Industry
Power & infrastructure
Client
Marubeni Corporation
Solution
Clean Energy & BESS
About the client

Supporting Marubeni's commitment to disciplined capital allocation

Marubeni is a global trading and investment conglomerate with deep roots in power, infrastructure, and energy. The company expanded its U.S. wholesale power trading business through strategic investments in battery energy storage systems (BESS) and second-life EV battery innovation.

Consistent with its disciplined capital allocation strategy, Marubeni rigorously assessed the revenue potential of solar-plus-storage trading operations in the CAISO market supported by an AI-powered energy trading optimization platform. The company needed to determine whether projected revenue capture levels were achievable using real market data, operational constraints, and historical trading conditions.

In CAISO, power trading success depends on precision, speed, and the ability to respond to rapidly changing market conditions. Price volatility, asset operating constraints, and increasingly complex market dynamics challenged traditional analytical tools and manual trading processes, making it difficult to consistently identify and capture optimal trading opportunities.

The measurable impact of intelligent energy trading

80 - 85%

Percentage of perfect foresight (PoP) revenue.
Validated revenue potential using real CAISO market data, operational constraints, and historical trading conditions.

100% 

Regulatory compliance
Maintained compliance with utility regulations while incorporating battery health into trading decisions.

10% 

Improvement in PoP Revenue 
Improved PoP revenue from 75% to 85%, alongside lower imbalance penalties

10x

 Faster decisions
Accelerated decision cycles through decision-grade forecasting at 15-minute intervals and automated MLOps.
The business challenge

A trading model that had to keep pace with volatility, physics, and scale

Marubeni explored solar-plus-storage trading across CAISO, creating an opportunity to use advanced intelligence to navigate fast-moving market conditions, asset constraints, and increasingly complex bidding requirements.
Frame 4270
↑-1
Volatile market conditions
Prices swung sharply, sometimes below zero, while 15-minute markets left little room for delay.
↑-1
Physical asset constraints
Battery limits and degradation shaped every bid, while second-life assets added uncertainty.
↑-1
Complexity at scale
Bids had to balance forecasts, constraints, and penalties, while fragmented data slowed reaction time.

Where friction was building

As Marubeni explored solar-plus-storage trading in CAISO, the complexity extended beyond forecasting alone. Solar variability, fast-moving market signals, battery constraints, and fragmented data all had to come together quickly enough to support confident trading decisions
Frame (6)-1
Variable generation signals
Solar variability made stable trading signals harder to establish. Changes in solar output introduced uncertainty into the generation profile that trading decisions needed to account for.
Frame (12)
Fast-moving market conditions
Market signals moved faster than the operating limits of the assets. With CAISO markets operating at 15-minute intervals and prices moving sharply, the window for responding to changing conditions was narrow.
Frame (13)
Battery operating uncertainty
Battery behavior added another layer of predictability to every trading decision. Operating limits and degradation shaped each bid, while second-life assets introduced additional uncertainty.
Frame (22)
Fragmented decision inputs
Fragmented data slowed the path from market signal to execution. As forecasts, asset constraints, and penalties had to be considered together, disconnected information made rapid reaction more difficult.
The solution

What we built

Together, Nagarro and Marubeni co-developed an AI-powered trading solution that empowers electricity traders to make confident, data-driven decisions-balancing profitability, operational feasibility, and long-term asset health.
Price forecasting
  • DAM + FMM forecasting for CAISO
  • Enabled defensible market positioning at 15-minute speed.
Solar PV forecasting
  • Weather + satellite + production signals.
  • Reduced generation uncertainty and imbalance exposure.
Bid optimization
  • Multi-product bid schedule optimization.
  • Balanced profit, penalties, and battery limits in every bid.
AWS Cloud-native trading backbone
 Designed for high-frequency forecasting, scenario simulation, and secure trader access. 
Electricity market intelligence layer
  • Market signals: historical prices, grid congestion, ISO rules.
  • Energy signals: PV output profiles, irradiance, weather forecasts.
  • Economic signals: demand patterns, fuel indices, bidding strategies.
  • Asset signals: battery constraints, SoC behavior, degradation limits.
Frame 4267

Real market data.
Validated revenue potential.
Greater clarity for capital decisions.

 

What it unlocked

Marubeni set out to understand whether the projected revenue potential of solar-plus-storage trading in CAISO could hold up under real market conditions. By testing the opportunity against actual market data, operational constraints, and historical trading conditions, the engagement provided a grounded basis to assess commercial potential and support disciplined capital allocation.

grid2
renewable-1
grid-1
 
Icon non-hierarchical (1)

A validated view of revenue potential

PoP revenue levels of 80–85% were validated using real CAISO market and operational data, giving Marubeni a grounded view of the commercial potential of the opportunity.
Icon intelligent

Capital decisions grounded in operating reality

Projected revenue capture was assessed alongside actual market conditions and asset operating constraints, helping Marubeni evaluate the opportunity within the realities of solar-plus-storage operations.
Icon global

A foundation built to scale beyond the first asset

The platform was delivered in six months and built to scale beyond the first asset, creating a foundation that could extend across assets and markets.
The business outcome

Turning market intelligence into confident investment decisions

The platform gave Marubeni more than improved forecasting and trading performance. It created an evidence-based framework for evaluating the commercial viability of solar-plus-storage opportunities using real market behavior, asset economics, and operating constraints.
Commercial potential validated with real market evidence
Greater confidence in investment decisions
Historical and live CAISO market conditions could be evaluated against realistic operating scenarios, helping decision-makers assess whether projected returns remained attractive under real-world conditions.
Revenue opportunity translated into asset economics
Clearer understanding of value drivers
Market prices, solar generation, battery behavior, and trading strategies were evaluated together, providing greater transparency into what drives project economics and where value could be gained or lost.
Scenario-based evaluation of market opportunities
Better-informed commercial strategy
Teams could assess different market and operating scenarios before committing capital, enabling more informed decisions around bidding strategies, storage utilization, and future asset opportunities.
Reusable foundation for portfolio expansion
From a single use case to a scalable capability
The solution established a repeatable decision framework that could be extended to additional renewable assets and markets, supporting future portfolio growth without rebuilding the intelligence layer from scratch.
The technology stack

 

A technology foundation designed to turn infrastructure data into faster, more accurate maintenance decisions—helping the business scale predictive maintenance with confidence.
AWS Cloud-Native Trading Backbone
Designed for high-frequency forecasting, scenario simulation, and secure trader access
XGBoost

A price prediction under volatility.
Prophet
Time-series trend + seasonality forecasting.
LSTM
Sequential pattern learning for complex, non-linear time-series forecasting.