Physical State Estimation
Overview
An Automatic Market Maker determines prices from the current state of the market.
For a cryptocurrency exchange, that state may simply be the balances held within a liquidity pool.
For a prediction market, it may be the outstanding positions held by traders.
Electricity systems are fundamentally different.
The state of an electricity market is inseparable from the state of the physical electricity network.
An electricity AMM must therefore estimate not only supply and demand, but also the condition of the infrastructure through which energy is transported.
Prices can only be as good as the state estimate on which they are based.
The Electricity System is a Feedback System
Electricity networks continuously evolve.
Consumers switch appliances on and off.
Renewable generation changes with the weather.
Electric vehicles begin charging.
Batteries charge and discharge.
Power flows change every second.
The purpose of the AMM is to continuously observe these changes and adjust prices accordingly.
This creates a closed-loop feedback system.
The physical system influences prices.
Prices influence participant behaviour.
Participant behaviour changes the physical system.
The process repeats continuously.
What Are We Trying to Estimate?
The objective is not simply to estimate electricity demand.
Instead, the AMM seeks to estimate the current operating condition of the electricity system.
This includes questions such as:
- Is renewable generation abundant or scarce?
- Where is network capacity becoming constrained?
- Which feeders have spare capacity?
- Which substations are approaching their thermal limits?
- Where is flexibility currently most valuable?
- How much spare headroom remains?
Collectively, these quantities describe the state of the electricity system.
Sources of Information
Modern electricity systems generate vast amounts of information.
The AMM combines multiple sources to estimate the current operating state.
Market Information
Market participants reveal valuable information through their actions.
Examples include:
- buy requests,
- sell offers,
- accepted transactions,
- cancelled transactions,
- flexible demand availability,
- battery state of charge,
- renewable generation forecasts,
- demand forecasts.
These describe the intentions of participants.
Physical Measurements
The network itself continuously reports its operating condition.
Examples include:
- voltage,
- frequency,
- current,
- power flow,
- transformer loading,
- feeder utilisation,
- thermal headroom,
- line congestion,
- switch status,
- breaker status.
These measurements describe the actual physical state of the electricity system.
Inferring Scarcity
Scarcity is not directly measured.
It must be inferred.
Suppose demand requests begin increasing while renewable generation remains unchanged.
Accepted transactions begin consuming available network capacity.
Transformer loading increases.
Voltages begin falling.
The probability of congestion increases.
Collectively, these observations indicate that the system is moving towards scarcity.
The pricing function should therefore gradually increase prices to discourage additional demand and encourage additional supply.
Inferring Abundance
The opposite situation also occurs.
Renewable generation increases.
Feeders become lightly loaded.
Voltages begin rising.
Export curtailment becomes more likely.
Rather than allowing renewable energy to be wasted, prices should encourage additional consumption.
Buy prices fall.
Flexible devices naturally begin consuming energy.
Batteries charge.
Electric vehicles charge.
Heat pumps increase output.
The surplus is absorbed before renewable generation must be curtailed.
Physical Measurements Become Economic Signals
Individual measurements rarely tell the whole story.
For example,
low voltage alone does not necessarily indicate scarcity.
Similarly,
high transformer loading may simply reflect temporary demand.
Instead, the AMM combines multiple measurements to estimate the underlying operating condition of the network.
Physical measurements therefore become economic signals.
Examples include:
| Physical Observation | Economic Interpretation |
|---|---|
| Rising transformer loading | Capacity becoming scarce |
| Persistent undervoltage | Demand exceeds local capability |
| Persistent overvoltage | Excess local generation |
| Congested feeder | Consumption should be discouraged |
| Lightly loaded feeder | Consumption can be encouraged |
| Renewable curtailment | Demand should increase |
The pricing function operates on these inferred conditions rather than on any single measurement.
State Estimation Rather Than Rule-Based Control
Traditional control systems often rely on predefined thresholds.
For example,
If voltage falls below a threshold, disconnect loads.
FP-AMM adopts a different philosophy.
Rather than reacting to individual threshold violations, the market continuously estimates the overall operating state of the electricity system.
Prices then evolve smoothly as the estimated state changes.
This allows participant behaviour to adapt gradually before hard operating limits are reached.
The objective is not to respond to failures.
The objective is to prevent them.
A Continuously Updating State
The operating state of the electricity system is never static.
Every accepted transaction changes:
- available generation,
- available network capacity,
- power flows,
- voltages,
- future flexibility.
The AMM therefore updates its estimate continuously.
Each new transaction slightly changes the estimated state of the network.
That updated state produces new prices.
Those prices influence the next transaction.
The market therefore evolves continuously rather than through a sequence of isolated market clearing events.
State Estimation as the Foundation of Pricing
The defining characteristic of FP-AMM is that prices emerge from an estimate of the physical state of the electricity system.
Scarcity is not declared administratively.
Congestion is not resolved after the market clears.
Instead, both are continuously inferred from market activity and physical measurements.
The resulting state estimate becomes the input to the pricing function, allowing economic incentives to remain aligned with the real-time operating condition of the electricity network.
The following section explains how this continuously updated state enables the market to clear continuously, allowing prices and accepted transactions to evolve together as the network changes.