Deployed · wholesale power market
Market Edge has forecast day-ahead prices continuously in production since April 2026, currently ingesting 83 parallel input streams.
Research
What we have measured, and what we argue from first principles.
Everything we publish carries one of two labels, so you can tell a measurement from an argument at a glance.
Production
Evidence from continual learning systems running in production.
Empirical
Experiments with numbers, reported with their method and limitations.
The CLS Forecasting Benchmark on the 15-Minute Market
With the same inputs and decision time, the CLS forecasting function earned more simulated battery revenue than TabPFN 2.5 over the first year of 15-minute trading on the German-Luxembourg day-ahead market: 91.4% of the perfect-foresight maximum against 89.8%. We present the results, methodology and limitations.
Forecasting Traffic Without an Offline Training Phase
A continuous learning system matches the mean absolute error of the graph deep-learning models on the standard traffic benchmark without an offline training phase, by learning as it runs
Learning the Shape of a Rule It Was Never Shown
A continuous learning system recovers hidden cost rules from the consequences of its own decisions, revealing the shape of an exemption, a cap, a surcharge band and a progressive kink.
Keeping a Learned Rule Current When Nobody Announces the Change
A continuous learning system re-derives a hidden cost structure five times as it changes, keeping its learned understanding current without being told when the changes happen or what the new rules are.
Learning a Hidden Cost Structure from Outcomes Alone
Across a 26-year simulation, a continuous learning system recovers a hidden 25% capital-gains tax schedule from outcomes alone, calibrated to within about 6% at every bounded level of gain.
The CLS Forecasting Benchmark: Continual Learning Measured in Euros
The CLS forecasting function delivers approximately 14% higher cumulative trading P&L than TabPFN over 2,097 trading days on the German-Luxembourg day-ahead market. We present the benchmark results, methodology and limitations.
Theoretical
First-principles arguments and definitional work, with no new data.
Show all 20 articles Show fewer
