Share this
by Sarah Hatch
Daylight Savings Time (DST) and ISO Data
Twice a year, those that collect and manage data, face the dreaded extra or missing hour – in the spring when we "spring forward," and in the fall when we “fall backwards.” On the Short Day in spring most ISOs publish only 23 rows of data. On the Long Day in the fall most ISOs publish 25 rows of hourly data. Each ISO handles this differently – and this is true not only across ISOs but within ISOs. We've learned there are special nuances for each ISO website.
We Stay Awake So You Don't Have to
Considering we're collecting data from hundreds of websites, these two dates create a lot of work. For Yes Energy this work happens in our Market Data Ops Team. This team of superheroes monitors the data when it comes in – starting with forecasts earlier in the week, day-ahead on Saturday, real-time results on Sunday, and final/revision files after the DST date.
We take this so seriously that our team stays up and watches the data load in the database and updates in QuickSignals to confirm that we capture all of the data. We run secondary proofs on the raw data in the days and weeks following DST as well.
How Does DST Change the Power Markets?
Since DST changes our behavior so much we wondered how does it change the behavior of market participants in the virtual market. We took a look at our data to answer “Does Daylight Savings Time impact the participants in the virtual markets?” Here’s a look at what we found.
This chart shows the virtual supply MWs by ISO. Daylight Savings Time was on 11/3. On this day, most ISOs had lower supply MWs than other days in October and thus far in November.
Virtual supply MWs for ISO through October and so far in November.
Here’s a look at the percentiles for DST:
So, yes, it looks like in a few ISOs there is a possible impact of Daylight Savings Time on market participation. But we should keep in mind that many factors could be at play, including day of the week and market fundamentals such as supply, load, and outages.
We looked at how much load contributed to the MWs of virtual supply, and it appeared that load alone doesn't account for the virtual supply in any of the ISOs:
Want to dig into more analysis like this around Daylight Savings Time? Or are you looking to solve the hassle of collecting and managing data around this tricky day?
