Energy
Mexico’s 100 critical hours
The rule of Mexico’s capacity balance market, applied to a synthetic hourly power system.
Mexico pays for capacity after the fact. When a year ends, the 100 hours with the lowest generation reserve become the critical hours, and everything is measured against them. Thermal plants are recognized at their average availability in those hours, and hydro, wind and solar at their average output. The reserve margin that results sets a factor between 2 and 0, and the price is roughly levelized fixed cost × factor − market income of the reference plant.
I built this to see how those rules behave across different years. The rule is real. The system and its weather are invented, at about the size of the SIN.
Loading the model…
Checking the rule
Before trusting the model, I tested the market rule against published data. A straight line through the curve points reproduces the 2023 SIN factor of 1.827, and the price formula reproduces the published SIN prices from 2016 to 2023 to within about 2%, using the published inputs. The small gaps come from details of the settlement that the short formula leaves out.
What I found
The price is very volatile even with the fleet held constant. Across 24 synthetic years the factor runs from 0 to 2, and the 10th to 90th percentile of the price goes from about USD 22,000 to USD 286,000 per MW-year. The published history moved even more: from nothing in 2021 to about USD 273,000 in 2023. In a very wet 2025, Wood Mackenzie reported a settlement at a factor of 1.37, down from 2.0 the year before. So anyone who sells capacity forward is holding a risk that depends on the weather of a year that has not happened.
Solar gets very little credit. About 91% of the critical hours fall after 6 p.m., which matches what happened in the real market, where GME counted 98% of them between 6 p.m. and midnight in 2023. Solar ends up recognized at about 4.5% of its installed capacity, against 22% for wind and 86% for thermal. At 20 GW of solar, its recognized share falls below 1%, because more solar pushes the critical hours further into the night.
The most interesting result is on batteries. The first 0.5 GW of three-hour batteries is recognized at 62% of its capacity and earns about USD 95,000 per MW-year. A fleet of 10 GW is recognized at 18% and earns about USD 4,000. Two things happen together: batteries flatten the evening peak and push the critical hours outside their discharge window, and the extra margin lowers the price. Revenue falls faster than either effect alone would suggest, which is the difference between valuing the first battery and valuing the fleet. There is a side effect too: with 4 GW of batteries, solar’s recognized share rises from 4.5% to 14%, because the critical hours move toward the late afternoon.
What it leaves out
The numbers show mechanisms and are not forecasts. The model has one zone, no transmission and no investment response to prices, and batteries dispatch with perfect foresight of the day. The reference plant values stay at their published 2023 levels. The rules are also being updated under the 2025 electricity law, so details may change.
Sources: CENACE’s preliminary exercise of the maximum capacity price (January 2025), GME’s technical note 01-2024, and Wood Mackenzie’s summary of the 2025 settlement.
Personal project built with synthetic data. It uses nothing from any employer, and it is not investment advice.