Grid Alpha

The real-time LMP spread across CAISO hit $1,288.91/MWh at 10:15 AM PT on July 11, the widest intraday gap in the session. Node SPRPPGNODE2 cleared at $89.23/MWh while BSPHYD341N001 posted negative $1,199.68/MWh, a $1,289 delta driven entirely by congestion.

The energy component was uniform at $27.56/MWh across all top nodes. The spread is pure congestion: SPRPPGNODE2 carried $50.76/MWh positive congestion, while BSPHYD341N001 and four CONTROLX nodes each showed negative congestion of $1,238.15/MWh. That is a forced flow reversal — the ISO is paying generators nearly $1,200/MWh to push power into a constrained pocket, or paying loads that much to take it off the system. The identity of the binding constraint is not yet confirmed, but the CONTROLX substation cluster and BSPHYD34 hydro unit are the obvious candidates. CAISO logged 1,668 curtailment events over the prior seven days, though the sample data shows near-zero MWh volumes, suggesting the count may reflect many small or zero-MW events rather than a systemic oversupply condition.

If this spread persists into the next binding constraint update, traders should watch for a derate or outage notice on the BSPHYD34 unit or the CONTROLX bus. A sustained negative $1,200/MWh LMP at a hydro node implies the unit is being dispatched down or off, with the ISO paying to keep it from injecting. Watch the 15-minute real-time price re-runs: if the spread collapses to $200-300/MWh in the next interval, this was a transient SCADA glitch or a single-interval constraint relaxation. If it holds or widens, the afternoon peak will see cascading congestion into adjacent nodes, and the day-ahead versus real-time basis for that path will blow out.

> A $1,289/MWh spread with uniform energy costs is a transmission constraint, not a supply event; the next constraint update will tell you if this is a five-minute spike or a five-hour problem.

Not investment advice. For informational purposes only.

PJM Western Hub peak price hit $563/MWh on 2026-07-09, nearly 9x the 24-hour mean. That is not a summer afternoon spike from a thunderstorm. That is a market signaling that dispatchable capacity is thin and the marginal unit is burning expensive gas at a heat rate that implies delivered fuel costs well above the forward curve.

The mechanism is straightforward: forced outages are eating PJM's cushion. On 2026-07-04 alone, the anomaly log shows five forced outage events totaling over 56,000 MW at the reading level. The largest single reading was 18,267 MW. Maintenance outages added another 3,615 MW across five readings. Those are not planned refueling outages. Those are units tripping or being derated in real time, and PJM's operating reserve margin is not deep enough to absorb them without dispatching the stack's tail. When the last unit in merit clears at $563, the entire hub settles at that price for the hour. The EIA's forecast of 8% lower wholesale prices this summer is an average that masks the tail risk. NEADA's Mark Wolfe is correct: consumers see no savings when the peaks are this sharp.

Watch today for two things. First, the 2026-07-10 forced outage total versus yesterday's. If forced outages remain above 15,000 MW at any single reading, the afternoon peak will test $400 again. Second, watch Dominion South gas. If it ticks above $2.50/MMBtu, the marginal heat rate for the last coal unit in PJM West pushes toward $400 before any scarcity adder. The spread between PJM Western Hub peak and the 5-year average for July 10 is already at a 3-sigma level. If load forecasts for today come in at or above yesterday's actual peak, that spread widens further.

> The $563 print was not a data error. It was the market telling you the reserve margin is a fiction on hot afternoons with units down.

Not investment advice. For informational purposes only.

The maximum real-time LMP spread in CAISO hit $787.27/MWh on July 9, 2026, at 10:15 UTC. The gap between Dunnside at $48.53/MWh and Bishop at -$738.75/MWh is not a pricing anomaly. It is a transmission constraint signal with a congestion component of -$765.33/MWh at Bishop, meaning the node is paying nearly $766/MWh to stay online.

The mechanism is straightforward. Dunnside's congestion component was a modest $15.09/MWh. Bishop's was -$765.33/MWh, with energy at $19.34/MWh and losses at -$5.11/MWh. The negative LMP means generation at Bishop must pay to export. This is classic renewable oversupply in a constrained export path, compounded by battery storage behavior. CAISO logged 1,668 curtailment events in the prior seven days. The CPUC modified battery storage rules on July 8 to address "foldback" — the phenomenon where storage resources withdraw from charging during negative prices, removing the only local demand sink and driving prices even lower. The timing is tight: the rule change was one day before the $787 spread.

Traders should watch two things. First, the Dunnside-Bishop spread widens if solar irradiance is high and battery state-of-charge is full at the same transmission constraint. Second, the CPUC foldback rule may reduce the severity of future negative price events, but the first test will be the next high-solar, low-load weekend. If Bishop repeats -$738/MWh under the new rules, the problem is physical, not behavioral.

> A $787/MWh intraday spread is not a pricing error; it is a transmission constraint priced correctly, and the CPUC's foldback fix will be tested the next time solar oversupply meets full batteries.

Not investment advice. For informational purposes only.

CSADIAB7N001 settled at -$150.77/MWh in CAISO's real-time market at 10:15 UTC on July 8, 2026. The congestion component alone was -$173.57/MWh, more than offsetting a positive energy component of +$27.44/MWh and a -$4.89/MWh loss component.

Four adjacent nodes — CONTROLX1N003, CONTROLX1N004, LEEVINE1N001, and LUNDY7N003 — posted nearly identical prints at -$150.74/MWh, all sharing the same -$173.57/MWh congestion signal. This is not a single-node data glitch. It is a localized pocket where the market is paying generators to reduce output, or equivalently, paying load to consume more. The energy component remains positive across CAISO, meaning the negative LMP is purely a congestion phenomenon: a binding export constraint from this sub-area into the wider grid.

CAISO recorded at least 1,668 curtailment events over the prior seven days, per gridalpha anomaly data. That figure signals chronic oversupply conditions, likely driven by solar ramping into low load. If the binding constraint at CSADIAB is a transmission path that frequently binds during midday solar peaks, traders should watch whether the negative LMP persists across consecutive 5-minute intervals or remains a single-interval spike. Persistent negative prints would widen the real-time versus day-ahead basis at CSADIAB, creating a repeatable congestion capture trade: buy day-ahead at the positive hub price, sell real-time into the negative node, and collect the congestion rent if the constraint holds.

> The -$173.57/MWh congestion component at CSADIAB is the signal, not the -$150.77 LMP; watch for persistence across intervals to size the basis trade.

Not investment advice. For informational purposes only.

Real-time LMP at PODBISHOP1_UNITS-APND printed -$150.00/MWh at 11:00 UTC on July 7. The congestion component alone was -$177.08/MWh, more than offsetting a positive energy component of +$22.16/MWh. Four nodes in the Bishop/Control area all landed within a penny of -$150/MWh at the same interval.

The negative price is a transmission story, not a generation story. Energy value at Bishop was positive and in line with system-wide CAISO pricing. The -$177.08/MWh congestion component tells you that the node is trapped behind a binding export constraint, unable to push power to load centers. This is classic localized oversupply: solar and/or wind generation in the eastern Sierra is exceeding the physical capacity of the transmission paths out of the area. CAISO logged 1,668 curtailment events in the last seven days, though the sampled MW volumes appear near zero, suggesting many events are small or administrative.

For traders, the basis spread between Bishop and NP15 or SP15 is the relevant number. If the congestion component persists across multiple intervals, the spread widens structurally. That creates an arbitrage for dispatchable resources: charge storage at -$150/MWh, discharge into a positive-price hub. The Caballero 100-MW/400-MWh BESS, active since early 2025, is the type of asset that should be soaking up these intervals. Watch its state-of-charge data and day-ahead bids for evidence that the strategy is actually being executed. If Caballero and similar batteries are not cycling during these events, the market is leaving money on the table.

> Negative LMPs driven by congestion, not energy, mean the basis spread is the trade, not the absolute price.

Not investment advice. For informational purposes only.

The real-time LMP spread between COLWATER1N011 and PODBISHOP1_UNITS-APND hit $675.50/MWh at 10:15 July 6. That is not a normal midday ramp. It is a transmission constraint tearing the market in half.

COLWATER1N011 settled at $487.63/MWh with a congestion component of $458.78. On the other side of the Bishop constraint, PODBISHOP1_UNITS-APND went negative at -$187.87/MWh with a congestion component of -$207.13. The energy component at both nodes was nearly identical ($15.46 vs $14.29). This is pure transmission physics. The Bishop area is exporting into a constrained path, and the market is pricing the bottleneck at nearly $460/MWh on the high side while dumping excess generation at negative prices on the low side.

CAISO logged 1,668 curtailment events in the seven days ending July 6. Solar and wind curtailment appeared on June 29, though reported MW volumes in the sample were small. The pattern matches a familiar mechanism: high renewable output in the eastern Sierra region, insufficient transfer capability out of the Bishop area, and the ISO cutting generation instead of redispatching around a binding constraint. The $675 spread is the extreme tail of that distribution.

Watch the forward congestion on the Bishop path over the next week. If the constraint persists through the July 10-12 weekend when solar output peaks, expect repeated negative pricing at PODBISHOP and its sibling nodes. The spread widens if any transmission element in the Colwater-Bishop corridor trips or enters maintenance. Conversely, if CAISO issues a market notice lifting the constraint, the $460 congestion premium on the high side collapses fast. Traders holding virtual supply at COLWATER1 should monitor the ISO's outage calendar for the Bishop 115 kV lines.

> A $675 intraday spread with nearly identical energy components is a transmission signal, not a supply signal: the Bishop constraint is the only number that matters.

Not investment advice. For informational purposes only.

Yesterday's $2459/MWh print at PJM Western Hub marks the highest real-time price event since Winter Storm Elliott. PJM declared a capacity emergency and called all resources into service as load exceeded 162 GW, driven by a heat wave pushing temperatures past 100 degrees across the RTO footprint.

The 162 GW load figure already threatened PJM's 2006 summer hourly integrated record of 165,563 MW. That record did not fall yesterday, but the margin was thin enough that PJM secured FERC approval to curtail data centers and other large loads "as a last resort." No curtailment order has been issued yet, but the authority is live. Forced outages remain elevated: a single 04:00 snapshot on June 27 showed 10,948 MW of forced outages, with 144 anomaly events recorded across the last seven days. That is a material supply-side drag during a peak demand event.

The Champlain Hudson Power Express reactivated on July 2 after repairs to an equipment failure on the Canadian side. CHPE's return adds roughly 1,250 MW of hydro into New York City, which relieves one pressure point on the PJM-NYISO interface but does nothing for Western Hub congestion directly. Western Hub basis against AEP-Dayton and Dominion Hub will be the spread to watch today: if load holds above 160 GW, the AEP-Western Hub spread could widen further as western PJM units face both heat-rate degradation and transmission constraints into the eastern load pocket.

> The $2459 print is not a one-off spike; it is the market pricing in a structural supply deficit that forced outages and record heat have exposed, and today's session will test whether yesterday's emergency was a warning shot or the new floor.

Not investment advice. For informational purposes only.

West Hub printed a $287/MWh swing on 2026-04-26, a move big enough to reprice the rest of the week if it holds. The read-through is not subtle: basis widened hard enough to signal a local constraint, not a clean system-wide move.

There is not enough verified detail in the supplied facts to pin the exact day-ahead and real-time LMPs, the congestion component, or the interval that drove the move. [FACT NEEDED] on the West Hub price print, [FACT NEEDED] on the basis pair versus Zone A or adjacent hubs, and [FACT NEEDED] on whether the swing came from one interval or a multi-day run. That matters because a single spike can fade; a durable basis break usually needs a transmission limit, outage, or load-pocket imbalance that lasts past one dispatch cycle.

For now, the mechanics point to western New York supply tightness relative to load. If NYISO keeps posting west-side congestion notices, then West Hub should stay bid versus the rest of the state. If the move was tied to a short-lived outage or a temporary interface limit, then the spread can snap back quickly once the constraint clears. Watch West Hub versus Zone A, and watch whether the spread persists into the next operating day rather than just the next interval.

> When West Hub gaps this far, the market is usually telling you the wire, not the load, is setting the price.

Not investment advice. For informational purposes only.

HB_HOUSTON touched $1153/MWh last night, roughly 12x the 24-hour mean embedded in the thesis. That is not a rounding error in a hub that normally trades like a liquid reference point; it is a scarcity print, and it says the evening stack got thin enough to clear at punitive levels.

The backdrop is not subtle. ERCOT’s unplanned resource outage feed showed at least 19,306 events over the last 7 days, though that result is sample-limited, and the sample includes forced reductions across gas, wind, and water units: DEC was down 22 MW, SANTACRU 25 MW, BUCHAN 17 MW, AQUILLA 49 MW, and FOARDCTY another 10 MW on a forced-extension basis. That is not one big failure, just a steady accumulation of missing megawatts. ERCOT’s SCED shadow-price feed also printed at least 7,330 events over the same window, with constraint samples at 6.827 on WESTEX, 46.446 on 630_B, and 62.388 on BOWFMR1. In parallel, the LMP/DART feed logged at least 433,228 events, a reminder that price discovery is active and the system is already moving through a lot of congestion and redispatch.

The market question is whether last night was a one-off evening squeeze or the first signal of a repeatable pattern. If the same outage stack remains in place and the same generation stack is called again into the evening ramp, HB_HOUSTON can reprice quickly before the broader ERCOT screen catches up. Watch Houston versus the wider ERCOT hubs and the behavior of binding constraints into the load pocket: if congestion tightens while operating reserves thin, the spread can widen faster than outright hub averages suggest.

The clean read is that scarcity is no longer theoretical when Houston can print four digits on a routine evening.

> In ERCOT, the price is telling you the evening stack, not the average day, still owns the tape.

Not investment advice. For informational purposes only.

NYISO’s N.Y.C. LBMP hit $609/MWh in the evening, or 12.6x the 24-hour mean. That kind of move is not a noise print; it says the zone cleared against a thin margin and the market had to pay up for deliverability into Zone J.

The mechanics are straightforward even if the exact trigger chain is not yet public. In a constrained load pocket like NYC, a spike of that size usually means local demand, imports, and unit availability lined up badly for one interval, then eased as conditions normalized. The fact pattern points to a tight supply-demand balance rather than a broad system repricing. If the hour also came with binding transmission into Zone J, shadow prices would have done part of the work; if not, then the scarcity sits more squarely in local generation and load.

What I would watch today is whether the same setup repeats in the evening ramp. If load follows a similar profile and import capability stays limited, then the right comparison is not the day’s average but the marginal hour against nearby zones and the day-ahead strip. If real-time keeps clearing far above day-ahead in Zone J, that is the cleaner tell that the pocket remains tight and that traders should keep an eye on the spread into the rest of the state.

> A $609/MWh print in NYC is the market telling you the last megawatt in Zone J got expensive fast.

Not investment advice. For informational purposes only.