---
id: VOLT-HOME-WP-038
title: "Are home-battery savings spread across the year or concentrated in a few days?"
slug: are-home-battery-savings-spread-across-the-year-or-concentrated-in-a-few-days
description: "The top decile of positive zone-month-duration rows contributes 37.3301% of indexed value, showing concentration without identifying specific days."
published: 2026-08-30
cluster: "Home batteries"
status: measured
evidence_url: /research-data/home-papers/are-home-battery-savings-spread-across-the-year-or-concentrated-in-a-few-days.json
figure_url: /research-media/home-papers/are-home-battery-savings-spread-across-the-year-or-concentrated-in-a-few-days.webp
figure_alt: "Chart for Are home-battery savings spread across the year or concentrated in a few days?: share of positive monthly index value in the top decile of rows, shown as top decile, remainder."
source_ids:
  - energy-informatics-storage
  - acer-retail-2025
  - iea-electricity-2026
  - dynamic-tariff-viability
  - volt-research-content
peer_reviewed: false
---
## Abstract

The frozen evidence reports that the top decile of positive monthly BESS-v2 index rows accounts for 37.33012584790893% of positive monthly indexed value across 16,368 observations. The evidence interpretation defines the units as zone-month-duration observations. This demonstrates concentration in the aggregate distribution: 10% of positive rows contribute substantially more than 10% of positive value. It does not identify “a few days,” because the published metric is monthly-row concentration rather than daily-event concentration.

BESS-v2 is a normalized, per-power, wholesale-only index evaluated with perfect foresight. It excludes retail taxes, network charges, forecast error, and site-specific costs. The result indicates that averages depend materially on a high-value tail, but most positive value still lies outside that top decile. A practical annual plan should therefore report both central value and concentration, while avoiding the claim that particular household days generate the measured share. Daily concentration requires a separate daily aggregation with matched asset and zone identity.

## Plain-language answer

The value is meaningfully concentrated, but this paper cannot honestly say it is concentrated in a few days. The measured top 10% of positive monthly zone-duration rows contributes 37.33% of positive indexed value. The other 90% contributes the remaining 62.67% as an arithmetic complement. That pattern is far from evenly spread, yet it is also not a case where nearly all value comes from the top decile.

A row can represent a zone, month, and battery duration rather than one calendar day for one home. High-value rows may cluster in volatile months, specific zones, or longer-duration categories. Perfect foresight also makes the upper tail easier to capture than it would be with a forecast. The practical message is to stress-test annual averages against missed high-value periods and to preserve the exact unit of concentration. Calling these rows “days” would overstate what the evidence measures.

## Research question

The decision-relevant question is whether a battery’s annual value arrives steadily or depends on rare opportunities. High concentration matters because maintenance, outages, reserve requirements, or forecast misses during a small set of important periods can reduce realized performance disproportionately. It also makes short historical windows unstable: missing one unusual regime can change an average materially.

The registered primary metric ranks positive monthly index rows and calculates the share of positive total value contributed by the top decile. This is a standard concentration description, but its observational unit determines the answer. Because the rows are zone-month-duration combinations, the metric speaks to concentration across those combinations, not directly across days. This paper uses the title to motivate the issue while preserving the narrower evidence-backed conclusion.

## Data and provenance

The analysis uses frozen SELECT-only snapshot SHA-256 `7e97489fc8528c8cc8c38830e05b48d949ce1f67b98575dff26f5d7c321e4c67`, in a read-only transaction with a 180-second timeout. Registered source contracts are `day_ahead_prices`, `bess_index_daily`, `bess_forecast_daily`, `generation_mix`, and `capture_stats`. Daily prices cover 2021-01-01 through 2026-08-29; detailed intervals cover 2025-10-01 through 2026-08-29; and the long-history boundary covers 2015-01-01 through 2026-08-29. Publication cut-off is 2026-08-30T00:00:00Z.

The metric sample size is 16,368 and its unit is percent. Analysis code is SHA-256 `57c57de79cdab2b5b6d6c54c485cb5162598c5ba0b0bfe995da40d75e6c52ba9`, protocol `adb36bf6b447af9f96339249b8becaefc20422499cca1977242866347a97bd4b`, registry `7bcb91d7476d0a69fe9fa75a5c7782f8117e0153f82f9112b7e1d307d3943717`, and source registry `07949550ac443ff673fda5c0209b99f137544f3ffecf6775f109bb9d09663bd6`. The frozen evidence and figure hashes provide a public aggregate audit trail without exposing raw interval rows.

## Method

BESS-v2 normalizes battery index values per unit of power and uses perfect foresight of realized day-ahead prices. It therefore describes ex post market opportunity under common constraints, not what a forecast-driven household controller necessarily captures. Wholesale-only accounting excludes retail taxes and network charges.

For concentration, the analysis retains positive monthly index rows, ranks them by value, selects the top decile, and divides their sum by total positive value. The reported statistic is the share attributable to that top group. Negative rows are not part of the denominator described by the metric. The approach is descriptive. Within-family Holm control applies to inferential claims, but no unadjusted significance claim is made. The method does not identify causal drivers or convert monthly rows to days.

## Results

The top-decile share is 37.33012584790893%, rounded to 37.33%. This is 3.733 times the 10% share that the top decile would contribute under perfectly equal positive-row values, an arithmetic comparison rather than a new empirical estimate. The result shows a heavy upper tail in positive indexed value. The complement, 62.66987415209107%, remains outside the top decile, so the distribution is concentrated but not exclusively dependent on that tail.

The regenerated evidence reports `bootstrap_95_interval: null` and no interval method for this estimand. The figure publishes top decile = 37.33012584790893% and remainder = 62.66987415209107%, matching the arithmetic complement described above. No confidence interval is reported for the concentration metric; a future one would need a documented resampling unit that preserves dependence among zone, month, and duration rows.

No dates, zones, months, or durations are identified as top-decile members in the public JSON. The result cannot support a list of “best days” or a claim about how many days drive annual value.

## Robustness and placebo checks

Concentration should be recalculated with alternative units: day, zone-day, zone-month, and asset-year. It should also be shown with top 1%, 5%, 10%, and 20% shares, a Lorenz curve, and leave-one-regime-out checks. Those views would reveal whether the 37.33% share is driven by aggregation or a stable tail. The current evidence reports only the top-decile monthly-row statistic.

A useful placebo would randomize values across monthly rows while preserving row count; expected concentration would move toward the distribution implied by the randomized values, helping separate structural group effects. Another would calculate concentration under forecast-based dispatch. If perfect foresight disproportionately captures extreme days, forecast-based concentration may be lower in value or higher in dependence on rare correct calls. These checks are not in the JSON and are not presented as completed. The present paper’s main robustness safeguard is correct labeling of the observational unit and refusal to invent an uncertainty range.

## Limitations

The metric includes only positive monthly index value in its denominator. It does not net negative outcomes or show annual downside. Grouping by zone, month, and duration can cause large markets or duration categories to appear multiple times and can hide within-month daily concentration. A top-decile share is also sensitive to ties and sample composition, none of which is detailed in the paper-level object.

Perfect foresight overstates attainability of tail opportunities. BESS-v2 is normalized and omits retail taxes, network charges, degradation, efficiency variation, auxiliary use, and contract terms. The result is not a household bill measure or observed site cash. No interval is reported for the primary share. Most importantly, the evidence does not identify days, so the title’s daily wording cannot be answered directly.

A concentration metric should also distinguish opportunity concentration from capture concentration. Perfect foresight may identify the same rare high-value rows for every normalized asset, while forecast-based controllers could miss different subsets. Availability outages or backup reservations may further remove access exactly when the opportunity tail is largest. The resulting captured-value concentration can therefore differ materially from the opportunity-index concentration reported here.

Negative and zero rows deserve a companion view. Restricting the denominator to positive values answers how positive opportunity is distributed, but it does not show whether low or negative rows offset the upper tail over an asset-year. A complete risk report would present net value, positive-tail concentration, downside concentration, and leave-top-days-out sensitivity together. The current 37.33% statistic remains useful because its numerator and denominator are explicit; it should not be expanded into those unmeasured annual claims.

Cross-sectional and temporal concentration should not be conflated. If one bidding zone contributes many high-value month-duration rows, the aggregate top decile can be concentrated across geography even when each asset’s value is relatively steady through its own year. Conversely, an individual zone may depend on a handful of days without dominating the pooled monthly ranking. An asset decision needs the within-zone, within-duration time series, while a market comparison may legitimately use the pooled view.

The observation count also should not be treated as a count of independent months. Rows can share the same prices, month, zone, or duration and therefore be statistically dependent. A block bootstrap would need to preserve the relevant clustering. Since no interval is reported, this paper avoids an independence assumption and uses no p-value. The 37.33% result is a descriptive allocation of positive aggregate value only.

## Practical implication

When evaluating a battery case, publish concentration alongside average indexed value. Ask what happens if the asset is unavailable, reserved for backup, or scheduled incorrectly during the highest-value observations. A case that depends heavily on a narrow tail should use conservative availability and forecast assumptions and should not annualize a short favorable window without regime checks.

For this evidence, the right statement is that 37.33% of positive monthly zone-duration value sits in the top decile of rows. Analysts who need daily operational risk should recompute on daily asset-level rows. Controllers should also compare perfect-foresight and forecast-based concentration: missing a rare high-value period may matter more than small errors on ordinary days. Such reporting makes uncertainty visible without claiming that a particular home will encounter the same distribution.

## Reproducibility

Retrieve `/research-data/home-papers/are-home-battery-savings-spread-across-the-year-or-concentrated-in-a-few-days.json`. Verify status `measured`, value 37.33012584790893%, sample size 16,368, figure values, absence of an interval, metric wording, and interpretation. The manifest records evidence SHA-256 `d33027f25744a7365dbff27435a4bcb1c949d0dfa888bf0f9a49d965d14ac460` and figure SHA-256 `d6c8f6ab9650c70d911e82233500f706bbded6a530f8a221b6d106bc094a4820`.

Reproduction must use the exact snapshot, code, protocol, registry, windows, and publication cut-off. Preserve positive-row filtering and the zone-month-duration observational unit. Any daily concentration result is a new analysis and must not reuse this paper’s metric label. Licensing references are `/legal/data-licensing` and [Voltcast data licensing and redistribution](https://github.com/ossedk/voltcast/blob/main/docs/voltcast/LICENSING.md).

## Disclosure

Analysis and drafting were model-assisted. The difference between monthly rows and days, the absence of an interval, and all accounting boundaries are disclosed. This working paper is not peer reviewed. It is not trading advice, financial advice, an investment recommendation, or operational authorization. Voltcast has no live traders or live capital, and the concentration result does not authorize action.

## References

- `energy-informatics-storage` — Energy Informatics. [Risk and reward: evaluating household energy storage for optimizing demand-side flexibility under dynamic tariffs](https://doi.org/10.1186/s42162-025-00602-9). Kind: peer-reviewed.
- `acer-retail-2025` — ACER and CEER. [Rewarding flexibility: How retail contract choice can help unlock consumer flexibility](https://www.ceer.eu/wp-content/uploads/2025/11/ACER-CEER-2025-Retail-monitoring.pdf). Kind: official.
- `iea-electricity-2026` — International Energy Agency. [Electricity 2026](https://www.iea.org/reports/electricity-2026). Kind: official.
- `dynamic-tariff-viability` — Advances in Applied Energy. [Assessing the conditions for economic viability of dynamic electricity retail tariffs for households](https://doi.org/10.1016/j.adapen.2024.100174). Kind: peer-reviewed.
- `volt-research-content` — Voltcast. [Voltcast Research Content Plan](https://github.com/ossedk/voltcast/blob/main/docs/voltcast/RESEARCH-CONTENT-PLAN.md). Kind: canonical.