Price the Tail. Bid the Curve.
Battery revenue in ERCOT is not earned on the average day. It is concentrated in scarcity hours, price spikes, and the spread between day-ahead and real time. We forecast the full distribution of prices, value every megawatt-hour in your battery against it, and turn that value into the offer curves you actually submit.
What We Do
Strategy, architecture, shipping, and continuous improvement. Pick what you need, or let us take it end to end.
Day-Ahead Bid Optimization
Our optimizer solves the storage bidding problem the way it is actually structured: a stochastic program whose decision variables are the bid curves themselves, fed by fat-tailed price scenarios and a state-of-charge value curve computed by backward induction. Energy and ancillary services are co-optimized, not bolted together. Offer curves are ready before the day-ahead deadline, with a full audit trail of why each segment is priced where it is.
Probabilistic Price Forecasting
A point forecast cannot price a battery, because the money is in the tails. We forecast calibrated distributions of day-ahead and real-time prices with spikes modeled explicitly, then generate scenario paths that preserve the temporal structure arbitrage actually depends on. Calibration is verified continuously: when we say P90, prices exceed it ten percent of the time.
Real-Time Re-Optimization
The day-ahead award is a position, not a plan. Our rolling-horizon real-time engine re-forecasts and re-optimizes around the frozen day-ahead book as conditions change, trading deviations when the real-time price justifies them and protecting state of charge when it does not. Every real-time decision respects your physical limits, cycling budget, and ancillary service commitments.
Advisory & Revenue Audits
Not ready for an autobidder? We start with an audit: a walk-forward backtest of your battery's realized revenue against what a disciplined strategy would have captured, using only information available at each decision time. Attribution separates forecast misses from execution misses, so you see exactly where revenue was left. We also advise on market entry, ancillary strategy, and offtake structure.
How It Works
A daily loop from market data to settled offers, with honest measurement at every step.
Forecast the Distribution
Every morning we build calibrated distributional forecasts of day-ahead and real-time prices: quantiles for the body, an explicit spike model for the tail, and scenario paths that preserve the hour-to-hour structure a battery arbitrages. Point-in-time discipline is absolute: no forecast ever sees data that arrived after the decision it feeds.
Value Stored Energy
Backward induction over the price scenarios produces the marginal value of energy in your battery at every hour and state of charge. That value curve is the object everything else is derived from: it tells the optimizer when a megawatt-hour is worth holding, releasing, or committing to ancillary services.
Construct the Offers
The value curve becomes compliant offer curves for energy and ancillary services, with your physical limits, cycling budget, and risk rules enforced as constraints rather than afterthoughts. Curves are delivered before the day-ahead deadline, and the real-time engine keeps re-optimizing around the frozen day-ahead book as the operating day unfolds.
Settle & Learn
After settlement, every decision is attributed: what the forecast said, what the market did, what the strategy captured versus published benchmarks and the hindsight optimum. Forecast misses are separated from execution misses. Recalibration is continuous, and you see the same numbers we do.
Talk to Us
Tell us about your assets and how you bid today. We'll get back to you within one business day.