Research & Evidence
PMRA Systems publishes engineering results when they are supported by primary artifacts, defined measurement scope and stated limitations.
Component benchmarks remain component benchmarks, and historical measurements are clearly distinguished from current runtime evidence.
Detailed technical evidence is available for qualified review.
Methodology
Research hypotheses are documented before evaluation.
Dataset integrity, completeness and timing semantics are validated before analysis.
Market effects are separated from transport and execution effects.
Results are evaluated on data outside the fitting window.
Execution models account for queue position, fees and cancellation behavior.
Costs and economic effects are attributed by component.
Inputs, methods and classification criteria are reproducible.
Simulation, shadow observation and runtime records are treated as distinct evidence classes.
Evidence cards
Historical benchmark environments are reported separately from current runtime measurements.
Research families
Book dynamics, trade flow, spread and liquidity state.
Fill quality, slippage and execution-cost measurement.
How prices form and propagate across venues.
Depth formation, replenishment and withdrawal.
Post-execution price behavior associated with adverse selection.
Reproducible processing, replay and reconstruction.
Out-of-sample testing and evidence gates.
Controls that act while the system runs.
Decision-to-fill measurement and transaction cost analysis.
FPGA-assisted processing and precision timing.