). Non-zero displacement energy signals non-equilibrium behavior.
The entropic temperature of the system.
Body temperature of the system — same as the reservoir temperature. Thermodynamic expressions define relationships between E, free energy, T, free entropy, and T_B through the second law: ⟨F⟩ ≡ T⟨A⟩/T_B = E − TS, offset from equilibrium.
When in thermal equilibrium, T = T_B and classical thermodynamics is regained. Equilibrium is not assumed and is infrequently observed in time-series.
Expected Supply ⟨ξ⟩. Combines the expected count and the expected strain (energy stored in extension). Defined by the affine connection as part of Natural Dynamics.
Expected strain component of the supply measurement.
Expected Demand ⟨η⟩. The expected value minus the expected strain — an extension that stores energy. Defined by the affine connection as part of Natural Dynamics.
Sustained supply count.
Sustained strain component.
Sustained displacement energy.
Sustained demand.
Variance of supply count.
Variance of strain.
Variance of displacement energy.
Variance of change in supply count.
Covariance of supply count and strain.
Free energy: ⟨F⟩ ≡ T⟨A⟩/T_B = E − TS, offset from equilibrium.
Entropy: S(λ,E) — the number of different counting configurations possible when subjected to the constraints of counting the time-series. A function of the Lagrange multiplier λ and displacement energy E. In general, S and entropic temperature 1/T = ∂S/∂E are not constant.