Boutique consulting
for commodity trading
risk management

We developed a power load prediction tool for ERCOT market. This tool can predict both short term and long-term power load for all ERCOT zones, based on weather forecast data. We trained our ML model with ERCOT historical load data as well as historical weather data. We continue to train our model as latest load and weather data becomes available. Please email us for a live demo! ***We are currently upgrading our tool to use ERA5 source data for top qualify weather data input ***

Client was migrating from direct system integration to a decoupled architecture. We designed and developed an Azure serverless function — triggered at configured frequency — that extracts AR/AP data from Endur, generates a CSV file, and saves it for Oracle ERP ingestion. The design is completely source and target agnostic, making it fully compliant with the decoupling principle.

Client's in-house web app for trade capture and risk calculation needed a new technology platform to handle future trading volume growth. We prototyped a container-based risk engine running on Apache Spark. Benchmark testing showed a 20% performance improvement with only 2 nodes deployed on a 16 GB memory laptop.

A Power Utility Company signed a new PPA to store and sell solar power. We created a custom instrument to model the contract, designed custom charging and discharging valuation curves, and implemented a complex daily settlement price calculation across 4 dispatch scenarios to determine the daily settlement price on these curves. We designed a swap-based instrument whose receive side is mapped to the discharge curve, pay side is mapped to charging curve. We created a new price loading interface for curve population, as well as custom settlement views, DMS invoice templates for settlement team and new accounting rules to ensure correct journal posting for accounting team.

Client deployed Snowflake as their enterprise data lake to centralize data from all trading systems. We developed custom SQL queries to extract reference, contract, market data, trades, settlement, and accounting data from Endur — serving as the data source for Endur data pipelines feeding into the Snowflake platform via the client's in-house ETL framework.

Client wanted to deploy Salesforce as the system of record for customer accounts within their trading ecosystem. We created a MuleSoft-based integration interface that synchronizes data changes between the source system and Salesforce in real time — ensuring consistent, up-to-date customer data across the entire trading operation.

Client was upgrading their Natural Gas trading desk from Endur V9 to V25. In V9, pricing formula templates were unsupported — the client had built custom user tables to automate fee generation. We rewrote their legacy scripts to leverage V25's native formula pricing templates, fully refactored the codebase to adhere to object-oriented design principles, and ensured compliance with all internal SONAR rules for optimal performance.