Italy Methodology

Modo Energy provides benchmark data for battery energy storage systems across global energy markets, applying a standardized methodology to ensure consistency and transparency across all produced Indices.

1. Introduction

The Modo Energy Italy Methodology Framework outlines the calculation and construction of Modo Energy's battery energy storage Indices in Italy. This document covers:

• 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.

1.1 A simulated Index

The Modo Energy Italy Index is simulated, not measured. It represents the revenue a representative battery could have earned in Italy's markets, rather than the metered revenue of any single asset. In the wholesale markets it represents a well-run operator; in the balancing markets it is bounded by its pro-rata share of the zone's activated balancing volume, and by the observed call depth of storage units (§5.6).

A simulated representative-asset Index is adopted for three reasons:

• Comparability. It defines one consistent archetype whose revenue compares across bidding zones, across durations, and against Modo Energy's Indices in other markets.
• Completeness. It can be computed consistently across the market's current configuration. The per-unit data GME publishes for the dispatching-services and balancing markets is offer- and award-level rather than metered settlement, and arrives with a publication lag.
• Market opportunity. It measures the revenue available in the market, not an individual operator's availability or commercial arrangements.

1.2 What the Italy Simulated Index represents

The Modo Energy Italy Index family (the ME vBESS IT Indices) represents the simulated revenue of grid-scale lithium-ion batteries in Italy's wholesale and balancing markets. It is published per duration (4H / 6H) and per bidding zone, because Italy's zonal price structure is pronounced enough that a single national figure would represent no real asset. Revenues are gross of site-specific grid charges, warranties, and operating costs.

Currency: all revenues are computed and published in EUR.

1.3 Terminology and market timing

Clearing times follow the pan-European SDAC (day-ahead) and SIDC (intraday) schedules as applied by GME; exact clock times should be read against GME's current market timetable.

2. Index construction

2.1 The ME vBESS IT Indices

The headline Index is the 4-hour variant, per zone. Four hours is the duration standard of Italy's contracted storage procurement and of the majority of the build-out pipeline. The 4H and 6H variants are published alongside for comparability with Modo Energy's other markets.

2.2 Representative BESS specification

Note: The parameters marked standard convention (rated power, round-trip efficiency, depth of discharge, degradation treatment and the grid limit) are set identically across every Modo Energy BESS Index, which is what makes the Indices comparable across countries. The choices specific to Italy are the 4-hour headline duration and the per-zone construction.

2.3 Index calculation methodology

The Index is calculated by:

1. Running Modo Energy's optimisation model over the assessed period, per bidding zone, using the asset specification in §2.2 and the inputs in §4.2.
2. Computing simulated revenue across each market and direction, in EUR, at 15-minute resolution.
3. Summing per settlement period and normalising by rated power (50 MW) to obtain EUR/MW per period.
4. Aggregating into monthly, quarterly, and annual indices.

2.4 Index value representation

• EUR/MW (period): total Index revenue for the period (e.g. EUR/MW/month). All values are per MW of the asset's rated AC power (the 50 MW of §2.2).
• EUR/MW/year (annualised): period revenue divided by the number of days in the period, multiplied by 365.

2.5 Publication cadence and revisions

Index values publish monthly, in a single final edition, once GME's per-unit balancing publications for the month are complete, in practice about two months after the month ends (M+2). No provisional edition is published. At first publication, the entire December 2025 to June 2026 history is computed in a single batch under the parameter set then in force, archived with Stable Version 1.0. Published values are not restated by default; the circumstances in which Modo Energy does restate, and how a restatement is notified, are set out in §7.4 and §7.3.

2.6 Market context and published history

Italy's markets reached their current shape in stages. The table is context for how balancing revenue became available to storage and why the Index starts where it does. It describes the market, not a published series.

3. Revenue components

3.1 Revenue components included

An energy-only balancing stack. The Italian balancing markets included here remunerate accepted energy, not held capacity: MB pays only for the offers Terna accepts and activates. The Index therefore has no capacity-payment stream, and the balancing streams are modelled as directional activation-energy products with volume caps (§5.6). The MB stream covers mFRR activations only (sessions GR1–GR4); Terna's unit-specific redispatch instructions are excluded, because storage has never been awarded redispatch volume (zero MWh across twelve months and seven zones). aFRR energy is modelled in the six zones where activation is observed; Sardinia records none and carries a zero (§5.6.6).

3.2 Excluded revenues

• MSD: storage may participate, but observed storage revenue in MSD is immaterial, so it is not modelled. The MSD products carry a throughput coefficient of zero (§5.6) and therefore contribute no revenue and no state-of-charge movement. Reviewed as storage participation develops.
• Fast Reserve / FCR pilot: a competitive fast-reserve pilot was introduced in June 2026. It is a pilot rather than an established open-market stream and is not included; inclusion will be assessed as the mechanism matures.
• Capacity Market and MACSE (contracted storage capacity): contract-specific, and excluded. MACSE reviewed as it reaches delivery.
• Mandatory primary reserve: not remunerated as a market product.
• Imbalance settlement: positions are assumed closed out (§3.1).
• Grid fees, levies and dispatching charges: site-specific, and outside gross market revenue.
• Bilateral contracts, and operator, warranty and O&M costs: outside the scope of a gross market-revenue Index.

4. Data inputs and use of discretion

4.1 Data visibility

The Index is built from publicly available market data, except the representative-asset parameters (§2.2), the pooled call-depth values and the revenue calibration factor (§4.3), which Modo Energy sets under the governance in §6–§7.

4.2 Data inputs and sources

4.3 Use of discretion

Modo Energy applies discretion strictly within predefined parameters, only where market data is incomplete or requires interpretation. Each item below is rule-bound and subject to internal governance oversight.

• Fleet definition: the active flexible fleet, the denominator of the MB κ, is a capacity rather than an average of activated volume, and it counts only the units that compete regularly. A unit enters F_z if it was awarded upward mFRR volume in the zone in at least 3% of the settlement periods in which the zone activated over the measurement window. Its size is the 95th percentile of its offered power where it submits explicit quantities and its maximum awarded power otherwise, capped at 250 MW so that a single very large thermal unit cannot dominate the denominator. F_z is the sum of those sizes, read from GME's per-unit offers. The 95th percentile rather than the maximum because a unit's largest single award can be an outlier, while the 95th percentile is the power it demonstrably makes available.

  The 3% condition applies to the numerator as well, so N counts the activations of the same units and no others and both sides of the ratio describe one population. It is set so that the units retained account for at least 90% of the volume their zone activated. Units below it are awarded a handful of times a year, typically for a local or must-run reason, and counting their full rated power against a battery overstates the competition it faces.

  The measurement window is twelve months, which keeps F_z stable from one quarterly refresh to the next. The aFRR fleet is measured from the aFRR award publication, using the reserved band per unit, and is recomputed each month. It carries no participation condition: a unit appears in that publication only when it holds awarded band, so the occasional participant the 3% condition removes from the mFRR fleet does not arise there. Systematic reconciliation against Terna's qualified-unit registry is planned governance, and until it is in place the identification of units is a stated limitation (§5.6.7).

• Call-depth cap: d̄_z is the unweighted mean, over every (storage unit, activated settlement period) observation in the zone, of the unit's awarded power divided by its rated power, capped at 1. Rated power is proxied by the unit's 95th-percentile offered power over the window. It is measured per zone where a multi-unit sample supports it (North, Centre-South, Sardinia); the remaining zones, including the single-unit Centre-North, carry the pooled value (0.377 for MB, 0.25 for aFRR). Reasonable estimator variants (per-unit means averaged across units; volume-weighted) read higher in the North (approximately 0.43–0.46 against the value in force), so the estimator in force is the conservative choice; changing it follows §7.3. This is the one judgment element in the coefficient and is declared as such.
• Measured, not fitted: the MB κ contains two measured parameters, the flexible fleet F_z and the call-depth cap d̄_z, both read from GME's per-unit award file over the same twelve-month rolling window, refreshed quarterly and ending at the per-unit publication boundary, 30 June 2026 at this refresh. Neither is tuned to a revenue level; the coefficient is checked out-of-formula by comparing its implied call frequency with the measured call frequency of awarded storage units (§5.6.2, property 2). The aFRR κ is built identically from the aFRR award publication.
• Coverage: every zone carries a balancing component, because the coefficient rests on the volume the zone activates, not on the presence of storage in the zone. Zones with no or one storage unit use the pooled call depth; Sardinia's aFRR component is zero because no aFRR activation is observed there (§5.6.6).
• Reproducibility: every cap can be reproduced by a third party from GME's per-unit award files: the activated volume, the fleet F_z and the call depth d̄_z all come from the same public source. The validation reference, the realised remuneration of awarded storage units, comes from the same file.
• Pay-as-bid price proxy: the volume-weighted average accepted bid stands in for the achievable MB price (§5.8).
• Intraday continuous price point: the last-hour traded price stands in for the achievable XBID price (§3.1), rather than a full-session average or a best single execution.
• Revenue calibration factor: the 0.90 on the day-ahead leg (§5.9) rests on a measured floor plus one declared judgment. The walk-forward percent-of-perfect frontier of a pre-declared family of ex-ante day-ahead strategies, replayed on the Index's own price inputs, reaches 0.825–0.866 by zone (fleet-value-weighted 0.84) using calendar and price history only; the factor sits above that floor to stand in for the weather and load information real desks hold, which the measurement cannot replay. The scope (day-ahead only) carries no judgment. Applied as a single scalar, never per-zone, never tuned to a revenue level, re-examined against the re-measured frontier at the quarterly parameter refresh and reviewed annually. No other leg carries a factor (§5.9).
• Regulatory windows: each balancing product enters and leaves the Index on the dates its mechanism was in force (§5.7).
• Parameter refresh: at each monthly GME publication (M+2), N is taken per settlement period as published; F_z and d̄_z are recomputed on a 12-month rolling window, refreshed quarterly, d̄_z subject to a minimum-sample threshold, below which the pooled value applies (currently Centre-North, South, Calabria and Sicily). The parameter set in force applies to all editions produced until the next quarterly refresh. Refreshes apply prospectively only. They change how subsequent months are computed and never touch a published value; the parameters in force are archived with the Stable Version (§7.1). As a continuous control, ω (storage units' captured share divided by fleet share) is recomputed each batch; if it settles durably above approximately 1.2, introducing it explicitly into the formula goes through the methodology-change process (§7.3).

5. Modelling methodology

5.1 Overview

The Index simulates the dispatch of the §2.2 asset against historical prices using Modo Energy's optimisation model, the same model used in Modo Energy's other markets. It is a mixed-integer linear programme that maximises revenue across the §3.1 markets, subject to the battery's physical limits and each market's rules.

5.2 What the model is solving for

Given the day's prices, the model maximises the revenue available to a well-run battery, allocating its power and stored energy between wholesale energy (day-ahead, MI-A2, XBID) and balancing energy (MB and aFRR, paid for the energy Terna activates). Revenue is the sum, net of charging cost. The constraints in §5.4 exclude schedules that a real battery could not execute.

5.3 The three-step sequence: how intraday is modelled

The model solves in three sequential steps, each taking earlier positions as fixed, reflecting that a position is committed day-ahead and then adjusted twice as delivery approaches:

1. Day-ahead step: sets MGP energy positions. 60-min before 1 Oct 2025, 15-min after. Perfect foresight of day-ahead prices.
2. Intraday auction step (MI-A2): rebalances in the auction, 15-min, with imperfect foresight of MI-A2 prices. Bids into MI-A2 are placed before the auction clears, so the clearing price is not yet known when the position is decided; the step optimises against a centred 2-hour rolling mean of the MI-A2 price, not the exact outcome.
3. Real-time step (XBID + balancing): continuous trading and balancing, 15-min, on a 2-hour rolling horizon. The continuous (XBID) price is replaced by a centred 2-hour rolling mean, representing a trader's near-term view rather than the realised outcome.

Where foresight remains, and what covers it. Two price series are still seen as realised rather than forecast:

• the day-ahead price, in the day-ahead step, which is the modelling baseline shared with Modo Energy's other Indices. This is the one leg whose optimisation retains hindsight, and it is exactly the leg the §5.9 calibration deflates, at its measured percent-of-perfect.
• the balancing prices (MB and aFRR energy), in the real-time step, together with the timing of activations: the κ cap of §5.6 is built from the volume each settlement period actually activated, so the model can see in advance which periods will carry an activation. The smoothing above applies to the wholesale energy series; because Italy's balancing services are modelled as directional energy products (§3.1), their price series are not smoothed with them.

The balancing legs carry no foresight deflator, because their revenue is already bounded by observed outcomes on both volume and price: κ anchors the volume to the zone's realised activations and the observed storage call depth, and the price basis is the volume-weighted average accepted bid, below what a strategic bidder captures (§5.8). What hindsight buys the optimiser inside that envelope is selection: it concentrates its permitted volume in the periods where balancing pays best relative to wholesale, much as a strategically bidding operator does through its offer curve, but with certainty rather than skill. This residual is declared as a limitation rather than corrected by a factor: a deflator on an observed-anchored leg would double-count (§5.9), and the two approximations lean against each other, since the price basis understates the achievable price while the selection overstates the achievable timing.

5.4 Market stacking and physical limits

Italy's market design permits a battery to hold simultaneous positions across day-ahead energy, intraday energy and balancing, subject to physical limits on the battery and its grid connection. The model enforces:

1. Rated power: committed power no greater than 50 MW (and the grid limit) each period.
2. State-of-charge feasibility: every scheduled and activated flow must be deliverable from stored energy, including the SoC movement from balancing activations.
3. Directional consistency: up offers backed by discharge capability and stored energy, down offers by charge capability and headroom.
4. Non-physical trade guard: a penalty on simultaneous buy/sell that nets out, so revenue reflects executable trades, not solver artefacts.
5. Day-ahead footroom reservation: the day-ahead solve may hold back discharge headroom to serve the MI-A2 auction in the same settlement period, at a reservation cost of 20 EUR per MW-h held, so headroom is only reserved where the intraday spread justifies it.
6. Continuous-trading tether: within each real-time solve, total XBID buy volume may not exceed total balancing discharge volume, and total XBID sell volume may not exceed total balancing charge volume. Continuous trading rebalances the state of charge around balancing activations rather than running as a standalone arbitrage stream on realised prices.

5.5 Battery physical model

• 88% round-trip efficiency, applied to charging: 1 MWh from the grid stores 0.88 MWh usable.
• Continuous state of charge: within a day, energy is carried period to period, including flows from accepted balancing offers, so delivering the services it sells moves its SoC.
• Daily solve, empty at the day boundary: the optimisation is solved one calendar day at a time, each day independently, so no stored energy is carried across midnight. The asset starts each day at zero state of charge. The day-boundary state in force is recorded with the Stable Version (§7.1).
• Cycling limit: total daily discharge no greater than 1.5 cycles of usable capacity, across wholesale and balancing combined. The limit is an upper bound, not a target: the asset cycles only where prices justify it.
• No minimum-spread hurdle: the asset trades on any positive spread. A calibrated cycling hurdle is used in Modo Energy's forecast products but not in the Index, where it would act as an undeclared parameter inside a published benchmark.
• Degradation disabled, to keep duration constant across the history.

5.6 Balancing services: the activation-throughput coefficient κ

A battery earns in balancing only on the energy Terna activates. κ caps the Index asset's balancing dispatch, settlement period by settlement period, at its pro-rata share of the volume the zone activated, never more than the call depth observed for real storage units. The cap bounds simulated balancing revenue by what one battery could win, rather than by the volume the system procured in total.

5.6.1 The cap

Each settlement period, the battery's dispatched volume in market m, direction d is capped at:

z is the zone, t the settlement period, P^max the rated power (50 MW). In a settlement period where the zone activated no volume in that market and direction, the cap is zero.

5.6.2 How κ is built

From two published quantities and one measured bound:

• N(z, t) is the balancing volume activated in the zone in settlement period t. For MB, this is the accepted mFRR offers of the zone's regular participants (§4.3), upward and downward read separately, all technologies, from GME's per-unit public offers at native 15-minute resolution. Redispatch instructions (which storage has never been awarded) are excluded.
• F_z is the zone's active flexible fleet: the combined rated power of the units regularly awarded mFRR volume over the measurement window (the 3% participation condition, §4.3), from the same table.
• d̄_z is the measured call depth of storage units: the unweighted mean, over the zone's (storage unit, activated period) observations, of awarded power over rated power (formula and estimator sensitivity in §4.3), from the same table.

In words: in each activated settlement period, the battery serves its pro-rata share of the zone's activated balancing volume, its megawatts over the zone's flexible megawatts, and never more than the call depth observed for real storage units.

The cap is applied per settlement period. Reading the activations period by period preserves both the total volume and its timing; an averaged coefficient would convert activations concentrated in a minority of periods into an entitlement available in every period.

The three variables are measured on different clocks. F_z and d̄_z are measured once, on twelve months of published per-unit data, and are then fixed for the quarter (§4.3).

• F_z answers how many megawatts of flexible competitors share this zone.
• d̄_z answers when a storage unit is called, what fraction of its power is asked of it: it is an average, but a conditional one, taken only over the settlement periods in which a unit was actually activated and never over quiet periods.

The third is read afresh for every settlement period. N is the volume the zone actually activated in that quarter-hour, as published, never averaged and never smoothed. Where the zone activated nothing, N is zero and so is the cap.

A settlement period is therefore priced from two numbers that were fixed before it happened and one number that describes what happened in it. The coefficient inherits the rhythm of N: one value per settlement period, per market, per direction. Everything that makes balancing seasonal, or concentrated at particular times of day, reaches the Index through N at 15-minute resolution.

All three quantities are drawn from the same published source, and N and F_z are read on the same set of units, defined in §4.3. GME publishes two balancing series, zonal results at day-plus-one and per-unit offers and awards approximately two months later, compiled on different bases (the zonal series aggregates mFRR with redispatch, and reports energy rather than offered power).

The coefficient has four properties by construction: (1) it contains no parameter tuned to revenue: F_z and d̄_z are read from the public award file, and the participation threshold is set by the volume-coverage rule in §4.3; (2) it is checked out-of-formula at each refresh: the call frequency the formula implies, min(1, N/(d̄·F)), is compared with the frequency at which the zone's storage units are actually awarded; (3) all zonal differentiation derives from the source data, with no per-zone tuning; (4) the cap per MW is expected to fall as the zone's fleet grows (§5.6.4).

Worked example (Sardinia, upward MB). In a given settlement period Terna activates 100 MW of upward mFRR in Sardinia. The zone's regularly competing fleet is 1,140 MW, so the pro-rata share is 100 / 1,140 = 8.8%, and the 50 MW Index asset may dispatch up to 8.8% × 50 ≈ 4.4 MW in that period. In the largest activations the call-depth bound takes over: the cap never exceeds d̄ = 37.5% of rated power. In a period with no upward activation the cap is zero. The cap is a permission, not an obligation: to use it, the asset must hold stored energy at that moment, and have cycling budget left, and find the balancing price better than the wholesale alternative (§5.2).

5.6.3 Zone coverage

All seven zones carry an MB component: the coefficient rests on the volume the zone activates, which exists independently of storage participation. What varies by zone is the measurement of the call depth d̄: it is measured directly where a multi-unit storage sample supports it, in North (12 units), Centre-South (3) and Sardinia (6), and carries the pooled value (0.377) in Centre-North, South, Calabria and Sicily, where storage awards are absent or rest on a single unit. The pooled bound is declared discretion (§4.3) and its effect is small: the pro-rata term, not the depth bound, sets the cap in 84% to 100% of activated periods depending on the zone and direction. Restricting the fleet to its regular participants roughly doubles the pro-rata term, so the depth bound engages more often than it did. On the upward leg it binds in under 2% of activated periods everywhere except Centre-North, at 3.2%. On the downward leg it reaches 15.8% in the South and 8.1% in Centre-North, and stays under 3% in the other five zones.

5.6.4 Fleet growth and publication timing

Because the fleet is the denominator, the cap per MW declines as the zone's fleet grows: the volume requiring balancing is set by demand, forecast error and outages, not by the quantity of flexible capacity available to serve it. Storage that qualifies and is awarded volume often enough to meet the participation condition joins F_z (§4.3). Each settlement period is computed against the fleet active at that time, so published history is not rewritten by a later, larger fleet.

GME publishes the per-unit award records in whole-month batches approximately two months after delivery (observed at two to two-and-a-half months). The Index therefore publishes in a single final edition at M+2, once the month's batch is complete (§2.5).

5.6.5 Measured levels: MB

κ is a per-settlement-period series. The table below summarises it over the MB measurement window, July 2025 to June 2026. "Periods with activation" is the share of settlement periods in which the zone activated volume in that direction; the mean cap is the average share of rated power available in those periods.

MB (mFRR):

Two features of the distribution. Activation is far more frequent than storage awards alone would suggest, with zones activating upward volume in 5% to 47% of settlement periods, but the pro-rata share in a large fleet is small: the North's 4.2% mean cap against Sardinia's 9.2% is the fleet-size effect (5,604 MW vs 1,140 MW), and it is what differentiates zonal balancing revenue. The distribution is also strongly skewed: a minority of large activations carries much of the volume, which is why the cap is applied per settlement period rather than as an average.

In thin zones the net MB figure for a period can be small or even negative: the downward leg is an energy purchase (paid recharge), and its value is realised on the wholesale legs rather than inside the MB line itself.

5.6.6 Measured levels: aFRR energy

The aFRR coefficient is built with the same formula from GME's per-unit aFRR award publication, over the PICASSO window, 25 November 2025 to June 2026: N is the zone's activated aFRR power, F_z the zonal aFRR fleet (reserved band per awarded unit, recomputed each month; the table shows the June 2026 fleet), and d̄ the measured storage call depth in aFRR (0.25 pooled; measured December 2025 to May 2026, one month inside the activation window below).

One construction difference from MB, in the direction of understatement: N for aFRR is the hourly mean activated power, applied to each of the hour's four settlement periods, where MB reads N at native 15-minute resolution. aFRR activations are near-continuous but shallow (see the table), so the hourly mean is close to the periods it replaces; the effect is to flatten intra-hour peaks rather than to change the level. A 15-minute reconstruction is under review.

Six zones record activation; Sardinia records none and carries a zero, an absence of observation, reviewed as participation develops.

aFRR is near-continuous where it runs (the North activates in 89% of hours) but shallow: the mean cap sits at 1 to 6% of rated power. MB is the opposite, episodic and deeper. The two streams are modelled independently, each against its own activated volume, and both compete with wholesale energy for the same state of charge and cycling budget (§5.4, §5.5).

The fleet denominators are refreshed as further units are awarded (§4.3). The cap min(κ, 1) × P^max applies in every regime window (§5.7). MSD carries a coefficient of zero (§3.2).

5.6.7 Scope of the coefficient

κ bounds the balancing volume the Index asset may dispatch; it does not predict which individual unit wins a given activation. Every megawatt of the flexible fleet receives the same pro-rata claim, so the coefficient represents the average awarded unit rather than the best-performing operator. Both statements are checked against the award data: the acceptance rate of storage offers relative to the fleet average measures 0.84 to 1.22 across the multi-unit zones, showing no systematic preferential activation in either direction, and the call frequency implied by the formula is compared with the frequency at which real storage units are called at each refresh (§5.6.2, property 2).

κ is defined as a share of rated power and applies identically to the 1-hour, 2-hour and 4-hour Index variants: duration does not change a unit's pro-rata claim on activated volume, and the effects of duration, meaning how long the permitted volume can be sustained, are carried by the physical model (state of charge, cycling limit, §5.4–§5.5), not by the coefficient.

One effect the coefficient does not carry is the size of the asset. κ is a share of rated power, so it reads the same for an asset of any size, while observed participation does not. Holding the zone constant, the share of activated periods in which a storage unit is awarded rises close to proportionally with its rated power: in the North the largest units are called an order of magnitude more often than units a tenth of their size, at a comparable bid price and a comparable vintage. The mechanism is that the system operator meets a lumpy requirement in merit order, an average upward call in the North being a few hundred megawatts spread over a handful of units, so a small unit is seldom worth calling. The Index represents a 50 MW asset, close to the median rated power of the Italian storage units that have won mFRR volume. A custom asset materially larger will have its balancing revenue understated on this basis, and a materially smaller one overstated. Making the coefficient conditional on rated power is a candidate for a future version.

Two conservatisms are retained: the volume-weighted average accepted bid (§5.8) is below what a strategic bidder captures, and the cycling limit (§5.5) makes balancing compete with wholesale energy for the same daily throughput, so the asset does not always take the full permitted volume.

5.7 aFRR energy

The published history begins December 2025, entirely within the PICASSO regime. Activated volumes and prices for the PICASSO window are read from GME's per-unit aFRR award publications; the ENTSO-E Transparency Platform series for Italian aFRR is not used, as its publication coverage is materially incomplete. The national-scheme treatment applies to internal history and custom studies only, and carries a stated limitation: it is computed from GME's zonal results series rather than per-unit awards.

5.8 Pay-as-bid pricing (MB)

MB is pay-as-bid: each accepted offer is paid its bid price, so there is no single clearing price. The Index uses the volume-weighted average accepted mFRR bid, computed by Modo Energy from GME's per-unit awards, per zone, direction and settlement period, over the same accepted offers that form the κ numerator.

It is neither the highest accepted bid, which would assume bidding at the ceiling and still clearing, nor the marginal accepted offer. GME's aggregated zonal price series is not used: it mixes mFRR with redispatch and other services priced on a different basis.

5.9 Revenue calibration

After §5.3, one leg is still optimised with hindsight: day-ahead. Every other leg is either decided on smoothed prices (MI-A2, XBID), bounded by observed activations and priced at observed accepted bids (MB, aFRR), or zero (MSD). The calibration therefore applies to the day-ahead leg only:

A measured floor, and a declared judgment above it. A pre-declared family of ex-ante day-ahead bidding strategies (persistence/climatology blends, same-weekday means, an exponentially weighted profile and an adaptive picker, all walk-forward, using only information available at D-1 gate closure) is replayed on the Index's own day-ahead price inputs. Each day the asset's dispatch is decided on the strategy's forecast and settled at realised prices; percent-of-perfect is settled revenue over perfect-hindsight revenue. The best strategy reaches 0.825–0.866 depending on zone (fleet-value-weighted 0.84). These strategies use calendar and price history alone, with no weather or load forecasts of the kind real desks hold, so the frontier is a floor on a competent desk's capture, not an estimate of it. The factor is set at 0.90: the gap between the measured floor and the factor is the calibration's one declared judgment, standing in for the desk information the measurement cannot replay. It is applied as a single scalar across zones (never per-zone), in the production pipeline before publication, so published Index values are already calibrated. The frontier is re-measured at each quarterly parameter refresh and the judgment margin re-examined against it; reviewed annually by Modo Energy's benchmark oversight function; a change beyond rounding follows §7.3.

Why no other leg carries a factor. The MI-A2 and XBID legs already embed imperfect foresight in the optimisation itself (§5.3), and XBID is additionally the balancing counter-leg under the continuous-trading tether (§5.4): deflating it while balancing revenue stands would misstate the net margin of balancing trades. The balancing legs are anchored to observed activations and accepted prices, with their residual hindsight declared in §5.3. Deflating any of them would double-count a correction already made at source.

What the factor does not cover. Asset availability and execution costs are outside the Index's definition: it is a gross market-opportunity benchmark of an available, well-run asset (§1.1, §2.2), not a net-realised-revenue estimate for any operator.

6. Governance and compliance

Modo Energy is committed to transparency across calculation methodologies, revenue components, and updates. Measures include:

• Publication of methodology documents.
• Historical data updates.
• At least 30 days' written notice of material methodology changes (§7.3).
• Documentation of stakeholder feedback, available on request via team@modoenergy.com.

7. Methodology changes

7.1 Stable Version

Each published Index value is produced by a numbered Stable Version of this methodology and of the model that implements it. The Stable Version in force is recorded alongside every published value, so any figure can be traced to the exact rules and model build that produced it (§8.4).

Modo Energy updates the Stable Version no more than once per calendar quarter. Changes in market structure outside Modo Energy's control, such as new products or changes to gate-closure timing or granularity, may require an update ahead of that cadence. The Stable Version in force at first publication is 1.0 (Appendix I).

7.2 Review process

The methodology is reviewed at least annually by Modo Energy's benchmark oversight function. A review is also triggered by:

• A change in Italian market design or regulation (GME or Terna market rules, ARERA resolutions), where the day-ahead move to 15-minute granularity on 1 October 2025 is the worked example.
• A change in the availability or definition of a source data series (§4.2).
• An excluded revenue stream reaching maturity (§3.2), currently MACSE and the Fast Reserve pilot.

7.3 Notification of changes

A Stable Version update is material where the cumulative movement in the Index across all Stable Version updates in that calendar quarter would exceed ±2%.

Material methodology changes are published with at least 30 days' written notice before they take effect, together with an impact analysis. The notice states the change, its rationale, its effective date, and whether previously published values are affected (§7.4). Stakeholder feedback received during the notice period is documented and available on request (§6). All methodology changes are recorded in Appendix I, so historical comparability can be traced.

7.4 Recalculation and restatement

A methodology change applies prospectively from its effective date; published history is not restated by default. Modo Energy restates published values only where a change corrects an error or a material data-quality issue (§8.3). Restatements are notified per §7.3, and both the superseded and the restated value are retained.

8. Consistency and continuity

8.1 Quality assurance

• Automated validation checks on input data.
• Automated regression tests on every change to the optimisation model.
• Quarterly manual audits of data sources and methodology compliance.

8.2 Data integrity

• Secure data management, access controls, regular backups.
• Clear traceability from raw source (GME, ENTSO-E) into the model feed.

8.3 Handling data quality issues

1. Verification with the upstream provider. For example, the month-scale publication holes found in the ENTSO-E Italian aFRR series, which led to sourcing aFRR activation from GME's per-unit AFRE databooks instead (§4.2). A second example: GME's silent republication of the January 2026 aFRR databook, which revised activated upward volumes from about 28 to about 68 GWh, was caught by cross-checking against ENTSO-E's national activation series, re-downloaded at source, and incorporated before first publication.
2. Customer communication of confirmed issues and the corrective action taken.
3. Data unavailability: customers notified; the Index publishes once finalised data is available.

8.4 Traceability and verification

• Comprehensive records of input data, Stable Version (§7.1), and outputs.
• Reproducibility: every published value can be reproduced from archived inputs and the Stable Version in force when it was published.
• Public disclosure of material methodology changes (§7.3).

Appendix I — Methodology changes

Methodology changes since first publication are tracked here. Version numbers correspond to Stable Version updates of the model (§7.1).

Disclaimer

This document, including the methodologies and indices described herein, is the proprietary work of MODO ENERGY LIMITED (“Modo Energy”) and is provided solely for informational purposes. These indices are designed for use in financial analysis, benchmarking, and decision-making. They do not constitute investment advice or a recommendation regarding any specific financial instrument, asset, or strategy.

While Modo Energy strives to ensure the accuracy, reliability, and transparency of the indices and methodologies, all information is provided “as is”, without any express or implied warranties, including but not limited to warranties of merchantability or fitness for a particular purpose. Users should be aware that the indices are derived from publicly available market data that may be subject to revisions, delays, or inaccuracies, and that the Italy Indices are simulated from a representative asset rather than observed from underlying transactions. Past performance is not indicative of future results, and external factors such as regulatory changes, market conditions, and asset-specific characteristics may impact index performance.

Modo Energy encourages users to conduct their own due diligence and consult with qualified professionals before making any investment or operational decisions based on the indices or methodologies herein. Modo Energy disclaims any liability for direct, indirect, incidental, or consequential losses or damages arising from the use of the indices, methodologies, or related data.

It is not possible to invest directly in an index. Indices are intended to represent performance benchmarks, and exposure to an asset class represented by an index may be available only through separate investable instruments. Modo Energy does not sponsor, endorse, or manage any financial products that aim to track the performance of its indices.

All intellectual property rights to the indices and methodologies are owned by Modo Energy. Unauthorised use, reproduction, or redistribution of this document, in whole or in part, is strictly prohibited without prior written consent from Modo Energy.

This document and the indices it describes are subject to updates and revisions. Significant changes will be communicated to stakeholders as appropriate. For further information, including licensing inquiries, please contact Modo Energy directly.

By accessing or using this document, you acknowledge and accept the terms of this disclaimer.