Perspective from the desk, the grid, and the data platform.
Your battery is paid at one node. Stop planning it at the zone.
Zonal and hub views average away exactly the congestion that determines a storage asset's return — and the nodes that pay are frequently not the famous ones. Capture rate at the node is the only honest measure of the edge.
A backtest with a perfect record saw the future
A trading signal earns trust by showing where it's wrong — its losing days, its confidence band, its drawdown. A 100% hit rate isn't reassuring; it's evidence the model was allowed to peek at the answer.
The grid's demand shock is showing up as nodal basis — and batteries are the natural arbitrageur
AI data-center load and a stalled interconnection queue are widening persistent price spreads at specific nodes. Energy arbitrage is already most of what a battery earns — and the edge is seeing that spread forward, at the node, not the zone.
One freeze made six percent of the year: underwrite storage revenue as event risk, not seasonality
A single freeze event produced 6% of a battery's entire annual arbitrage; the ten best days made a fifth to a quarter of it. That concentration is something to underwrite, not a seasonal pattern to weight toward.
Why regulatory reporting is a data problem before it's a compliance one
Filing season turns into a fire drill when the data is gathered by hand every time. The fix isn't more reviewers — it's a reporting foundation that makes "prepare a filing" and "answer an audit request" the same fast, traceable operation.
The revenue you already earned: finding leakage in meter-to-cash
The cheapest revenue a utility can find is the revenue it already earned but never collected. Leakage hides between the meter and the bill — and the win is less about a smarter model than about attribution you can act on.
Churn scores are cheap. Knowing where to spend retention is not.
A churn probability on every customer is easy to produce and easy to waste. The harder, more valuable question is which of those customers are worth keeping — and that means tying churn to lifetime value, not just risk.
Predicting outages starts with connecting OMS, AMI, and SCADA
Outage prediction gets pitched as a machine-learning problem. On most grids it's an integration problem first — the model is only as good as the OMS, AMI, and SCADA data you can actually bring together in time.
Seeing risk at every level of the desk: market and credit together
A single VaR number for the whole book hides more than it reveals. The useful view is risk decomposed — standalone, incremental, component — down to the trader, desk, and commodity group, with credit exposure sitting right next to it.
Where forecasting models actually earn their keep in day-ahead trading
Not every forecast improvement is worth the money. A look at where model accuracy translates into real day-ahead P&L — and where it quietly doesn't.
Catching the trade that becomes a fine before the regulator does
Trade surveillance built only on structured data misses the behavior that matters. The patterns that turn into regulatory exposure — collusion, cross-market benefit — live in the messy signals most systems throw away.
Segmentation that moves load, not just mailing lists
Most utility segmentation stops at demographics. The segmentation worth building is behavioral and margin-aware — because the point isn't a tidier mailing list, it's shifting load and putting the right product in front of the right customer.
Valuing a power plant like an option, not just an asset
A generation asset's worth isn't only what it produces — it's the optionality of when and whether to run it. Bringing market data into asset modeling turns a plant from a cost center into a position you can optimize.
Collections is a segmentation problem: cutting bad debt with better aging
Chasing every overdue account the same way wastes money and still writes off debt. Better collections start with aging you can trust in near real time and a segmentation that predicts who pays — and who won't.
What breaks first when you move ISO market data to the cloud
The migration rarely fails on compute. It fails on the assumptions baked into how ISO feeds arrive, reconcile, and get trusted — here's what to check first.
When better weather forecasts stop paying — and hedging starts
The instinct with renewable uncertainty is to buy a finer weather forecast. Past a point, the better return is treating weather as a risk to hedge — with derivatives, insurance, and models that explain the variance.
An authenticated view of the fundamentals your desk can actually trust
Every desk has data. Fewer have a fundamentals view they trust enough to act on without re-checking it. The competitive edge in market intelligence is less about more data than about a single authenticated source and the historic trends behind it.
Which customers buy the EV first? Competitive intelligence for a changing grid
Electrification, rooftop solar, and efficiency are rewriting load profiles customer by customer. The utilities that see it coming aren't guessing — they're predicting adoption at the segment level and planning load and revenue around it.
Shadow settlement: knowing your P&L before the ISO tells you
Waiting for the market operator's settlement to learn true profitability is settling blind. Shadow settlement reconstructs charges and credits at the transaction level, so profitability attribution is something you calculate, not receive.
Scorecarding fuel suppliers against the whole market, not last year's contract
Judging a fuel supplier against your own prior contract tells you whether you improved, not whether you're competitive. Benchmarking suppliers against market-wide data — and scorecarding their risk — is where procurement cycles and cost both come down.
Trimming the fuel bill without blowing the emissions budget
Fuel cost and emissions pull against each other, and optimizing one in isolation usually worsens the other. The win is correlating supply, transport, burn mix, and emissions in a single view — and optimizing the budget across all of them.
Scheduling the crew for the outage you haven't had yet
Field service sits on mountains of data and still schedules reactively. Predictive scheduling matches the right task-and-resource combinations to demand — including the unplanned outages that haven't happened yet.
Making gross-margin and cash-flow forecasts boring (in the best way)
Gross-margin and cash-flow forecasts are late and manual because they're rebuilt from scratch every cycle. Sitting them on an integrated financial data layer makes them timely, accurate, and — finally — routine.