Stromfee DV Audit Methodology · Whitepaper v1.1

Forensics of Direct Marketing (DV) performance for renewable plants from 500 kW
Lotter-Schweighof-II as reference case · Methodology globally applicable

For German operators, the German master version carries the canonical legal language. This English version is the operational + methodological translation for non-German audiences.

Executive Summary

Direct marketers (in German: Direktvermarkter, DV) earn revenues by selling renewable generation on spot markets and pass a contractually fixed share to the asset owner. The pass-through methodology is typically not specified in the contract text but left to the DV's internal interpretation. The German EEG §51 peak-shaving rule additionally permits a volume exclusion during negative-price hours that, under ambiguous interpretation, works against the asset owner.

This whitepaper documents a reproducible audit methodology to quantify DV performance against a fair multi-market optimum. Reference case: PV plant Schweighof II (9,998 kW gross, operator ABJ Solar GmbH & Co. KG, DV Trianel under contract V24PV033 dated 2024-05-06).

Key findings for 2025 (ClickHouse-verified):

The delta between Stromfee forecast and Trianel payout (≈ €339,385/year) is not fully attributable to DV methodology — it includes spread effects, balance group costs, market premium calculation, and §51 exclusion. Our forensics decompose this delta into three categories (see Section 3). A first damage estimator corridor lands at €17,000 – 35,000/year under explicit assumptions (Section 3 Step 5) — NOT a final claim figure.

Business consequence: Verified over-declaration beyond the §51 EEG mandatory volume creates a disclosure and potentially damages claim by the operator against the DV. The methodology directly applies to other German renewable assets and to EU/UK/AU/US markets after substitution of regulatory anchors.

1 · The Problem: Structural Audit Vacuum

Three independent research paths conducted on 2026-05-24 (German-language operator forums, English-language global communities, DV performance comparison platforms) converged on a consistent picture:

1. No public asset-level DV-revenue database exists for Germany. Modo Energy ("ME BESS DE", May 2026) confirms this in writing: "Public asset-level operational data, which underpins Modo's GB index, doesn't exist in Germany." (Source: modoenergy.com)

2. Industry aggregates (Bundesnetzagentur Monitoring Report 2025, BMWE Direct Marketing Annual Report 2024 by Energy Brainpool + Öko-Institut) report only by market segment, never by DV name, never publishing median outperformance per provider.

3. Operator forums (Photovoltaikforum, Akkudoktor) are dominated by residential users. The Twaice "BESS Pros Survey" Q1/2026 quantifies the gap: 47 % of operators consider supplier performance not enforceable, 50 % lament missing data foundation, 45 % experience monthly issues.

4. Structural causes: NDA clauses in DV contracts prohibit revenue publication. Whoever finds trading edge does not publish it. Industry associations (BVES, BSW, BWE) are bilaterally staffed and structurally block operator-vs-DV conflicts.

Consequence: As of research date 2026-05-24, no published, reproducible methodology for forensic DV-performance auditing at asset level for German renewable operators was found. We do not claim such methodology does not exist — we observe that none is published, hence none is reusable for other operators.

2 · Reference Case: PV Schweighof II

• Plant: MaStR SEE945995639982 · 9,998 kW gross · full feed-in · COD 2024-07-16 · Höhengauer Weg, 92224 Amberg, Bavaria · DSO Bayernwerk
• Operator: ABJ Solar GmbH & Co. KG, Schweighof 1, 92224 Amberg (Lotter family)
• Direct Marketer: Trianel GmbH, Krefelder Straße 203, 52070 Aachen · contract V24PV033 of 2024-05-06 (PV market premium under EEG-2023) · supplemental agreement 2025-06-29 (methodology change to clause 5.3 referencing BNetzA BK 6-20-059 + BDEW RD 2.0 — litigation-relevant)

Data sources (audit input):

• 5-min meter data from utility meter Lotter Veitsbronn 863517 (lotter.lastgang_5min, 89,747 slots 2025-05-23 → 2026-03-30)
• Day-Ahead prices from ENTSO-E (current through tomorrow 21:45 UTC)
• Trianel monthly invoice data (lotter.bilanz_2025_betreiber, 12 months 2025)
• §51 block forensics from Stromfee pipeline (lotter.block_def_forensik, 23 months of analysis)

Contract clauses (audited 2026-05-10, lotter.dv_vertrag_klauseln, 16 clauses): max applicable EEG value 71 €/MWh · contract model Market Premium Code 689 · peak-shaving (Spitzabrechnung) Annex 5 SELECTED · special compensation Annex 6 EMPTY · balance group Trianel MPM · global radiation supply obligation YES · German law · jurisdiction Aachen.

3 · The Generic Methodology in 5 Steps

The methodology is market-agnostic. German-specific terms are in parentheses.

Step 1 · Reconstruct ground-truth generation

Input: 1-minute or 5-minute meter data from certified instrument (utility meter or independent power analyzer like Janitza UMG / CA8336).
Output: Hourly energy generation per plant (kWh).

Hourly aggregation via sum(power_w × Δt / 3600) / 1000. Spikes above plant rating are flagged as measurement error (excluded from audit). Curtailment events documented separately (DV-caused vs. DSO-caused).

Country anchor (DE): Bundeswide Metering Code (Messstellenverordnung) · TR3 load profile capture · BDEW MSCONS format.

Step 2 · Compute fair market valuation

Input: Hourly generation × hourly spot price (Day-Ahead) PLUS multi-market optimum from Intraday-Continuous, Intraday-Auction (IDA1/2/3), balancing energy (aFRR/mFRR Activation+Capacity), imbalance.
Output: Theoretical gross revenue maximum (€).

The multi-market optimum uses Perfect-Foresight (DA + ID-Full + IDA1+IDA2+IDA3) as upper comparison value. Important: Multi-market optimum is a theoretical maximum, not achievable by real DV performance (liquidity constraints). Hence comparison value = approximation, not duty value.

Step 3 · Model regulatory exclusion regime

Input: Hourly spot prices + statutory exclusion rule.
Output: Mandatory exclusion volumes (MWh) per month.

For Germany, EEG-2023 §51 applies: if more than 4 consecutive hours of negative spot price occur, the right to market premium for the entire calendar day is lost (literal wording: "Operators have no claim to market premium for the respective calendar day if … the hour contract was negative for more than 4 consecutive hours"). Operators with peak-shaving (Spitzabrechnung) can curtail in advance to avoid this trigger.

The 4h-mandatory volume is the unambiguously legally required exclusion share. Anything beyond that (e.g., pre-hour cushion, post-hour cushion) is negotiation room, not mandatory.

Step 4 · Three-source forensics

For each negative-price spot hour, three volume classifications are derived:

Forensic statement:

• Trianel-suspicion > EEG-4h-statute → over-declaration suspicion, claim-eligible
• Trianel-suspicion ≈ EEG-4h-statute → contract-compliant interpretation
• Trianel-suspicion < EEG-4h-statute → DV acted in operator's favor (rare)

For Schweighof II 2025: EEG-4h = 1,408.8 MWh · Trianel-suspicion = 1,078.7 MWh · SoK24 = 1,636.4 MWh. The SoK24 figure suggests the regionally applicable exclusion recommendation lies considerably above the statutory 4h-minimum — this requires legal clarification of the additional-volume rationale.

Step 5 · Damage extrapolation with confidence interval

Assumptions for Schweighof II 2025 (explicit, simplifying):

(a) Average lost market premium per MWh is conservatively set at €20/MWh. Real market premium varies by Solar quarterly market value (BNetzA-published) and by §51-avoidable hours distribution. A range €20-25/MWh is more plausible than a point value.

(b) Model assumes 100 % of EEG-4h-mandatory volumes would have been paid in absence of over-declaration with functioning peak-shaving. This is a theoretical maximum — real realization rate likely 60-90 %.

(c) Balance group costs and forecast penalties are not included and must be added per-plant.

Damage estimator corridor Schweighof II 2025 (with assumptions above):

• Lower band: 1,408.8 MWh × €20/MWh × 0.6 realization ≈ €17,000/year
• Middle: 1,408.8 MWh × €22.5/MWh × 0.8 ≈ €25,000/year
• Upper band: 1,408.8 MWh × €25/MWh × 1.0 ≈ €35,000/year

These figures are indicator, not final claim figure. Legal preparation must additionally consider interest, attorney and court costs, plus the specific contract setup (peak-shaving vs. flat-rate).

4 · Global Applicability · Translation Matrix

The methodology is market-agnostic. Only the regulatory anchors (Step 3 + 5) are country-specific:

Recommended adaptations per market:

• NL/BE/LU: Similar exclusion regime to DE, but no explicit 4h-anchor. Forensics adaptable once local anchor is defined.
• UK: Ofgem CfD-Clawback more transparent (Strike-Refund vs. Tag-Index), but aggregator performance still NDA-protected. Methodology directly applicable.
• AU (Hornsdale pattern): Neoen quarterly investor reports already provide asset-level data — methodology serves to validate these reports against spot + AEMO data.

5 · Reproducibility

The companion open-source repository stromfee/dv-audit-de (Apache-2.0, planned 2026-06) provides:

dv-audit-de/
├── README.md                    EN quick-start
├── methodik/                    DE methodology docs
├── parser/
│   ├── mscons_parser.py         BDEW MSCONS format (Z36-Z39, Z74, ZT8, ZG3)
│   └── trianel_pdf_parser.py    Trianel monthly invoice PDF
├── lib/
│   ├── p51_detector.py          §51 4h-rule detector
│   ├── multi_markt_opt.py       DA + ID + IDA + balancing optimum
│   └── bootstrap_konfidenz.py   90 %-CI extrapolation
├── examples/
│   ├── schweighof_ii_demo.ipynb anonymized Lotter demo notebook
│   └── synthetic_pv_5mw.ipynb   fully synthetic test case
└── tests/                       ≥80 % test coverage

Apache-2.0 license permits commercial use with patent protection clause. Anonymization obligation remains (GDPR compliance).

6 · Limitations + Next Steps

Limitations:

1. Contract-clause variation: §51 interpretation is DV-specific. Three-source forensics provides suspicion, not proof — proof requires contract disclosure litigation.
2. n=1 in reference case: Schweighof II is one asset. Statistical validation requires ≥30 comparable audits — currently collected by Stromfee.
3. Balance group effects: Methodology does not fully separate spot damage from balance group damage. Significant BK contributions require separate audit layer.
4. Peak-shaving availability: Damage extrapolation assumes peak-shaving was contractually possible. For flat-rate billing, the §51-avoidability component is removed.

Next steps:

• 2026-06: Open-source repo stromfee/dv-audit-de v0.1 (MSCONS parser + §51 detector)
• 2026-07: Legal review of methodology by Pavlos Konstantinidis (vbvh Berlin, BWE Direct Marketing Advisory Board — request pending)
• 2026-Q3: EN translation of supporting materials + pilot acquisition NL and UK
• 2026-Q4: First published litigation pursuit on methodology basis (Lotter vs. Trianel) as precedent case

Sources

Stromfee-verified data (ClickHouse, as of 2026-05-24):
- lotter.lastgang_5min (89,747 5-min slots, 2025-05-23 - 2026-03-30)
- lotter.bilanz_2025_betreiber (12 months Trianel payout + Stromfee forecast)
- lotter.block_def_forensik (23 months three-source forensics)
- lotter.curtailment_events (23 events, 115,459 kWh loss)
- lotter.dv_vertrag (10 plan/actual fields) + lotter.dv_vertrag_klauseln (16 clauses)

Regulatory sources:
- EEG 2023 §51 (juris.de · gesetze-im-internet.de/eeg_2023)
- BNetzA Notification No. 12 (BK6-23-241) mixed-price methodology
- BDEW Redispatch 2.0 Implementation Guide v3.0
- BNetzA Monitoring Report 2025 (data.bundesnetzagentur.de)
- BMWE Monitoring Direct Marketing Annual Report 2024 (Energy Brainpool + Öko-Institut)

MaKo / MSCONS / EDIFACT fundamentals (upstream for §51 forensics):
- Enerchy by Kinergia · Pillar page market communication (Thorsten Zoerner / STROMDAO GmbH, 2025-01) — UTILMD/MSCONS/INVOIC/ORDERS format docs, GPKE processes, AI tool "Willi Mako" for EDIFACT analysis
- BDEW MaKo portal · edi-energy.de · stromhaltig blog (Zoerner, long-running source)

Market validation of the structural gap (research 2026-05-24):
- Modo Energy ME BESS DE launch May 2026 (modoenergy.com)
- Twaice BESS Pros Survey Q1/2026 (twaice.com)
- Konstantinidis vbvh Berlin · publication 33. Windenergietage 2025-11-12 (windenergietage.de)
- ARENA Large-Scale Battery Storage Knowledge Sharing Report (arena.gov.au)

Operator communities (research finding: gap real, no active DE discourse):
- Photovoltaikforum.com (90 % residential users, 1 relevant MW thread)
- BWE Working Group Direct Marketing (bilateral, non-public)
- 0 hits: Reddit DE-MW-BESS · LinkedIn groups · Discord · Telegram

License

This whitepaper is published under Creative Commons Attribution 4.0 International (CC BY 4.0). The companion open-source repo stromfee/dv-audit-de will be released under Apache-2.0.

Citation:

Stromfee / HR Energiemanagement GmbH (2026): DV Audit Methodology · Whitepaper v1.1. Bünde, 2026-05-24. URL: stromfee.ai/whitepaper/dv-audit-methodology-2026-05-24-en.html

Contact for adopters:
HR Energiemanagement GmbH · Holger Roswandowicz · Bünde, Germany · +49 5223 4921030 · stromfee.ai