REMIT prohibits market manipulation and insider trading in wholesale energy, and requires market participants to monitor for both. TradingSurv produces the evidence an analyst needs to decide whether an hour deserves a closer look, and is explicit about what that evidence does not establish.
Regulation (EU) No 1227/2011 on wholesale energy market integrity and transparency (REMIT) prohibits market manipulation and attempted manipulation (Article 5) and the misuse of inside information (Article 3), and obliges participants to publish inside information (Article 4). REMIT II, Regulation (EU) 2024/1106, extended the regime and strengthened ACER’s role. The UK operates a retained equivalent supervised by Ofgem.
The obligation is continuous. It is not discharged by an annual review, and it is not discharged by a tool that produces alerts nobody can explain the derivation of. A surveillance function has to be able to say, about a specific delivery hour, what was observed, what was expected, and how far apart the two were.
Each screen below maps to a REMIT typology and runs against a reconstructed merit order for every delivery hour in the covered bidding zones.
Article 5: market manipulation
The cleared price for a delivery hour sits materially above or below the price a merit order reconstructed from published fundamentals would have produced, after accounting for the model's own calibrated error.
Article 5: market manipulation
Capacity that the reconstructed merit order places in-merit at the cleared price is reported unavailable or simply absent from the outturn, without a published outage explaining it.
Article 5: market manipulation
A participant's capacity is arithmetically necessary to meet demand in the hour, so the price cannot clear without it. Pivotality is a structural fact, not a finding about conduct.
Article 5: market manipulation
The shape of the aggregated supply curve around the clearing point implies that a small volume shift would have moved the price by a large amount.
Article 5: market manipulation
The blocks setting the price in an hour are unusually concentrated relative to the same hour across the calibration window.
Article 5: cross-border manipulation
Prices across an interconnected border diverge in a direction the available coupling would normally have arbitraged away.
Every output of every screen is an indicator for human review. It is not a determination that manipulation occurred, and the product does not make one. A price that a reconstruction cannot explain is a price the reconstruction cannot explain: the missing explanation is very often a fundamental the model does not see: an unpublished outage, a cross-border constraint, a fuel or carbon position, a hedge being unwound.
This distinction is load-bearing rather than cautious. A screen that asserted abuse would be wrong most of the time it fired, and a surveillance function that escalated on it would burn its credibility with its own regulator. So every finding carries the model error measured on the calibration window, and a finding inside that error is reported as inside it.
A merit-order model fitted on the same data it is then evaluated against will report an error far smaller than the one it actually has, and every screen thresholded on that error will fire too often or not at all. TradingSurv calibrates walk-forward: parameters for a delivery day are fitted only on days before it, and the reported error is out-of-sample.
The consequence is visible and deliberate. Model error is shown next to the finding, not buried in a methodology note, and a zone with too little history to calibrate says so instead of returning a confident number.