Event Intelligence
EventTransmission.
Measure the reach, strength, and timing of an event’s impact.
Follow how an event affects connected markets, industries, and economic conditions. Measure its reach, strength, and delay as new information changes the scenario outlook.
Nodes represent illustrative economic objects and states. A flash marks arriving information; scenario probabilities revise as evidence accumulates. Faint bars show the estimate from 0.6 seconds earlier, with changes in percentage points (pp). Paths, values, and outcomes are illustrative.
Impact responseMethod
Where does the impact reach?
An event can affect supply, costs, demand, and expectations at different times. Its impact can strengthen, fade, or take another path as new information becomes available.
Event Transmission follows these relationships through algorithms that measure the reach, strength, and delay of an event’s impact. Scenario estimates are revised as observations accumulate.
Reach
Identify the markets, industries, and economic conditions an event may affect.
Strength
Compare the intensity of the effects along different paths.
Delay
Measure how long each response takes to arrive and how it unfolds over time.
In practice
After a new announcement
What responds first, where is the effect strongest, and how long does it last?
Start with a financial announcement and follow the connected observations. Compare the order of their responses, the size of the change, and its duration.
Model Studies
How information changes a calculation.
Two candidate methods, explored through small numerical examples. These studies illustrate the mathematics; they do not specify the architecture of Event Transmission.
Message passing
Information from neighbours.
A node combines its own state with information from its neighbours. Change one neighbour’s value to see how the updated state responds.
mᵥ = mean(h₁, h₂, h₃)
h′ᵥ = ReLU(0.5hᵥ + 1.2mᵥ)
One simplified GNN layer with fixed example weights. ReLU keeps positive values and sets negative values to zero.
Evidence & probability
Revise a scenario estimate.
Start with three alternative scenarios. Choose an observation and compare its likelihood under each scenario to update their probabilities.
P(Sᵢ | E) = P(E | Sᵢ) P(Sᵢ) / P(E)
P(E) = Σⱼ P(E | Sⱼ) P(Sⱼ)
The scenarios are mutually exclusive and cover all outcomes in this example. Each observation starts from the same prior; all values are illustrative.
Data & refinement
Improved through observation.
We collect event and response data to refine the connections, transmission measures, and timing estimates used by the instrument.