GENCO VPP

A continuously-updating 7-day price forecast for India's Day-Ahead and Real-Time electricity markets — a single best-estimate (median) price per block, re-corrected in real time as the market settles.

DAM RTM Live

What this is

Every fifteen minutes, our system pulls the latest cleared prices from the Indian Energy Exchange and refreshes a 7-day forecast covering both the Day-Ahead Market (DAM) and the Real-Time Market (RTM). For each block it issues a single best-estimate (median) price, continuously re-centred by a rolling bias correction and pushed toward the ceiling or floor by dedicated regime classifiers as market conditions shift.

Why a self-correcting point forecast

Power traders, gencos, and DISCOMs need a clear current best estimate for each block — and one that adapts as the day settles, not a single morning number left to go stale. Our forecast issues one median price per block and keeps correcting it: a rolling bias tracker re-centres each slot from the last 14 settled days, and ceiling/floor classifiers shift the estimate when the market enters those regimes. The result is a forward curve that tracks reality and updates every 15 minutes.

How it stays current

The forecast is not a static morning publication. It is regenerated whenever the market gives us new information:

• Every 15 minutes — newly settled real-time blocks are pulled in; the live actuals line on the chart extends as the day progresses. The rolling bias correction updates automatically, so the point forecast self-adjusts as the day unfolds.
• After each DAM auction publishes (around 3 PM IST) — the next day's cleared DAM prices enter the feature space, and a new 7-day forecast is generated using them as a leading indicator.
• Each morning — yesterday's complete actuals feed back into the bias tracker. Regime classifiers for extreme prices (ceiling events at ₹10,000/MWh, floor events below ₹500/MWh) are monitored continuously and recalibrated when the market shifts regime. The model never stops learning.

What the chart shows

Markets covered

• DAM — Day-Ahead Market. Hourly blocks, cleared at ~3 PM IST for the following delivery day.
• RTM — Real-Time Market. 15-minute blocks, cleared continuously throughout the day with gate closure 30 minutes before each block.

Methodology

Separately trained gradient-boosted (LightGBM) models for DAM and RTM, each producing a point (median) price per block. They learn from price history, CSEP grid-dispatch data, and weather across multiple grid cities. DAM uses 24 per-hour models; RTM uses a single global model.

Three post-model layers run on every forecast before it reaches the chart:

• Rolling bias correction — a per-block exponential-decay tracker computes systematic bias from the last 14 settled days and applies an additive shift to each 15-minute slot. Overnight slots and evening peaks carry their own independent corrections.
• Ceiling-regime overlay — a dedicated classifier estimates the probability that each block will clear at the ₹10,000/MWh CERC ceiling. When ceiling probability is elevated, the point forecast blends toward the ceiling proportionally. An hour-stratified correction handles overnight blocks, which have structurally weaker momentum signals even during heat-wave ceiling events.
• Floor-regime overlay — a separate classifier estimates the probability of sub-₹500/MWh floor events driven by monsoon RE surplus. The overlay activates during June–September when solar and wind generation can push prices to floor.

For partners and prospective customers

This public page shows a single-snapshot view. The full product includes forecast horizon control, hourly bid recommendations, backtested accuracy statistics by horizon and segment, API access for integration into your bidding pipeline, and regime alerts when ceiling or floor classifier probability exceeds configurable thresholds. For commercial enquiries please get in touch.