Anyone running a battery on the imbalance market makes a decision every quarter of an hour: charge, discharge or stand still. To support that, Elia publishes an imbalance forecast every minute, for the current and for the next quarter. We looked into how useful those forecasts really are. We compared the forecasts from 28 April to 24 September 2026 with the settled prices, and analysed the prices from 1 October 2025 to 24 September 2026. Together that is about 14,250 quarters.
First, the rules of the game
Imbalance is settled per quarter of an hour. When the Belgian grid is short, with too little power, the MIP applies: the price of the most expensive reserve Elia had to activate. When the grid is long, with too much power, the much lower MDP applies. So a forecast is mainly valuable when it picks the right side: short or long.
The side is usually right
At the start of the quarter, the current-quarter forecast picks the right side in 82% of quarters. After 5 minutes that is 88%, after 10 minutes 93%. The next-quarter forecast is at 75% a quarter of an hour ahead and at 81% one minute before the start.
| When the forecast is made | Side right | Avg. error | False long | False short |
|---|---|---|---|---|
| 15 min before the quarter | 75.0% | 103 MW | 15.3% | 9.6% |
| 8 min before the quarter | 78.5% | 91 MW | 13.9% | 7.6% |
| 1 min before the quarter | 81.4% | 80 MW | 12.1% | 6.5% |
| At the start of the quarter | 81.6% | 79 MW | 11.8% | 6.6% |
| 5 min into the quarter | 87.6% | 53 MW | 9.1% | 3.3% |
| 10 min into the quarter | 92.5% | 27 MW | 6.1% | 1.4% |
The first three rows are the next-quarter forecast, the last three the current-quarter forecast. False long: forecast long, but the quarter ended short. False short: the other way round.
How often the forecast side is right
% of quarters, by minute relative to the start of the quarter. Before the start this is the next-quarter forecast, after it the current-quarter forecast.
One caveat: before the quarter starts, the forecast is only 1 to 3 percentage points better than simply looking at the imbalance measured so far. Only within the quarter does it pull clearly ahead.
You can't see the price itself
Elia forecasts the imbalance, not the price. Quarters forecast short settle at €160 to €168/MWh on average, quarters forecast long at €72 to €79/MWh. But exactly how high the price gets depends on the bids on the balancing market, and those are not in the forecast. Converting the forecast into a price leaves you €54/MWh off on average. That is no better than taking the usual price for that hour. Even knowing the imbalance exactly still leaves you €48/MWh off.
The costliest mistake always goes the same way
The forecasts err towards long more often than towards short: false long happens 1.6 to 4.4 times as often as false short. And that is exactly the mistake that costs money. Anyone counting on a cheap long quarter that turns out short pays about €137/MWh on average, while a real long quarter settles at €59 to €72/MWh.
For the price, both forecasts are structurally about 15 MW too long. The probabilities Elia publishes also underestimate short: when the current-quarter forecast gives a 35% chance of short, 51% of those quarters actually end short. The good news is that this bias is stable from month to month, so you can correct for it.
Small imbalances are noise
Most misses happen in quarters with a small imbalance. Below 50 MW, which is 29% of quarters, the side is right in only 59 to 68% of cases: barely better than guessing. Above 100 MW it is 86 to 100%. A silver lining: with a small imbalance, MIP and MDP are also closest together, so a miss costs less than its frequency suggests.
The sun sets the daily rhythm
Imbalance price by hour of day
Settled price per quarter, October 2025 to September 2026.
- Midday is long. From 08:00 to 17:00 there is too much power on the grid, peaking at 13:00 (+118 MW, with only 26% of quarters short). The median price then drops to €55/MWh and a quarter of those quarters settle below zero.
- Evening and night are short. From 18:00 to 02:00, and again from 06:00 to 07:00. At 20:00, 72% of quarters are short. The evening peak, with a median of €123/MWh at 19:00, is the most expensive time of day.
- Weekends are longer and cheaper. A median price of €85/MWh against €101/MWh on weekdays, and 13.6% negative quarters against 6.3%.
- Prices rise towards autumn. The monthly median price doubled from €80/MWh in April to €160/MWh in September.
Median imbalance price by month
From April to September 2026 the median price doubled.
Within a quarter the imbalance is restless: in 53% of quarters it crosses zero at least once. Between quarters it is surprisingly steady: 81% of quarters land on the same side as the quarter before. The final price is only settled late: halfway through, a third of quarters are still more than €10/MWh away from their final price.
Extremes are rare. 8.3% of quarters were negative and 1.9% were above €300/MWh. On Easter Monday, 6 April, five quarters settled at the price floor of −€15,000/MWh. The highest price was €3,115/MWh, on 15 August. The alpha surcharge Elia adds when the imbalance is large applied in 37% of quarters, with a median of €6.60/MWh.
Source: Voltmasters' own analysis of the public imbalance data from Elia Open Data (datasets ODS136, ODS147, ODS161, ODS162, ODS169 and ODS134). Forecasts from 28 April to 24 September 2026, prices from 1 October 2025 to 24 September 2026.