When there is more wind in Scotland than the cables can carry, the grid operator pays wind farms to switch off and other stations to switch on. This page shows what that cost each day and who got the money, worked out from the official published data with a method locked before the data was looked at. The latest day shown is 22 Sep 2026.
| Day | Paid out | To wind farms, for switching off | To gas stations, for switching on | Everything else | Price paid to gas, £/MWh | Above the market price by | Spent outside the main market | Grid operator's own figure | Official figure vs the headline number |
|---|---|---|---|---|---|---|---|---|---|
| £3.95m | £0.11m (2.9%) ‡ | £3.17m (80.3%) | £0.67m (16.9%) | — | — | £0.86m (21.8%) | £2.91m (published 26 Sep 2026) | 0.89× the headline number | |
| £6.56m | £0.14m (2.2%) ‡ | £5.28m (80.5%) | £1.14m (17.3%) | £246.38 | £73.87 | £0.40m (6.1%) | £4.31m (published 26 Sep 2026) | 0.79× the headline number | |
| £22.11m | £5.37m (24.3%) | £14.31m (64.7%) | £2.43m (11.0%) | £206.10 | £156.46 | £0.52m (2.4%) | £23.04m (published 26 Sep 2026) | 1.17× the headline number | |
| £29.32m | £6.82m (23.3%) | £18.37m (62.7%) | £4.12m (14.1%) | £223.54 | £182.43 | £0.96m (3.3%) | £31.83m (published 26 Sep 2026) | 1.26× the headline number | |
| £25.29m | £4.06m (16.0%) | £17.49m (69.2%) | £3.74m (14.8%) | £224.45 | £139.77 | £1.42m (5.6%) | £28.36m (published 26 Sep 2026) | 1.32× the headline number | |
| £28.81m | £4.09m (14.2%) | £20.46m (71.0%) | £4.27m (14.8%) | £229.79 | £111.49 | £1.68m (5.8%) | £29.29m (published 26 Sep 2026) | 1.19× the headline number | |
| £9.33m | £0.50m (5.4%) | £8.50m (91.1%) | £0.33m (3.6%) | £239.37 | £88.84 | £1.50m (16.1%) | £8.19m (published 21 Sep 2026) | 0.91× the headline number | |
| £11.52m | £0.44m (3.8%) ‡ | £9.89m (85.9%) | £1.19m (10.3%) | £239.22 | £90.96 | £1.35m (11.8%) | £8.88m (published 21 Sep 2026) | 0.86× the headline number | |
| £6.71m | £0.01m (0.2%) ‡ | £6.60m (98.4%) | £0.09m (1.4%) | £240.23 | £61.28 | £2.61m (38.9%) | £3.01m (published 21 Sep 2026) | 0.45× the headline number | |
| £4.28m | £0.00m (0.0%) ‡ | £3.86m (90.3%) | £0.41m (9.7%) | — | — | £0.00m (0.0%) | £0.38m (published 21 Sep 2026) | 0.10× the headline number | |
| £23.54m | £2.80m (11.9%) | £17.68m (75.1%) | £3.07m (13.0%) | £228.22 | £120.93 | £0.00m (0.0%) | £21.91m (published 26 Sep 2026) | 1.07× the headline number | |
| £9.78m | £0.19m (2.0%) ‡ | £7.72m (78.9%) | £1.87m (19.1%) | £252.39 | £83.61 | £2.11m (21.5%) | £8.33m (published 21 Sep 2026) | 1.05× the headline number | |
| £15.13m | £0.49m (3.2%) ‡ | £13.90m (91.9%) | £0.74m (4.9%) | £229.81 | £93.43 | £2.25m (14.8%) | £12.69m (published 21 Sep 2026) | 0.88× the headline number | |
| £8.13m | £0.17m (2.0%) ‡ | £7.85m (96.5%) | £0.12m (1.5%) | £224.54 | £73.89 | £0.25m (3.0%) | £4.13m (published 15 Sep 2026) | 0.52× the headline number | |
| seed | £33.61m | £3.77m (11.2%) | £26.82m (79.8%) | £3.02m (9.0%) | £229.53 † | £82.64 | £3.33m (9.9%) | £32.67m (published 15 Sep 2026) | 1.07× the headline number |
| seed | £19.75m | £1.49m (7.6%) | £16.99m (86.0%) | £1.26m (6.4%) | — | — | — | £19.83m (published 21 Sep 2026) | 1.07× the headline number |
| seed | £19.34m | £2.03m (10.5%) | £14.84m (76.8%) | £2.47m (12.8%) | — | — | — | £18.13m (published 15 Sep 2026) | 1.08× the headline number |
| seed | £17.51m | £3.84m (21.9%) | £9.07m (51.8%) | £4.61m (26.3%) | — | — | — | £17.02m (published 15 Sep 2026) | 1.32× the headline number |
| seed | £30.05m | £5.39m (18.0%) | £18.68m (62.2%) | £5.98m (19.9%) | £201.77 † | £110.65 | £1.46m (4.9%) | £31.66m (published 15 Sep 2026) | 1.32× the headline number |
| seed | £15.79m | £1.88m (11.9%) | £12.57m (79.6%) | £1.34m (8.5%) | — | — | — | £14.54m (published 15 Sep 2026) | 1.01× the headline number |
| seed | £4.76m | £0.04m (0.9%) ‡ | £4.52m (94.8%) | £0.20m (4.3%) | — | — | — | £2.19m (published 15 Sep 2026) | 0.48× the headline number |
| seed | £12.73m | £1.44m (11.3%) | £9.65m (75.8%) | £1.63m (12.8%) | — | — | — | £13.14m (published 15 Sep 2026) | 1.18× the headline number |
Money in millions of pounds; percentages are shares of the day's paid-out total. Rows marked seed are the eight days of 1 to 8 September 2026, reconstructed by the earlier investigation that set this page's method; they set the bar for the first record and are never themselves flagged. † On seed rows the gas price uses that investigation's original volume rule, which agrees with the settlement totals for switching-on but not for switching-off; every later row uses the volume type that reconciles with the settlement totals, and if none does the price is left blank. Not yet populated means the grid operator's file for that day still holds placeholder rows. Not yet published means the grid operator's own attribution has not yet reached that day; it is added, with the date it appeared, when it does. The headline number is wind payments plus gas payments, which is what the daily trackers add up. A blank means the figure could not be computed from what is pinned, never that it is zero. No day has yet exceeded the seed rows' highest paid-out total.
‡ Wind figure ambiguous: on the marked days the check that payments to wind farms for switching off ran in the expected direction failed, so the figure has two readings; both are in the evidence file.
Propositions
Sealed and timestamped before any tracker day was fetched; each is decided by its instances and reported as it stands.
- T1 — on record days the Disaggregated BSAD net share of paid-out exceeds 5 %: undecided (0 instances). Falsifier date 2027-03-31.
- T2 — NESO's Constraints figure, when published, exceeds the two cuts (wind-unit bids + gas-unit offers paid) for the same day: fails (22 instances, 14 deciding). Falsifier date 2027-03-31.
- T3 — on record days the gas-offer premium over APXMIDP exceeds £50/MWh: undecided (0 instances). Falsifier date 2027-03-31.
- T4 — NESO's Constraints figure as a share of the two cuts rises with the wind share of accepted bid volume (Spearman rho >= 0.50 over at least 20 qualifying days from 2026-09-16): undecided (6 qualifying days; decided once on 2027-03-31). Sealed separately, SHA-256
0f3b5a96….
T2 failed on 9 September 2026.
Before any tracker day was fetched, the declaration stated that NESO's own Constraints figure for a day would come out larger than two cuts of the mechanism's payouts: money paid on bids to wind units (paid to reduce output) plus money paid on offers to gas units (paid to increase it). The first day that could decide it was 9 September. NESO's figure is £4.13m; the two cuts come to £8.0m, so NESO's figure is 51.5% of them. NESO's file covers all 48 settlement periods for that day and is identical across five daily versions (15 to 19 September), so this is not a revision still to come. The declaration says this outcome is the publication. This note is that publication.
What it means: the two cuts are not contained in what NESO calls constraints. Money paid to gas units to increase output buys more than constraint management: on 9 September at least £3.72m of the £7.85m, close to half, sits outside NESO's Constraints figure. Where NESO books it has not been checked here.
What it does not mean: it says nothing about whether NESO's figure is right, and nothing about the size of the bill. The paid-out column is unaffected.
A reading, not yet tested: the ratio seems to follow the wind. On the windiest seed days (4 and 5 September) NESO's figure was about 130% of the two cuts; on the calmest (2 and 9 September) about half. That pattern was noticed after the numbers were seen, so it proves nothing. It will be sealed as a new proposition before batch 2 is fetched on 26 September and decided by later days only. T2 stays marked failed.
Two cautions on the table. The sign check on wind-unit bids failed on six of seven tracker days, so the wind-bids column is ambiguous on those days; the verdict above does not depend on it, because gas offers alone (£7.85m) exceed NESO's figure. And BSAD for 12 and 13 September reads close to zero (£72.86 and £47.48); that is correct under the declared rule, but NESO may not have finished filling those days, so read them as provisional.
How the numbers are made
The rules that pick the days, split the money and decide what counts were written down and sealed before any day's data was fetched, and the sealed text is never edited afterwards. Paid out is the gross money paid to generators and other units in Great Britain's Balancing Mechanism on a settlement day, from Elexon's published indicative cashflows. To wind farms is money paid on bids by units registered as wind; to gas stations is money paid on offers by gas units; everything else is the remainder. The gas price pairs those pounds with the accepted volumes; the market price is the APX day-ahead index for the same half-hours. Spent outside the main market is the grid operator's published adjustment trading for the day. Grid operator's own figure is its published constraints cost, laid beside ours when it appears. Cashflows are indicative and pre-settlement, and nothing here is a saving, a loss, or an attribution to any cable or company.
Declaration (SHA-256 d20d59203e7b7fda…):
investigations/013-the-cover-price-tracker/DECLARATION.md.
Evidence, every row and every source digest: evidence/.
Code and tests: github.com/jordan-dimov/grid-mysteries.
Page generated 2026-09-26 from
evidence/tracker.json; the page is a pure function of that file.
Record: every day above stands in a governed record that admits a computed row once and refuses to change it (bill/); checkpoint at tree size 84, root sha256:7a8db97df0d566c4…, signed with key bill-2026 and countersigned by DigiCert; anchor bill/anchors/tree-84.json.
Corrections
None so far. When a figure on this page is corrected, the correction is listed here with its date and the old and new values; the old value is never silently overwritten.