Imbalance settlement is a regime an electricity market participant enters by choosing own responsibility for imbalance and concluding a contract with the settlement operator. The public interface of the XMtrade®/ISZO system meanwhile knows five types of evaluation of the same day, not one (OKTE, a. s.; documentation of the public API, endpoint SystemImbalance, retrieved on 23 August 2026).
This text reconstructs the sequence of evaluations and names the inputs to the financial security. It is not a guide to entering the regime, not advice on contracts, and it does not state the meaning of the individual fields of the response: the documentation of the interface does not explain them in words, so they are not interpreted here.
Who is a balance responsible party and what a balance group is
A balance responsible party is, in the glossary of the operator, an electricity market participant who has chosen the regime of own responsibility for imbalance and concluded a contract on imbalance settlement with the settlement operator. Entering the regime is therefore a choice, not an automatic consequence of connection, and it attaches to a contract, not to a metering point.
A balance group is a group of metering and feed-in points of market participants for which one common balance responsible party has taken over responsibility for imbalance, and it is identified by an identification number. A single metering point therefore need not have its own balance responsible party — it may belong to a group led by someone else, and in that case all the evaluations described below concern that group, not the individual point.
The time unit of the whole mechanism is the quarter-hour. The glossary defines it as a sequence of four consecutive stretches of 15 minutes within a trading hour, and the response of the interface for a single day holds exactly 96 such periods.
Five types of evaluation of one day
The documentation of the interface lists the parameter evaluationType as mandatory and enumerates five permitted values: preliminarydaily, regulardaily, decadal, monthly and final. Each of them returns a separate response for the same day, so the result depends on which stage is requested; omit the parameter and the interface returns no response at all.
| Type of evaluation | Evaluation date of the day 15. 7. 2026 | Lag |
|---|---|---|
| preliminarydaily | 16. 7. 2026 | 1 day |
| regulardaily | 17. 7. 2026 | 2 days |
| decadal | empty response | — |
| monthly | 12. 8. 2026 | 28 days |
| final | empty response | — |
The field lastEvaluationType carries the value monthly in all three non-empty responses, which matches the monthly evaluation being the last one as at 23 August 2026. The data come from the machine response of the SystemImbalance endpoint.
The lags between the stages are uneven. The first two arrive within two days of the day itself, the third only four weeks later; 26 days therefore pass between the second and the third stage, and during them the value from the second one stands. Anyone working with daily figures in the course of a month is working solely with regulardaily.
What moved between the evaluations and what did not
Comparing the three responses for the same day shows two groups of values. The first did not move once: the field isp holds 174,174 in the first period in all three cases and the field si holds −2,341; the sum of the field si across all 96 periods is −459,826 in all three cases.
The second group moved at every further evaluation. The field srec of the first period holds in turn −0,5174, −0,5171 and −0,4742, so its absolute size in the monthly evaluation is 8,35 % smaller than in the first. The field pi rose from 26,102 to 26,195 and finally to 26,718, that is by 2,36 %, and the field ni shifted from −27,773 through −27,868 to −29,078, that is by 4,70 % in absolute value.
What those fields mean, the documentation of the interface does not state in words — it holds only a sample response with their names. Only what is verifiable is therefore set out here: which values move between the stages and by how much.
Summed up in one sentence: the day 15 July 2026 had three valid evaluations as at 23 August 2026, the newest of them created 28 days after the day itself. Seven observed quantities split into three that stayed fixed and four that shifted, the date of evaluation among them.
Where imbalance settlement stops
It stops at what the system has evaluated for a given day, not at how many stages the documentation enumerates. The types decadal and final returned an empty list for the day 15 July 2026, so as at 23 August 2026 those stages did not exist for it; the field lastEvaluationType confirms that with the value monthly.
From that follows a practical consequence for planning: a value taken a day after the day is not final, and a value taken a month later need not be either. Anyone comparing two days must compare the same stage of evaluation, otherwise they are comparing different things.
The same consequence holds for year-on-year comparisons. If a day from a year ago is compared with a day from a week ago, the first is most probably at a higher stage of evaluation than the second, and the difference between them then also holds the difference between the stages. Which stage it is can be checked in the field lastEvaluationType of every response.
Financial security: five inputs, one of them monthly
Entering the regime has a financial side too. For the calculation of the financial security the operator names five quantities: OO, Zp, Po, Kg and Co. The quantity OO is the higher value out of the sum of absolute values of agreed offtakes and the sum of absolute values of agreed deliveries in the daily diagram of the balance responsible party, and Zp is the period for which the security is determined.
The quantity Po is the proportional imbalance of the party for the current month, set to three decimal places; for parties concluding a contract for the first time and for parties without a metering point it is set at 0,5, while its minimum value is 0,005. The difference between those two values is a hundredfold. Kg is the coefficient of financial guarantee for the credit group of the party and refers to the operating rules, as does Co.
Co is the only one of the five quantities the operator publishes as a monthly series; the others depend on the specific party or on the operating rules. The list of inputs sits on the page on financial security.
Of the five quantities, two are therefore specific to the party — OO and Kg — one is a period, that is Zp, one is monthly and party-specific at once, that is Po, and one is common to all and published, that is Co. Anyone wanting to estimate the order of magnitude of the security has to trace four of them outside the public pages.
That is precisely why the amount of the security cannot be derived from public data. What is published is the monthly Co series and the bounds of Po, that is the value 0,5 for a first contract and the minimum of 0,005; the other inputs are either contractual or refer to the operating rules. Any figure for a specific party would therefore be an estimate, and none is made here.
The Co series for 2026
The value of Co moves month by month through 2026 and the range is not small. The lowest is in February 2026 at 135,9620 €, the highest in March 2026 at 217,9750 €; the difference between them is 82,0130 €, that is 1,60-fold.
The other months of the year lie between those two values: January 169,6890 €, April 143,3970 €, May 155,8730 €, June 142,5080 €, July 143,4020 € and August 176,4910 €. The average over eight months is 160,6621 €, so the August value sits above it by 9,85 % and the February one below it. The series is on the Co development page together with an export.
The denominator behind each of these ratios is one monthly value of Co in euro, not the price of electricity and not the price of imbalance in the settlement itself. It is a quantity for the purposes of the financial security, as the operator itself labels it.
The difference between the outer months is meanwhile larger than the difference between neighbouring ones. March 2026 is 60,32 % above February 2026, while August is 23,07 % above July; between April and June the value holds in a narrow band from 142,5080 to 155,8730 €. The series therefore has no single direction, only pronounced one-month swings.
Conclusions
A day is evaluated in five stages, not once. For 15 July 2026, three were available as at 23 August 2026: preliminarydaily dated 16. 7. 2026, regulardaily dated 17. 7. 2026 and monthly dated 12. 8. 2026. The stages decadal and final returned an empty response.
Some values do not move between the stages. The field isp stayed at 174,174, the field si at −2,341 and the sum of si across 96 periods at −459,826 in all three responses. The fields srec, pi and ni moved at every stage instead.
Financial security rests on five quantities. OO, Zp, Po, Kg and Co, with Po set at 0,5 for a first contract and a minimum of 0,005 — a difference of a hundredfold. Of those, only Co has a publicly available monthly series.
What to watch next: whether the stages decadal and final appear for the day 15 July 2026 and by how much the values srec, pi and ni shift at them. Equally worth watching is whether the August value of Co at 176,4910 € holds through the following months or returns to the average of 160,6621 € — the terms used in this text are defined by the glossary of the operator.
Analysis, not personalised advice. Before a transaction decision, verify the tariff value or the wording of the regulation directly with the network operator or the regulator.