Miss the monthly peak by a single interval and you pay for it for thirty days. PHALANX is built to not miss it, without wasting charge you didn't need.
It works out how likely a peak really is from your own interval data, not a rule of thumb. Peak-miss tail estimation by multiple-importance sampling.
Missing a peak is far more expensive than holding charge you didn't need. PHALANX is tuned for exactly that asymmetry. Asymmetric-cost dispatch objective.
It reads the load and temperature forecast to see the peak coming. Consumes the HELIOS forecast distribution.
During onboarding we validate the saving on your 12-month history, so you see how it holds up before you rely on it. Walk-forward validation on your own load.
During onboarding we validate the method on your own data — a 12-month walk-forward (non-overlapping folds, five benchmarks, Diebold–Mariano significance) — so you can see how it performs before you rely on it.
These are the real methods behind PHALANX, each in one plain line. We name the method as a credibility signal — we don't publish the recipe.
Estimates the true chance of a peak from your own load, not a heuristic.
Weights a missed peak far more heavily than a needless hold.