For millions across the Northeast last week, it may have felt that they were transported to the planet of Arrakis as the world turned an ochre hue when Canadian wildfire smoke made its way into the region. As shown in Figure 1 below, the commute was tough for many.

Figure 1 | New York Commuter Taking the 2/3 Express Downtown Last Week

Instead of newfound wealth from “spice” deposits, New York's power market told two intertwined stories this past week including a summer heat wave and a plume of Canadian wildfire smoke that arrived right on top of it. Day-ahead (DA) markets priced in a hot, tight system through midweek. What showed up in real time (RT) was more complicated. Smoke didn't just cloud the skyline, it dented behind-the-meter (BTM) solar output and appears to have capped how hot it actually got, even as the region remained under a heat advisory. The net result was a DA/RT spread that spiked mid-week and persisted at a lower amount in the following days.

The Smoke Timeline

Wildfire smoke from a cluster of large, actively burning fires in northwestern Ontario began moving into the New York City region on Wednesday, July 15, arriving on the back half of a multi-day heat wave that peaked that same day. Air quality alerts covered New York, New Jersey, and Connecticut, with readings on Wednesday ranging from "unhealthy" to "unhealthy for sensitive groups."

Thursday, July 16 was the worst day of the week for smoke. The Air Quality Index (AQI) reached roughly 200 — described by local outlets as the city's worst smoke event since June 2023 — even as the heat advisory technically expired Wednesday night. Friday brought a brief reprieve as skies cleared. Smoke returned Saturday, July 18, as did thunderstorms and by Sunday, July 19, skies were largely clear again.

Suppressing Behind the Meter Solar

That timeline shows up in NYISO's system-wide BTM solar output. Midday (HE10–HE17) average output for the week shows in Table 1:

Table 1 | NYISO BTM Solar Daily Average During Smoke Event

Wednesday's smoke arrival cut midday BTM solar output roughly in half versus the Monday/Tuesday baseline. Thursday — the smokiest day of the week by air quality readings — still ran well below baseline even though it wasn't the single worst day for solar output. That distinction belongs to Saturday, when returning smoke combined with severe thunderstorms to cut BTM output to barely a quarter of normal.

Smoke Suppressed Temperatures

Forecasters had flagged this possibility that heavy smoke aloft scatters and absorbs incoming sunlight before it reaches the surface, which can dim sunlight and cap daytime heating even during an active heat wave. That's consistent with what happened in New York City-area RT temperatures versus the DA forecast:

Table2 | NYC Forecast versus Actual Temperatures

Thursday, the day with the worst AQI readings of the week, also produced the single largest actual-versus-forecast average temperature miss, at 2.2 degrees below what was forecast the day before. Friday, once the smoke cleared, was the one day actual temperature came in above forecast. It's a reasonable interpretation of the data that the smoke plume itself, not just the passing of the heat wave's peak, helped keep Thursday's temperatures from reaching what had been forecast.

However, the story has another side. Smoke pulled system load in two opposite directions at once. Lower temperatures mean lower cooling demand thus pulling down load.  On the supply side, suppressed BTM solar means less generation that offsets customer demand. This pushes grid-served load up, since customers draw more from the bulk system when their own solar underproduces.

Table 3 looks specifically at midday hours (HE10–HE17), where the BTM solar effect is concentrated.

Table 3 | NYISO Forecast versus Actual Midday Load (HE10-HE17)

Wednesday, when BTM solar output was cut roughly in half, is the one day midday load came in above forecast, even at the peak of the heat wave. That's consistent with the solar-suppression effect where less rooftop generation meant more demand had to be met by the bulk grid, and that effect was strong enough on its own to outweigh any early cooling from the smoke that had just arrived.

By Thursday, the picture flips. BTM solar was still running well below baseline (which, in isolation, should have pushed grid load higher), yet midday load missed forecast by 8.3%, the sharpest miss of the week, on either a midday or full-day basis. That only makes sense if the temperature-suppression effect of the day's heavy smoke was large enough to swamp the load-boosting effect of lost solar. In other words: smoke's cooling effect on demand, not its dampening effect on rooftop solar, is what drove Thursday's load well below forecast, even though both effects were in play at the same time.

For the rest of the week (Friday through Sunday), BTM solar was back near normal, and load undershot forecast in line with cooler-than-expected, storm-affected conditions and typical weekend demand patterns.

The DA/RT Price Spread

That mismatch between forecast and reality shows up directly in the gap between DA and RT locational prices. Looking at NYISO's New York City zone (Zone J), on-peak hours (HE8–HE23):

Table 4 | NYISO Zone J (NYC) DA and RT On-Peak LMP and Spread

Every single day this week settled with DA prices above RT as the DA market was consistently pricing in more heat-driven demand than materialized. The spread peaked on Wednesday, the hottest day of the week, at $165/MWh on-peak, as DA prices near $278/MWh gave way to a RT settlement of roughly $113/MWh. That's the clearest sign of the week: the DA market baked in a hot forecast for Wednesday, and while it was hot, RT supply/demand balance didn't require anywhere near the premium the DA market assigned to it.

From there, the spread compressed rapidly. By Tuesday's counterpart hours later in the week (Friday, Saturday, Sunday), the DA/RT gap had fallen back into single-to-low-double digits, still positive, still showing DA running rich to RT, but nowhere near the mid-week blowout.

Putting it Together

The pattern across smoke, solar, load, and price data lines up into a single narrative:

●   Wednesday (7/15) was the actual forecast peak of the week on temperature, load, and the DA/RT spread alike, the hottest day, the highest load, and by far the widest gap ($165/MWh on-peak) as the day-ahead market priced aggressively for a hot, tight system. It's also the day smoke arrived and cut BTM solar output roughly in half, which briefly pushed grid load above forecast even as the smoke began its cooling effect.

●   Thursday (7/16) was the smokiest day of the week by air quality, and the day the smoke's cooling effect clearly took over: a temperature miss (-2.2°F on the daily average) and the sharpest load miss (-8.3% at midday) of the week, even with BTM solar still suppressed in a way that should have pushed load higher. The DA/RT spread nearly halved from Wednesday's peak.

●   Friday through Sunday saw BTM solar mostly back to normal (aside from Saturday's smoke-and-thunderstorm-driven crash), and both temperature and load consistently ran below the day-ahead forecast. The DA/RT spread continued to compress, though it stayed positive through Sunday. The persistence of a positive spread even on the mildest, smoke-free days (Saturday, Sunday) suggests day-ahead pricing retained a bit of a risk premium through the weekend even as the underlying weather and load fundamentals eased.

For traders and risk managers, the week is a reminder that DA NYISO pricing, especially around a summer heat pattern, leans heavily on forecast temperature and load assumptions, and that an unpredictable factor like wildfire smoke can move both of those inputs (and in the case of BTM solar, in opposite directions at once) well before RT settlement.