Spain Methodology

Modo Energy provides benchmark data for battery energy storage systems across global energy markets. Our Indices share a common framework, and the differences between markets are documented in each methodology.

1. Introduction

This document explains how Modo Energy calculates and constructs its battery energy storage Indices in Spain. It covers four things:

• the assumptions that define a representative asset;
• the components that make up published revenues;
• the optimisation model that simulates asset dispatch; and
• the data inputs and transformations that feed the simulation.

It also sets out how judgment is applied, how late or missing data is handled, how the methodology changes, and the main limitations of the Indices.

1.1 Why Spain requires a simulated index

The ME vBESS ES Indices differ from our GB Index in one fundamental respect: they are simulated, not measured.

In Great Britain, utility-scale batteries are registered as Balancing Mechanism Units (BMUs). BMU status comes with strict data-sharing obligations. This lets us compute revenues from publicly reported physical notifications, bid-offer acceptances, ancillary contracts, and settlement metered volumes. The GB Index reflects what a representative portfolio of real assets earned.

Spain has no equivalent data regime. Battery activity in the wholesale and ancillary markets is published only at programming-unit (UP) level, not per asset. Public BESS revenue disclosures are sparse and uneven across markets. The installed fleet is also still small enough that fleet-wide measurement would not be statistically representative.

To produce a benchmark under these conditions, we simulate the dispatch of a representative battery asset against publicly available market data. We use our Global Dispatch Model, a mixed-integer linear programme (MILP) that maximises battery revenues subject to market participation rules.

1.2 What the ME vBESS ES Indices represent

The ME vBESS ES Indices (together, "the Index") represent the simulated revenue performance of grid-scale lithium-ion battery energy storage systems in Spain. The family has three variants, one per system duration (1, 2 and 4 hours), so revenues can be compared across different types of utility-scale battery systems. The Indices reflect the revenues that a representative asset could have earned, not the revenues any real battery earned.

We use a simulation because per-asset revenue data is not public in Spain. Every input is published by the market operator (OMIE) or the system operator (REE, through ESIOS), and every parameter Modo Energy sets is listed in §2.2 and §4.1.

1.3 Definitions

2. Index Construction

2.1 The ME vBESS ES Indices

Modo Energy currently produces the following Indices for Spain:

All three Indices use the same published methodology. The only difference is the duration parameter of the representative asset. The 4-hour variant is the headline Index. Index values are available from 1 January 2022.

2.2 Representative BESS specification

The representative asset is specified as follows. Modo Energy sets these parameters, and the market owner maintains them (§4.1).

2.3 Index calculation methodology

We calculate each Index in four steps:

1. Run the Global Dispatch Model over the assessed period, using the representative asset in §2.2 with that Index's duration and the input data in §4.2.
2. Compute the simulated revenue of the representative asset in EUR, for each modelled market and direction.
3. Sum revenues per settlement period and divide by the rated power (50 MW). This gives an Index value in EUR/MW per settlement period.
4. Aggregate these values into monthly, quarterly and annual figures, as described in §2.4.

The calculation is rule-based. Where judgment is needed, it is applied as described in §4.3. No post-hoc calibration factor is applied (see §5.9).

2.4 Index value representation

All Index values are reported as revenue per unit of rated power. Revenue is market revenue net of the cost of charging energy, before operating costs, fees and taxes (§3.2). Index values can be represented in two units:

• EUR/MW/year (annualised): total Index revenue for the assessed period, divided by the number of days in the period, multiplied by 365.
• EUR/MW/hour: total Index revenue for the assessed period, divided by the number of hours in the period.

Index values do not scale linearly with rated power. Spanish ancillary and balancing markets have finite depth. A larger battery has to bid deeper into the merit order to clear the same share of its book, so per-MW revenue falls as size rises. In the model this applies to the ancillary and balancing markets (§5.6.1). Day-ahead revenue per MW does not change with size, and intraday trading is capped at traded volume (§5.3.2). The ME vBESS ES Indices report a 50 MW asset. §5.6.1 sets out how size is priced, and users who need a different size can build one as a custom index.

2.5 Publication cadence and revisions

The ME vBESS ES Indices update daily. We ingest source data from all markets in §4.2 every day and publish Index values for delivery day D-2, normally every calendar day (§2.7). The two-day lag gives settlement, late submissions, and backfills on aFRRE, mFRR, and TTRR P2 activated-energy series time to complete before we compute the Index for that day.

Previously published Index values may be revised in three cases: when ESIOS or OMIE correct or complete values for periods already published; when new source data materially improves a historical value; or when a methodology change under §7 applies retrospectively.

2.6 Market evolution reflected in the Index

The Index reflects each market structure only for the periods in which its native mechanism and data were in force. The key Iberian milestones are:

Because of these changes, Index values before and after each milestone are not fully comparable (§9).

2.7 Minimum data

The Index for a delivery day needs the price, activation and traded-volume inputs in §4.2 for that day. The offered volumes and the mFRR reserve requirement are used only for the static market-depth measurement in §4.3, so they are not needed for a single day. If a required input is missing or incomplete when the day is computed, the day is run again once the data arrives (§8.3). Within a day, the fallbacks in §8.3 fill gaps in individual periods.

3. Revenue Components

3.1 Revenue components included in the Indices

The ME vBESS ES Indices capture the primary revenue opportunities available to a representative BESS in Spain's wholesale and ancillary markets. The table below shows each component, the market it is modelled from, and the price signal used.

3.2 Excluded revenues and costs

The ME vBESS ES Indices exclude revenues and costs from the following sources:

• Primary reserve (regulación primaria): in Spain, primary frequency response is a mandatory, unpaid obligation for generators above a size threshold. It is not a traded product, so there is no revenue stream to model.
• Imbalance settlement: we assume positions are fully closed out in the real-time step, so no residual imbalance exposure is modelled.
• Capacity market payments: not modelled. Orden TED/966/2026 of 15 September 2026 created a capacity market for mainland Spain, but no auction has cleared, so there is no capacity revenue to include.
• Curtailment compensation: site-specific, so not modelled.
• Network access charges and taxes: site-specific, or set to zero for comparability (§2.2). They default to zero in the representative asset.
• Bilateral contracts (PPAs, tolls, floors): not modelled. Their terms are private and asset-specific.
• Battery operator fees, warranty costs, and O&M: the Index reports market revenue net of charging costs, not net of operating costs.

4. Data Inputs and Use of Discretion

4.1 Data visibility and parameters

The Index is built exclusively from publicly available market data, with the exception of the representative asset in §2.2 and the parameters below. Modo Energy sets these parameters rather than observing them. The table lists each parameter with its value, the reason for it and the section that defines it. Modo Energy's market owner for Spain sets and maintains both the representative asset and these parameters (§4.3).

4.2 Data inputs and sources

The inputs and their providers are listed below. The fallbacks applied to them are set out in §8.3.

Each input has a single source, and no second source is used as a backup. We ingest all inputs into our data lake daily.

4.3 Use of discretion and consistent judgment

The Index is rule-based. The parameters in §2.2 and §4.1 apply in the same way to every day and every duration, and no value is set by hand for a single day. Judgment is used only when a source or parameter is chosen or changed, and the market owner makes that call. We select sources based on reliability, relevance to BESS operations, and the activity, depth, and transparency of the underlying market.

Every parameter sits in the code of the Global Dispatch Model. A change is reviewed by a second person before release, is tested by the automated regression tests in §8.1, and follows the change process in §7.2.

Three inputs to the size-aware participation model in §5.6.1 are set by Modo Energy rather than read directly from a published series:

• Market depth. For each market and direction, we measure the volume that competes for what the system procures or activates, from the ESIOS sources in §4.2. For aFRR capacity and aFRRE this is the volume offered. ESIOS publishes no offered volume for mFRR or TTRR P2, so mFRR uses REE's reserve requirement and TTRR P2 its activated energy. Batteries are not the only bidder: hydro, thermal plant and demand-side flexibility compete for the same volumes, and the measurement covers all of them.
• Price-taker threshold. Below a fixed share of market depth, an asset takes its full pro-rata share of what the system procures or activates. We set that share from backtest sensitivity runs over January 2025 to June 2026 at 5, 50 and 500 MW. The aFRR capacity auction carries a higher threshold than the balancing energy markets: its prices are flat within a delivery period, so there is no price tail for a small asset to select into.
• Participation ceiling. aFRR capacity and TTRR P2 upward carry a ceiling of 30% of rated power. §5.6.1 explains why.

The values used are:

Depths are rounded. The capture formula depends on rated power and market depth only, not on duration, so the same parameters apply to all three Indices. Market depths are static measurements taken in July 2026 and applied across the assessed period. ESIOS publishes the aFRR capacity and aFRRE offer files with a lag of around 90 days, so their measurement ends in April 2026.

5. Modelling Methodology

5.1 Overview of the Global Dispatch Model

We produce the Index by simulating the dispatch of the representative asset in §2.2 against historical market prices, using our Global Dispatch Model (GDM). This is the same proprietary optimisation framework behind our Indices in other regions and our forecasts. For each day in the assessed period, the GDM works out how the representative asset would have scheduled its power across the markets in §3.1 to maximise revenue, given the physical limits of the battery and the rules of each market.

5.2 What the model is solving for

The GDM answers one practical question: given the prices observed on a given day, and the physical and regulatory limits of the asset, what is the highest revenue the asset could have earned with the foresight set out in §5.3?

To answer it, the model allocates the battery's power and stored energy across three families of market opportunities:

• Energy markets: the day-ahead market and OMIE's continuous intraday market.
• Ancillary capacity markets: aFRR (Banda Secundaria), where the battery is paid to hold capacity available for REE.
• Ancillary activated-energy markets: aFRRE, mFRR, and TTRR P2, where the battery is paid in EUR/MWh for the energy it actually delivers when called.

Revenue is the sum of these streams, net of the charging energy cost. The model also applies internal consistency rules that stop it from scheduling trades that look profitable on paper but that a physical battery could not execute (§5.4).

5.3 Three-step optimisation sequence

A real BESS trader in Spain does not make all decisions at the same moment. Each market gate closes before the next, and traders commit at each gate using only the information available at that point. The sequence of the Spanish power market is:

Times are Spanish local time (CET/CEST). Three timing features shape the optimisation sequence:

• The MIC has no single gate close. It is a continuous market. Positions roll until the T-60 hard stop for each delivery period. Spain keeps T-60 under a temporary CNMC exception to the EU cross-zonal gate closure of T-30.
• mFRR and aFRRE share the same BSP gate closure time (T-25 minutes). This is set by the ACER-approved Implementation Frameworks and transposed into PO 7.2 and PO 7.3. They differ in clearing speed: aFRRE clears every 4 seconds, mFRR every 15 minutes.
• Most markets now clear at 15 minutes. Intraday auctions and the MIC moved on 18 March 2025, and day-ahead and TTRR P2 on 1 October 2025 (96 periods per day, up from 24). The model keeps TTRR P2 at 60 minutes (§5.7).

The model reflects this sequence by solving the dispatch problem in three sequential steps:

1. Day-ahead step: sets aFRR capacity commitments, day-ahead energy positions, and TTRR P2 commitments, co-optimised at this step. Activated volumes are published only after delivery, so the model uses the realised activations instead of a forecast.
2. Intraday step: rebalances day-ahead positions in the MIC at the ICM price index.
3. Real-time step: sets aFRRE and mFRR activated-energy positions. The MIC is also traded in this step, at the ICM price index.

Each step takes the positions committed in earlier steps as fixed. Once a decision is locked in, the model cannot unwind it. This sequencing keeps the simulation closer to how a real trader operates than a single all-markets optimisation would.

5.3.1 Day-ahead step

• Granularity: 60 minutes before 18 March 2025. From that date the step runs on the 15-minute model grid (§5.3): aFRR capacity is set every 15 minutes, and day-ahead energy is held constant across each hour until 30 September 2025 and set every 15 minutes from 1 October 2025, reflecting the OMIE / SDAC move to a 15-minute day-ahead auction. TTRR P2 positions are held for a whole hour throughout (§5.7).
• Foresight: perfect foresight of day-ahead energy prices, aFRR capacity prices and TTRR P2 activation prices over the full day.
• Markets optimised: day-ahead energy, aFRR capacity, TTRR P2 activated energy.

5.3.2 Intraday step

• Granularity: 60 minutes before 18 March 2025; 15 minutes from 18 March 2025 onward.
• Rolling horizon: 2-hour windows, advanced sequentially.
• Foresight: perfect foresight of the intraday prices in §3.1 over each 2-hour rolling window.
• Markets optimised: the MIC at the ICM price index.
• Commitments carried in: day-ahead energy positions, aFRR capacity and TTRR P2 positions from Step 1 are fixed.
• Liquidity cap: in each period and each direction, combined intraday and real-time dispatch is capped at the volume traded in the MIC in that period, and never above twice rated power. The MIC does not run through the capture formula in §5.6.1, because continuous trading executes against a live order book rather than competing for acceptance in an auction.

5.3.3 Real-time step

• Granularity: 60 minutes before 18 March 2025; 15 minutes from 18 March 2025 onward.
• Rolling horizon: 2-hour windows, advanced sequentially.
• Foresight: perfect foresight of the ICM price index and of aFRRE and mFRR activation prices over each 2-hour rolling window.
• Markets optimised: the MIC at the ICM price index (real-time energy), aFRRE activated energy, mFRR activated energy.
• Commitments carried in: all prior steps' positions are fixed.

5.4 Market stacking and physical limits

Spain's market design lets a battery hold simultaneous positions across day-ahead energy, intraday continuous energy, real-time energy, aFRR capacity, aFRRE energy, mFRR energy, and TTRR P2 energy. Two things limit it: the physics of the battery and its grid connection, and the depth of each market. The model enforces stacking through four families of constraints:

1. Sufficient state of charge for ancillary delivery. The battery's state of charge at each timestep must be enough to deliver every discharging-direction ancillary product it holds at that moment. Each product has its own delivery horizon: 15 minutes for aFRR, aFRRE and mFRR, and 60 minutes for TTRR P2. These horizons come from the energy reservoir requirements in the REE prequalification conditions.
2. No double-counting between aFRR and aFRRE. aFRR (capacity reservation) and aFRRE (energy activation) are distinct products. The model applies a pair of parallel constraints so that each physical limit (grid, battery, SOC) is checked once including aFRR and once including aFRRE, but never both in the same inequality. Reserved aFRR capacity must bid into aFRRE but is not necessarily activated, and it cannot bid into other energy markets at the same time.
3. One direction at a time in activation markets. For the three energy-activation markets (aFRRE, mFRR, TTRR P2), the model allows only one direction (POS or NEG) per timestep per market, enforced through a binary variable. This stops the model from earning revenue on both sides of an activation product in the same settlement period, which is physically impossible.
4. Market depth. Each ancillary and balancing market absorbs only a share of what the asset offers into it, and that share falls with asset size. §5.6.1 sets out how it is measured and applied, and how the intraday continuous market is treated instead.

5.5 Battery physical model

The representative asset is modelled as a single lumped battery governed by four technical characteristics:

• Round-trip efficiency of 88%. Efficiency losses apply to the charging flow: 1 MWh drawn from the grid stores 0.88 MWh of usable energy.
• Continuous state of charge. Stored energy in each timestep equals the previous level, plus charging (net of efficiency), minus discharging. The SOC adjustment uses the net energy flow across energy markets within each timestep. Offsetting trades that net out within a settlement period therefore incur no round-trip efficiency penalty.
• State of charge boundaries. To allow day-by-day solving, SOC is assumed to be 0% at midnight each day. During the day, SOC can vary freely between 0 and 100%, subject to the cycling cap below.
• Cycling limit per day. Total discharge throughput per day is capped at the daily cycling limit (2.0 cycles) times the battery's usable capacity. Both wholesale discharge and ancillary activation count towards the cap. Warranties are typically written on total throughput, not on the market the energy was sold into.

Cell-level degradation is intentionally disabled in the published Index (§2.2). This keeps the asset's duration constant across the full history, so years remain comparable.

5.6 Ancillary services: capacity vs activated energy

Spain's balancing stack separates two things: reserving capacity (Banda Secundaria capacity payments) and delivering energy when called (aFRRE, mFRR, and TTRR P2 activation payments). The two product families bid into different gates, and the model treats them differently:

• aFRR capacity (Banda Secundaria). REE pays the battery for holding capacity available. The bid is set in the day-ahead step. Revenue accrues at the cleared capacity price, in EUR/MW/h, on the capacity the auction accepts. That is a share of the asset's bid, set by the capture rate in §5.6.1. Capacity held reduces the power and SOC available for energy markets. Holding capacity is independent of activation: it does not drain SOC or count towards cycling on its own, so the capacity payment is the only revenue from aFRR itself.
• aFRR capacity and aFRR energy. Since the 20 November 2024 SRS reform, aFRR energy bids are formally decoupled from the capacity allocation. A provider could, in principle, price its energy bid out of activation. In practice, REE expects allocated capacity to stay available. The model therefore assumes that capacity allocated in the day-ahead step remains fully available to the aFRRE energy market.
• Activated-energy markets (aFRRE, mFRR, TTRR P2). The battery is paid in EUR/MWh on the energy it actually delivers when called. Revenue is recognised on the activated MWh, not the contracted MW. The energy flow moves SOC and counts towards the daily cycling limit.

5.6.1 Market depth and capture rate

Spanish ancillary and balancing markets are shallow next to the wholesale market. A battery that could sell its full rated power into every one of them would win a share of national volumes that no real asset achieves. Size is the binding constraint. A small battery clears its volume without moving the bid ladder; a large one has to bid deeper into the merit order and earns increasingly average prices.

The model prices this with a single formula, applied to aFRR capacity, aFRRE, mFRR and TTRR P2. Three quantities define it:

• Market depth: the volume that competes in that market and direction, measured from published data (§4.3).
• Price-taker threshold: a fixed share of market depth, below which an asset is a price-taker and wins its full pro-rata share of the volume procured or activated.
• Participation ceiling: an upper bound on the share of rated power a market can absorb, applied after the formula.

Effective power is the power the market lets the asset compete with:

Below the threshold the asset is a price-taker, and effective power equals rated power. Above it, effective power grows with the logarithm of size: each ten-fold increase in rated power adds roughly the same number of megawatts. The curve is continuous through the threshold.

The cap on what the asset can win in a given settlement period follows from effective power, the volume the system actually procured or activated in that period, and market depth:

Dividing this cap by rated power gives the capture rate: the share of rated power the asset wins. The capture rate is not itself logarithmic. Effective power is, and the rate falls out of the division. The shape of the curve depends only on the ratio of rated power to market depth, so one curve prices a 5 MW and a 500 MW asset, with only the threshold share and ceiling in §4.1 set per market. The level in each period also scales with the volume procured or activated (V_t/D). It does not depend on duration, so the same cap applies to all three Indices. Because depth differs by market and direction, the size at which the constraint starts to bite differs too.

The intraday continuous market is excluded. In the four markets above, offers compete for acceptance in a merit order, so a larger asset has to bid deeper to be accepted, and the logarithmic term approximates that. The MIC is pay-as-bid and continuous: trades execute against resting orders at their own prices, with no merit order to bid down and no acceptance to compete for. Size still constrains it, but through the liquidity available in the order book rather than through cannibalisation. The model caps MIC dispatch at the volume actually traded in each period instead (§5.3.2).

The participation ceiling

aFRR capacity is deep enough that the logarithmic term returns close to full rated power for small and mid-scale assets. That result is unreachable in practice. Across the market, the auction allocates roughly half of the band offered to it: 55% upward and 49% downward between January 2025 and April 2026. No asset sustains its full rated power in aFRR capacity in every period of the year. Capacity held for aFRR is also not available for intraday trading.

Modo Energy therefore applies a ceiling of 30% of rated power to aFRR capacity and to TTRR P2 upward, taken after the logarithmic term. For aFRR capacity, 30% sits below the roughly half of the offered band that the auction allocates. The ceiling and the curve do opposite things to the spread between asset sizes: the curve is size-dependent and widens it; the ceiling is size-blind and flattens it across the small and mid-scale range. They are separate controls, and we set them separately.

The ceiling can bind only while effective power stays above 30% of rated power, and then only in periods where enough of the band is procured. For aFRR capacity, effective power stays above 30% up to about 250 MW upward and 280 MW downward. Above that the curve takes over: effective power falls to 18–20% of rated power at 500 MW and 10.5–11.5% at 1,000 MW, before the scaling by volume procured in each period. aFRRE, mFRR and TTRR P2 downward carry no ceiling, so their capped and uncapped curves are identical.

The ceiling is a modelling assumption, not a limit on what an asset may offer. Custom indices can restrict participation further, and the two apply on top of each other.

5.7 Technical constraints (TTRR P2)

TTRR P2 is phase 2 of the technical constraints process that REE runs immediately after the day-ahead market clears. REE checks the day-ahead schedule (PDBF) for network constraints and publishes a viable schedule (PDVP) by D-1 14:45. The process runs in two phases:

• Phase 1 is locational. REE dispatches specific assets to resolve network constraints in the PDBF. Upward dispatch is paid as bid. Downward curtailment is not paid.
• Phase 2 is open country-wide. Any unit in Spain can participate. REE rebalances the system as a whole, keeping Phase 1 commitments fixed.

Phase 1 mostly dispatches thermal generation upward to resolve constraints. Phase 2 then nets that energy off through downward activations. The result: Phase 2 is structurally a downward market. Downward activations exceeded upward activations by more than an order of magnitude in 2025. This creates a systematic interaction between day-ahead positions and Phase 2 activations, which is why the model co-optimises day-ahead energy and TTRR P2 in the same step (§5.3.1). In dispatch, the model uses TTRR P2 downward activations in two distinct ways:

• Same-period netting. Energy sold in the day-ahead market is bought back in TTRR P2 within the same settlement period. The position nets out financially without moving the battery's state of charge. Under the net-flow SOC rule in §5.5, it incurs no round-trip efficiency penalty.
• Charging source. The battery charges at the TTRR P2 downward activation price and discharges the stored energy later into the day-ahead, intraday, or balancing markets. In these periods, TTRR P2 is simply an alternative to charging at the day-ahead or intraday price.

Unlike the frequency-control markets, TTRR P2 activations are persistently asymmetric, especially in periods of high renewable curtailment. The Index reflects these properties as follows:

• Granularity: the model holds TTRR P2 positions for a whole hour throughout. TTRR P2 has cleared every 15 minutes since 1 October 2025, so from that date this is a simplification (§9).
• Modelled from realised data: the model uses realised ESIOS activation data rather than a forward signal, and applies the capture rate in §5.6.1 to national volumes. Phase 2 is open to any unit in Spain, so national volumes are the base a real asset competes for.
• Participation ceiling: upward activations carry the 30% ceiling in §5.6.1. Phase 2 upward is called in a small fraction of periods, and a market called that rarely should not drive dispatch. Downward activations carry no ceiling.
• Asymmetric coverage: where ESIOS publishes only one direction (commonly NEG), the symmetric-band fallback in §8.3 does not apply. TTRR P2 is intrinsically asymmetric. The gap is real, not a measurement artefact.

5.8 mFRR (Reserva de regulación terciaria)

The mFRR market compensates manual frequency restoration as an energy-only product. REE calls mFRR-qualified plants per event and pays the cleared marginal price on activated MWh. Spain has no separately paid mFRR capacity product. The market connected to the European MARI platform on 10 December 2024, with clearing per 15-minute MTU and a BSP gate closure of T-25 minutes.

The mFRR price signal is asymmetric and tail-heavy. Upward activation prices average well above day-ahead levels and spike hard in scarcity events. Downward activation prices average close to zero. Capturing this asymmetric signal is a key reason mFRR is included in the stack. The model treats mFRR as follows:

• Granularity: 15 minutes from 18 March 2025. Before that date the model reads mFRR at the first quarter-hour of each hour (§5.3).
• Modelled from realised data: ESIOS publishes mFRR activations only after the fact, so the model uses realised ESIOS activation data.
• Size-aware participation: activated volume is capped by the capture rate in §5.6.1. mFRR carries no participation ceiling.
• Energy-activation only: revenue is recognised on activated MWh. There is no capacity payment.

5.9 Revenue calibration

In other Modo Energy indices (for example ME vBESS DE), we apply a flat calibration factor to all simulated per-timestep revenues. It bridges the gap between perfect-foresight modelled revenues and what a real fleet can earn, covering residual foresight, availability, and execution gaps.

For the ME vBESS ES Indices, no post-hoc calibration factor is applied (calibration factor = 1.00). Public per-asset BESS revenue disclosure in Spain is too sparse to anchor a Spain-specific factor, and Modo Energy has chosen not to transfer a factor observed in another market. The Index may therefore overstate what an operating battery earns (§9).

The size-aware participation parameters in §5.6.1 are listed in §4.1. Market depths are measured from ESIOS data, and Modo Energy sets the price-taker threshold and the participation ceilings (§4.3). They act inside the optimisation, on how much volume the asset can win, rather than as a scalar applied to the revenue it earns.

6. Transparency

We publish detailed explanations of calculation methodologies, revenue components, and benchmark updates, so stakeholders can understand how the benchmark works. Transparency measures include:

• Publication of methodology documents outlining calculation processes and revenue components.
• Revisions of historical values, as described in §2.5.
• Advance notice of material methodology changes in line with §7.2.
• A contact point for questions about the methodology: team@modoenergy.com.

7. Methodology changes

7.1 Review and update process

The methodology is reviewed when market rules, market conditions or data availability change, and when inputs or results look out of line. Examples include the OMIE move to 15-minute day-ahead auctions, the SRS reform of aFRR procurement, and ESIOS publication of new activated-energy series. Any change that follows goes through §7.2.

Direct back-testing against observed transactions is not possible, because per-asset revenue is not publicly disclosed in Spain. Instead, we cross-check the day-ahead revenue of the 4H Index against Modo Energy's hourly TB4 spread for Spain. The Index earned about EUR 90k/MW/year in the day-ahead market in 2024 and EUR 129k/MW/year in 2025, against a TB4 spread of about EUR 88k/MW/year and EUR 121k/MW/year. The Index can sit above TB4 because it may cycle twice a day, trades 15-minute prices from 1 October 2025, and can sell day-ahead energy that it buys back in TTRR P2 (§5.7). Both are perfect-foresight measures, so this checks the day-ahead step, not realisable revenue. We also compare the revenue stack composition with Modo Energy's published Iberian BESS market analysis.

7.2 Material and non-material changes

A change is material where it moves any of the three Indices by more than ±2% cumulatively in a calendar quarter. Material changes are announced with not less than 30 days' written notice before they take effect, together with an impact analysis. We welcome feedback on proposed material changes during the notice period at team@modoenergy.com.

Other changes are non-material. They include corrections and clarifications that do not alter the substance of the methodology. They may be made without prior notice and are disclosed as soon as practicable.

Emergency changes needed to protect the integrity of the benchmark or to comply with law may be made without prior notice, with notice given as promptly as circumstances permit.

All changes are recorded, with their effective date, in the change log (Appendix I).

7.3 Recalculation of published values

When a methodology change applies with retrospective effect, Index values may be recalculated across the published history, as they were for version 0.4 (Appendix I). Only the current version of results is available at any time.

8. Consistency and continuity

8.1 Quality assurance

Modo Energy uses these checks:

• Automated validation checks to identify discrepancies in input data.
• Automated regression tests on every change to the Global Dispatch Model.

8.2 Data integrity

We maintain data integrity through:

• Access controls on the data lake and on the model code.
• Traceability from each raw source (ESIOS, OMIE) into the model feed.

8.3 Late, missing or distorted data

The two-day lag in §2.5 normally gives every input in §4.2 time to arrive. If an input for a delivery day is still missing or incomplete when the Index is computed, the day is run again once the data is available, and its value is updated (§2.5, §2.7).

Inputs are used as published, including extreme prices and periods of system stress such as the Iberian blackout of 28 April 2025. Index values are not adjusted by hand for market stress. The only adjustments are these:

• Price and allocation gaps. Where a price or the aFRR capacity allocation is missing for a period, the model uses the last published value.
• Intraday volume gaps. OMIE omits periods with no MIC trades. A period with no published traded volume therefore gets a liquidity cap of zero, and the asset cannot trade intraday or real-time energy in it (§5.3.2).
• Symmetric-band fallback. Before 20 November 2024, ESIOS published aFRR capacity and aFRRE in one direction only. The model mirrors the published direction onto the absent one (§2.6).
• No activation without data. Where ESIOS publishes no aFRRE, mFRR or TTRR P2 activation for a period and direction, the model treats it as no activation. For TTRR P2, where ESIOS often publishes only one direction, the published side is not mirrored, because the market is intrinsically asymmetric (§5.7).

If a source corrects data for a period already published, the Index for that period may be recalculated (§2.5).

8.4 Traceability and verification

Every benchmark calculation is traceable:

• We record the model version that produces the published Index, and keep the outputs of each run.
• Reproducibility: current published Index values can be recomputed from the current input data and that model version.
• Material methodology changes are disclosed publicly (§7.2).

9. Limitations

The Index is a simulation. These are its main limits:

• Simulated, not measured. It shows what a representative asset could have earned, not what any real battery earned. Per-asset revenue is not public in Spain, so the Index cannot be back-tested against real results (§7.1).
• Foresight. The model knows day-ahead, aFRR capacity and TTRR P2 prices for the full day when it commits day-ahead positions, and intraday, aFRRE and mFRR prices two hours ahead (§5.3). No calibration factor corrects for this (§5.9), so the Index may overstate what an operating battery earns.
• Shallow, concentrated markets. Spanish ancillary and balancing markets are shallow, and competing offers come from a limited number of providers, mainly hydro and thermal plant. The size model in §5.6.1 prices this with static market depths, taken in July 2026 (§4.3) and applied to the whole history. It cannot capture how depth changed over time or how a real asset's bid prices affect its acceptance. TTRR P2 activations are seasonal and have grown since 2023, so a single depth figure fits that market least well.
• Price-taking in energy markets. The asset's day-ahead and intraday trades are assumed not to move clearing prices or the ICM price index.
• Index prices. Intraday and real-time revenues are valued at the ICM price index, an average of continuous trades, with volume capped at MIC traded volume (§5.3.2). A real asset trades at its own prices.
• Concentrated inputs. Each input comes from a single publisher: ESIOS for the day-ahead price, which it republishes from OMIE, and for all balancing and ancillary data; OMIE for the ICM price index and MIC traded volume. An error or change at that source flows into the Index (§4.2).
• One asset design per duration. 50 MW at 1, 2 or 4 hours, full grid connection, no grid fees or IVPEE, no degradation, and 0% state of charge at midnight (§2.2, §5.5). The revenue streams in §3.2, such as capacity payments and bilateral contracts, are not included.
• Coarser timing in parts of the history. Before 18 March 2025 the model reads markets that already cleared every 15 minutes at the first quarter-hour of each hour (§5.3). Since 1 October 2025 it holds TTRR P2 positions for a whole hour although the product clears every 15 minutes (§5.7).
• Changing market design. The history spans the SRS reform, the move to 15-minute intraday and day-ahead markets, and the connection to MARI and PICASSO (§2.6). Values before and after these dates are not fully comparable. Before 20 November 2024, aFRR revenues rely on the symmetric-band fallback.

Appendix I: Methodology changes

Methodology versions are recorded here, with their effective date and classification under §7.2.

Versions 0.2 (11 June 2026) and 0.3 (18 June 2026) were revisions made before first publication on 2 July 2026.