MISO Market Mechanics¶
Detailed reference on MISO data types, pricing conventions, profit calculations, clearing rules, and lessons from arbitration.
Data Types¶
Day-Ahead ExAnte LMP¶
The DA exAnte LMP is the clearing price used to determine whether offers and bids are accepted. It is identical to the clearing price displayed on the MISO trading portal.
Day-Ahead ExPost LMP¶
The DA exPost LMP is the settlement price used for profit calculations. This is distinct from the exAnte price and reflects post-market adjustments.
Real-Time LMP¶
RT LMP files are published in two stages:
- Preliminary files: Published first; available for recent days.
- Final files: Published later; the authoritative settlement price.
RT Data Availability Timeline
On OD-1 (when placing trades for OD), RT final files are available only up to OD-7. For OD-6 through OD-2, only RT preliminary files are accessible. This means strategies using recent RT data must work with preliminary values for the most recent week.
MISO Does Not Retain Preliminary Files
MISO removes preliminary files once the final files are published. To mitigate this, preliminary files have been stored internally since 2024-08-08. Some files may contain dollar-formatted values with commas that require parsing.
Profit Calculations¶
The DA exPost LMP is used to calculate gains:
- Offers (virtual supply):
DA exPost LMP - RT Final LMP - Bids (virtual demand):
RT Final LMP - DA exPost LMP
Dead Node Adjustment
When a node-hour is reported as dead on OD, the profits for those node-hours must be voided. There is currently no known equivalent to dead nodes in SPP or ERCOT.
Dead Node Files¶
Dead node files report the list of hours that were dead for a given date.
Key details:
- The "Mkt Hour" column shows the beginning hour:
00:00:00indicates the first hour;23:00:00indicates the last hour. - Dead node information is published the next day. Dead nodes for 2024-12-12 are reported on 2024-12-13.
- The filename includes the publication date, not the date of the dead node hours (e.g.,
Dead_Node_Report_20241213.xls). - Dead node files for OD-2 are published at 10:20 AM EPT.
Tight Timing Window
If strategies depend on OD-2 dead node data, the submission window is less than 10 minutes (market closes at 10:30 AM EPT).
Market and Trading Details¶
Market Timings¶
The MISO Day-Ahead market operates from 6:00 AM to 10:30 AM Eastern Prevailing Time (EPT). These timings remain unchanged regardless of daylight saving.
Submitting Bids and Offers¶
Overwrite Behavior
Unless explicitly structured with multiple blocks, submitting multiple offers or bids for the same node-hour will overwrite previous submissions. To avoid this, combine all offers and bids into a single file before submission.
Clearing Process¶
Conditions for Clearing¶
A node-hour clears if all of the following are true:
- The node-hour does not appear in the dead node file.
- For offers: the offer price is <= DA exAnte LMP.
- For bids: the bid price is >= DA exAnte LMP.
- The net volume on the node-hour is not zero.
Net Volume Example
If a node-hour has a supply offer of 1 MWh and a demand bid of 1 MWh, the net volume is 0. Neither the offer nor the bid will clear for that node-hour.
Impact of Net Volume on Calculations¶
- Gains: Net volume discrepancies do not affect gains because gains are based solely on LMPs. In the example above, the supply and demand gains for the 1 MWh would offset each other.
- Expected volume: The expected volume will differ if net volume conditions are ignored during simulation.
Arbitration¶
The following section documents lessons learned during the arbitration process.
Rounding Errors¶
- Rounding errors can compound, causing discrepancies in P&L calculations.
- Not rounding at all may also produce discrepancies.
- Rounding to 4 decimal places was adopted for the median-based strategy.
Tie Breaking¶
When ranking values, ties may arise. A clear tie-breaking rule must be defined in advance. For the median-based strategy on MISO, alphabetical order is used as the tie-breaker.
General Arbitration Approach¶
- Compare daily P&L between the simulated and actual results.
- Compare selected nodes (or node-hours, depending on the strategy) on days where node selection diverges.