Loading market view
Scanning European outage signalsA single Spanish publisher tops the 30-day UMM latency league with an 88,270-minute median lag and 100% of records >120 min late — but it rests on just 5 backfilled records from an IIP operator EIC. The genuine disclosure tail belongs to the high-volume generators: Iberdrola and Endesa each carry P90 lags of 378 and 225 minutes across 67-87 records, with 12-14% published more than two hours late.
Data source notice: This analysis is based on publicly available ENTSO-E Transparency Platform data. Latency patterns described here are observations from public disclosure records and do not constitute regulatory findings or determinations of wrongdoing.
Spain's 30-day UMM latency league is topped by a publisher posting an 88,270-minute median publication lag with 100% of its records more than 120 minutes late — a rank built on just 5 records, every one flagged as synthetic backfill of an IIP operator EIC. Strip the artefact and the league median is 0.0 minutes across nine publishers. The genuine disclosure tail belongs to the two highest-volume generators: Iberdrola (P90 378 minutes, 11.9% late over 67 records) and Endesa (P90 225 minutes, 13.8% late over 87 records).
Ranked on raw numbers, the latency league has an unambiguous leader: a Spanish operator EIC showing an 88,270-minute median publication lag, with 100% of its records published more than 120 minutes after the event. Against a league median of 0.0 minutes across the nine publishers in the 30-day window (2026-05-18 to 2026-06-17), that looks like a standout disclosure problem.
It is not. The sample behind it is five records — and the entity profile flags every one as synthetic backfill of an IIP operator EIC. The lag measures how those rows entered the book, not how a live publisher behaves under event pressure.
Machine-readable sidecar (JSON) - confidence n/a · 0/0 claims backed
| 88,270 |
| 88,273 |
| 100.0% |
| 5 |
| A00117961.ES | 0.5 | 45 | 3.6% | 110 |
| IBERDROLA ENERGIA ESPAÑA, S.A.U. | 0.0 | 378 | 11.9% | 67 |
| ENDESA GENERACION SA | 0.0 | 225 | 13.8% | 87 |
| A0018305D.ES | 0.0 | 8 | 3.0% | 33 |
| Gas Natural Comercializadora, S.A. | 0.0 | 4 | 6.3% | 16 |
| A00036658.ES | 0.0 | 3 | 0.0% | 25 |
| A00035684.ES | 0.0 | 0 | 0.0% | 5 |
| Unattributed | 0.0 | 46 | 6.3% | 302 |
The contrast is the point. Rank 9 carries 88,270 minutes on five rows; the other publishers sit on samples of 110, 87 and 67 records with medians at or near zero. A league table sorted on a single statistic floats the backfilled EIC to the top every time — which is precisely why the headline rank needs to be read against its provenance before it is read as a signal.
The size of the gap is itself the tell. The subject's P90 of 88,273 minutes is 1,936x the league P90 median of 45.6 minutes, and accounts for 98.8% of the total P90 mass across all nine publishers. A live disclosure failure does not concentrate that completely in one entity; a five-row backfill against samples of 67 to 110 rows elsewhere does.
The entity profile carries a confidence of 0.7 with a confidence_basis recorded as iip_umm_participant_eic_backfill. That basis is the explanation: the records were loaded retrospectively, so their timestamps reflect the backfill mechanism rather than the interval between an event and its disclosure. A multiple this extreme, resting on a sample this small with a provenance flag this explicit, is a value to discount. Set the entry aside and read what the remaining publishers show.
Remove the backfilled EIC and the centre of the distribution effectively disappears. Seven of the nine publishers carry a 0.0-minute median, and the highest non-artefact median is 0.5 minutes — a publisher with 110 records. Most Spanish UMMs publish near-instantly.
That collapse reframes the question. If the median is zero, the disclosure story sits not in the centre of the distribution but in its upper tail: the minority of records that arrive late, and which publishers carry them. The relevant statistic is no longer the median lag but P90 — the level the slowest tenth of records reaches — and where that tail concentrates by volume.
The tail has two protagonists, and both publish at high volume. Iberdrola shows a P90 of 378 minutes with 11.9% of its 67 records more than 120 minutes late. Endesa shows a P90 of 225 minutes with 13.8% of its 87 records late. Those P90 levels run 8.3x and 4.9x the league P90 median of 45.6 minutes respectively — and unlike the headline leader, they rest on real, high-volume samples.
This is the volume-weighted distinction that matters. A 12-14% late-share on samples of 67 and 87 records reaches far more events than a 100% late-share on five backfilled rows. Because these are the two highest-volume generators in the set, their tail records map onto the largest share of actual disclosed events — the disclosures most likely to be consequential for anyone tracking Spanish generation and interconnection. Near-zero medians sit alongside hundred-minute P90s: the signature of publishers that are fast almost all the time but carry a meaningful slow tail on the events that slip.
The outage record grounds the pattern in a concrete event. The ES-FR interconnector asset (1,965 MW) carried a 30-day outage published with a 2,115-minute delay and accumulated 12,130 revisions — a single event absorbing both a multi-hour publication lag and a heavy revision load.
Against the rest of the book, it is the exception. Most large Spanish events publish at or before their start time, registering negative delays — disclosure that lands ahead of the interval it describes. Late publication is not the norm across the Spanish book; it concentrates in specific publishers and specific events, consistent with the P90 tail rather than any broad drift in disclosure timing. The mechanism is congestion and revision pressure on individual high-impact assets, not a system-wide lag.
Track whether the backfilled operator EIC accumulates enough live, non-backfill records to either confirm or clear its rank — until it does, treat its 88,270-minute median as provenance, not behaviour, and read past it to the next entries in the table.
Watch the Iberdrola and Endesa P90 tail across the next 30-day window: whether the 378-minute and 225-minute P90 levels hold, rise or compress, and whether the 11.9% and 13.8% late-shares move on samples of this size. For surveillance and trading desks, the watch-item is that volume-weighted tail of the active generators, not the headline rank. The rank tells you how the league sorts; the tail tells you which late records actually land on the events you cover — and that is where the position-relevant signal sits.