Key Insight
Snowtrail's PEMI index signalled extreme grid stress across 11 European zones three days before peak power price dislocation, giving traders an actionable window ahead of a multi-week cascade.
PEMI decomposes grid stress into five event types, each with distinct trigger logic. During the March 2026 crisis, all five fired simultaneously for the first time in the dataset, the signature of a systemic market event, not an isolated outage.
The crisis did not start in one zone and slowly spread. Events appeared across multiple European zones from the very first day. What distinguishes this episode is the intensification: early February saw scattered, low-severity alerts across the continent, but by late February and into March the event density concentrated in Nordics (Sweden, Norway, Finland) and Benelux (Belgium, Netherlands), with severity escalating from isolated outages to persistent, multi-channel stress. The heatmap below shows this shift from broad-but-mild to concentrated-and-severe.
Grid stress is only actionable if prices respond. PEMI's price impact layer confirms whether outage-driven supply loss is reaching the day-ahead market.
Timing note: The initial price shock (early February) reflects the acute onset of nuclear fleet failures in the Nordics. The sustained elevation through late February and into March reflects chronic stress as outages persisted and spread geographically. Both phases are part of the same episode. The signal value is that PEMI flagged the transition from acute shock to chronic stress before prices fully adjusted.
Following the PEMI signal escalation, European power markets experienced significant stress across multiple dimensions:
Price dislocation: Day-ahead power prices spiked to extreme statistical levels, with Spain recording a z-score of 4.37 (i.e., more than four standard deviations above its 30-day average). France reached z=3.14, Denmark z=3.00, and Belgium z=2.46. These are not normal fluctuations. They are the type of moves that trigger risk limits.
Cross-commodity transmission: As generation capacity went offline, gas-for-power substitution surged across 8 countries. Norway absorbed the heaviest burden at 64.9 GW, followed by Sweden at 55.1 GW. This is the bridge between PEMI (power grid stress) and GBSI-EU (gas market stress). The substitution demand directly tightened European gas balances.
Volatility persistence: Unlike a single-day price spike, this event produced elevated volatility for weeks. PEMI's persistence analysis shows that CRITICAL-tier stress episodes had a 43% survival rate at 3 days, consistent with chronic fleet failures rather than transient outages. Traders who treated this as a one-day event were wrong-footed by the sustained elevation.
When generation capacity goes offline, gas turbines fill the gap. PEMI's gas substitution layer quantifies this cross-commodity linkage: the bridge between power grid stress and gas market impact. During this crisis, the Netherlands and Belgium saw the largest total outages, while Nordic countries absorbed the heaviest gas-for-power substitution burden.
PEMI's primary signal layer combines all five event channels into a 5-tier risk classification (NONE / LOW / MODERATE / HIGH / CRITICAL). This chart shows how the signal evolved across the most affected zones during the crisis.
PEMI's grid stress signal first flagged elevated risk in early February 2026, when Nordic bidding zones began reporting concurrent generation outages. By mid-February, 11 zones had triggered at least one elevated-risk day, well before power markets fully priced the supply shortfall.
The crisis produced 27 elevated-risk zone-days across 11 European zones. All five event channels fired simultaneously for the first and only time in the dataset. The grid stress signal reached its highest-ever severity concentration, with critical infrastructure alerts, surprise clusters, and persistent stress reinforcing each other across the Nordic-to-Benelux corridor.
A single-channel alert (e.g., one nuclear unit offline) is routine. When three or more channels fire in the same zone-day, PEMI is flagging systemic stress. When all five fire across multiple zones, the signal is unambiguous: this is not a maintenance event, it is a market regime shift. The contagion heatmap above quantifies how many days of lead time you have before the stress reaches your exposure zones.
PEMI's backtest asks a specific question: when the grid stress signal fires at a given tier, does supply pressure actually escalate over the following 1, 3, and 7 days compared to unconditional baselines?
Important caveat on sample sizes: PEMI covers a shorter history than the gas products (GBSI, GLMI). The CRITICAL regime has been observed, but with limited samples. Escalation multipliers are reported transparently and flag where confidence intervals are wide.
| Regime | N | Sufficient | 1d Mult | 3d Mult | 7d Mult | Avg Duration |
|---|---|---|---|---|---|---|
| NO_RISK | 1933 | Yes | 0.7x [0.9-1.6] | 0.7x [1.0-2.0] | 0.8x [1.0-2.0] | 33.3d |
| LOW | 48 | Yes | 3.4x [4.9-6.6] | 3.7x [5.2-10.9] | 3.3x [5.2-7.1] | 1.1d |
| MODERATE | 7 | NO | 7.6x | 6.4x | 6.0x | 1.0d |
| HIGH | 1 | NO | 14.6x | 15.1x | 16.0x | 1.0d |
| CRITICAL | 17 | NO | 23.1x | 28.4x | 26.7x | 2.4d |
The five event channels are not redundant. Each reveals a different dimension of the crisis. This table breaks down exactly what each channel contributed during the March 2026 episode.
| Channel | Events | Zones | Active Days | First | Last | Avg Magnitude | Max Magnitude |
|---|---|---|---|---|---|---|---|
| Critical Infra | 70 | 15 | 24 | Feb 02 | Mar 15 | 82.5 | 100.0 |
| Mkt Tightness | 144 | 20 | 32 | Feb 02 | Mar 15 | 2390.1 | 9760.0 |
| Surprise Cluster | 122 | 21 | 36 | Feb 02 | Mar 15 | 97.8 | 100.0 |
| Persistent Stress | 152 | 24 | 34 | Feb 02 | Mar 13 | 663.0 | 3783.8 |
| Weather Amplified | 30 | 11 | 11 | Mar 04 | Mar 15 | 76.9 | 100.0 |
PEMI's event escalation signal breaks down why the market tightened: was it supply pressure, event clustering, chronic outages, or price spikes? During this crisis, the triggers shifted over time as the episode evolved from acute shock to chronic stress.
A rising PEMI composite score historically precedes actionable moves across power and gas markets:
The channel count is the severity gauge: single-channel alerts are noise, 3+ channels in one zone is actionable, all five channels firing simultaneously has occurred exactly once in the dataset.
The March 2026 Pan-European Grid Crisis was the only period where all five PEMI event channels fired simultaneously, escalating from scattered alerts across 20+ European zones into concentrated, multi-channel stress in Nordics and Benelux, with price z-scores exceeding 4 and gas-for-power substitution spiking across 8 countries.
Power Trader: - Watch the channel count: scattered single-channel alerts are noise. When 3+ channels fire in the same zone, that zone is transitioning from isolated outage to systemic stress. All five firing simultaneously has only happened once - The concentration shift is your positioning signal. When event density migrates from dispersed to focused on specific zones, those zones are where forward curves need repricing - Persistent stress lasting >7 days means this is not a one-week maintenance event. Re-price your forward curve accordingly
Gas Trader: - Gas substitution features tell you how much incremental gas demand each GW of generation offline creates. Cross-reference with GBSI-EU to understand whether the gas market can absorb it or whether TTF spikes - Weather amplification + critical infra = worst case for gas demand. Size your exposure to gas-for-power before the cold snap arrives, not after
Risk Manager: - The channel count is your severity indicator. One channel firing = localized event. Three+ channels = systemic risk. All five = the only systemic market event in the dataset - Persistence data tells you whether to treat this as a volatility spike (LOW: resolves in days) or a regime shift (CRITICAL: 43% still active after 3 days, p90 of 4 days) - Supply pressure reaching 100% in Belgium means the zone has exhausted its reserve margin. That is when interconnector dependency becomes critical
PEMI does not just say "the grid is stressed." It tells you how (which channels), where (which zones), why (which triggers), how long (persistence and survival), and whether it is getting worse (escalation trajectory). During March 2026, that decomposition was the difference between seeing a single headline and understanding a multi-week, multi-country cascade with quantifiable gas market consequences.
Request a demo to explore PEMI signals, events, and backtests with your own use cases.
Request Demo