AI-based forecasting and optimisation
Our machine learning models analyse consumption separately for each meter in your portfolio. Forecasts are optimised for the expected system direction (whether the system will be long or short) and sent to the day-ahead market (GÖP); revisions updated through the day go to the intraday market (GİP). The aim is to keep your imbalance cost as low as possible. The process runs automatically, from data collection to sending the revisions.
With the DemandCast day-ahead forecasting module (average)
Forecasting, system direction and intraday revisions combined
Forecasting, intraday revisions and submission run automatically
Baseline: the supplier's own forecast · 2026, live operation · Method and details
Market context
On 1 January 2026, the netting rate within balance responsible groups (DSG) fell from 10% to 5%. The penalty coefficient for the side that adds to the system imbalance rose from 3% to 6%. The same forecast error now costs more.
On 1 January 2027, balance responsible groups are expected to be removed or to have a reduced effect. Imbalance that is netted within the group today will then fall directly on the supplier. That cost will then depend on the supplier's own forecast quality.
We can calculate your post-DSG cost with your own data →DSG netting 10% · penalty coefficient 3% in both directions
DSG netting 5% · coefficient 6% on the side that adds to the system imbalance
DSG effect reduced further or removed · each supplier carries its own imbalance alone
Measured results
Six months of live operation. The baseline is the cost produced by the supplier's own forecast. The chart shows the effect of each layer separately.
Supplier's own forecastStarting point
Meter-level forecastDemandCast
+ Imbalance OptimisationMetaCast
+ Intraday revisionUpdates during the day
Reduction in imbalance cost with all three layers combined
2026, live operation. Index: imbalance cost under the supplier's own forecast = 100. Imbalance cost is calculated before DSG netting. *The cost effect in the intraday market depends on the supplier's trading strategy and algorithmic trading rules. Results vary with portfolio structure.
Products
The data and integration layer that feeds the forecasting models. It collects data from EPİAŞ (the market operator), OSOS (the distribution companies' meter reading system), organised industrial zones (OSB) and weather sources automatically, fills gaps and removes faulty readings. It sends forecasts to EPİAŞ and to trading platforms, and produces reports and dashboards. Included with the forecasting products.
Consumption forecasting that models each meter in your portfolio separately. It produces the next day's hourly forecast for the day-ahead market. Once the intraday market opens, it revises the forecast for each contract every hour with the latest data.
An optimisation layer that runs on top of the forecast. Forecast Optimisation corrects the forecasts you already use by learning from their past errors. Imbalance Optimisation predicts the system direction and adjusts the final forecast to minimise imbalance cost.
Benefits
Meter-level forecasting reduces imbalance volume. The remaining imbalance is positioned according to the system direction and revised during the day in the intraday market.
Data collection, forecasting, submission and reporting run automatically every day on a fixed schedule, including weekends and public holidays. This includes the hourly revision of every intraday contract. Your team monitors the process from the dashboard; trading decisions stay with the supplier.
OSOS data from different distribution regions is collected in one place. For OSB meters that are not in OSOS, data is pulled from the OSB portals. Forecast performance is tracked monthly and daily on the dashboard.
Getting started
The first four steps are free and carry no purchase commitment.
We sign a mutual non-disclosure agreement (NDA) before we see your data.
Through a read-only EPİAŞ user account that you set up for us, we access your meter list and consumption history. No installation is needed on your side.
We simulate a past period day by day under real operating conditions. This takes about a week once access is in place.
We go through a report with you that compares the results with your current forecast on accuracy and imbalance cost.
If the results make sense for you, we prepare a proposal based on your portfolio size and the modules you choose.
OSOS access is set up and the system is connected to your portfolio. The first month, during integration, is free.
Integrations
Forecasts are sent directly to EPİAŞ and to trading platforms such as SmartPulse and Senkron. They can also be received as Excel or CSV files or by email; performance is tracked on a web dashboard.
In the first meeting we talk through your portfolio and the backtest process.
Request a meeting