Japan BESS Arbitrage: A 15-Year Reexamination

In 2025, a 1 MW / 4-hour Battery Energy Storage System (BESS) operating in the Tokyo area — charging during the lowest-priced settlement blocks of the day and discharging during the highest — would have generated theoretical annual arbitrage revenue of ¥10,466/kW. That figure is 69% below the 2022 peak (¥33,940/kW), but 13% above the 10-year average for 2011–2020 (¥9,243/kW).

JEPX arbitrage has reverted toward its long-term average since 2023, following the historically volatile period of 2021–2022. However, the landscape it has returned to is not the same as it was ten years ago. The ranking across the 9 areas has shifted substantially. This article organizes these structural changes based on a benchmark calculated from 15 years (2011–2025) of 30-minute JEPX spot market public data.

15-Year Trend

Theoretical 4-hour arbitrage revenue across all 9 areas from 2011 to 2025 shows the following pattern. The 10-year average for 2011–2020 was ¥8,964/kW/year, with all areas staying in a relatively narrow range of ¥6,000–¥14,000/kW. In 2021, values jumped sharply (average ¥21,012/kW/year), reaching a historical peak in 2022 (average ¥32,996/kW/year), before rapidly reverting to normal levels in 2023 (average ¥14,928/kW/year). 2024–2025 have stabilized in the ¥13,000/kW/year range (+47% vs. baseline).

Measured from the 2022 peak, current levels appear extremely low. But viewed across a 15-year timespan, it is more natural to interpret 2021–2022 as historically anomalous outliers, and to read 2023 onward as a reversion to near the long-term average.

The Anomaly of 2021–2022

The degree of anomaly can also be confirmed quantitatively through the annual intraday price standard deviation (intraday std). The ranking of average std across all 9 areas over the past 15 years is as follows:

The 2022 average std (8.15) is more than twice that of 2023 in third place (3.77). 2021 (5.25) ranks second, just behind 2022. These two years overlapped multiple structural factors: demand recovery from COVID-19, the power supply tightness alert in January 2021, the surge in LNG spot prices following the Russia-Ukraine war in February 2022, and fuel procurement constraints for domestic thermal power generation.

The Rise of Kyushu

The most dramatic shift in inter-area revenue is the surge in Kyushu.

Kyushu's increase vs. baseline is +85.8%, the largest among all 9 areas. Other areas range from +23% to +66%, making Kyushu clearly an outlier.

This is because the solar curtailment that became widespread from 2017 onward has structurally widened midday price suppression, pushing intraday spreads wider than in other areas. The frequency of settlement blocks priced at ¥0.01/kWh or below — as measured in this benchmark — is also highest in Kyushu across all 15 years (2–3 times that of other areas). Kyushu's relative advantage is tied to the fundamentals of renewable penetration and grid interconnection capacity, and is unlikely to reverse in the near term.

Hokkaido Loses Its Throne

The reverse of Kyushu is happening in Hokkaido.

Hokkaido was the historical champion of 4-hour theoretical arbitrage revenue, ranking #1 for 9 of 10 years between 2011 and 2020. In 2023–2025, Hokkaido's absolute level remains high at ¥14,474/kW/year, but its growth rate (+27.3%) is the lowest of all 9 areas — a relative decline as other areas post gains of 47–86%. The factors behind Hokkaido's historical advantage (high wind penetration, area price independence due to interconnection constraints) remain in place, but the pace of structural change happening in other areas appears comparatively faster.

Tokyo's Underperformance

Tokyo has ranked near the bottom of the 9 areas for much of the past 15 years.

The Tokyo area is structurally characterized by stable supply-demand conditions and thin intraday spreads, driven by factors including heavy reliance on thermal power without nuclear restarts, the largest power demand scale in the country, and relatively low renewable penetration. From a generation standpoint this reflects successful stable supply, but in the context of BESS arbitrage, it means fewer revenue opportunities.

Widening Spread Between Areas

For 4-hour theoretical arbitrage revenue across all 9 areas, the gap (in absolute terms) between the highest-earning area and lowest-earning area has clearly widened between the baseline period and the reversion period.

The ratio is roughly the same, but the absolute gap has expanded by +37%. As a concrete example, for a 1 MW BESS project, the annual revenue difference between building in the highest-earning vs. lowest-earning area expanded from approximately ¥3.45 million/MW/year (= ¥3,452/kW × 1,000 kW) in the baseline period to approximately ¥4.74 million/MW/year in the reversion period. Accumulated over a 20-year operating life, the implications of area selection amount to a cumulative revenue difference of approximately ¥100 million. The impact of site selection is larger than it was 10 years ago.

Summary

The findings of this article can be summarized in two points.

First, the spreads of 2021–2022 were historically anomalous outliers within the 15-year dataset. Since 2023–2025, values have reverted to near normal levels, averaging roughly ¥13,000/kW/year across all 9 areas.

Second, the area-level landscape shifted substantially around 2017. From 2011–2016, Hokkaido led the country in terms of spread; from 2017 onward, Kyushu has emerged as a result of accelerating renewable penetration. Inter-area revenue gaps in absolute terms have also expanded by +37% vs. the baseline period. Rather than concentrating in a single area, a multi-area siting strategy that accounts for future supply-demand trends is likely to be more effective over the medium to long term.

What This Benchmark Does Not Cover

This benchmark represents a theoretical upper limit and differs from the net revenue of an operational BESS. The main factors that create this gap are the following four points.

Foresight gap: The benchmark assumes perfect ex-post foresight. In actual operations, 30–50% is lost depending on forecast accuracy.

Round-trip efficiency: Assumes RTE = 100%. Practical Li-ion efficiency is 85–90%, resulting in a net reduction of 10–15%.

Degradation: Capacity fade is not modeled. In actual operations, annual capacity decline of 1.5–3% occurs.

Non-spot revenue not included: The DSAM, capacity market, and imbalance revenue are not reflected. In Japan today, DSAM revenue potentially accounts for 20–40% of total revenue.

Accordingly, actual net spot arbitrage revenue for an operational BESS is expected to fall in the range of 40–60% of this benchmark. Quantification of total revenue including non-spot sources will be addressed in v0.2 and beyond.

Calculation Methodology and Reproducibility

Data source

JEPX spot market settlement results

Coverage period

January 1, 2011 – December 31, 2025 (15 complete calendar years, approximately 5,475 delivery days)

Calculation formula

For each day and each area, 48 half-hour prices are sorted in ascending order. For N-hour duration, the daily spread is defined as the difference between the average of the top 2N blocks and the average of the bottom 2N blocks. Daily revenue = N × daily spread. Annual benchmark = sum of daily revenue for that year.

Granularity

Calculated at JEPX's native granularity of 30 minutes.

Assumptions

Ex-post optimal dispatch (perfect foresight, charging at the lowest price and discharging at the highest), 1 cycle/day, RTE = 100%, price taker, spot only, gross of tax / wheeling charges / OPEX, degradation ignored. Full assumptions are disclosed in the Methodology page.

Closing

This article is v0.1 of the Japan BESS Arbitrage Index (30-min) by Meridian138°. Based on 15 years of data, the scope covers only the theoretical maximum from the JEPX spot market. Future releases are planned to sequentially add an RTE 85% sensitivity variant, a DSAM revenue layer, and a realised revenue index integrating actual dispatch data.

Feedback on this benchmark is always welcome.

— Meridian138° / April 2026