Intraday Markets & Continuous Trading
The day-ahead auction establishes schedules and prices — but between DA results (~13:00 D-1) and real-time delivery, the world changes. Wind forecasts update, units trip, demand deviates. The intraday market exists to let participants adjust their positions continuously, reducing imbalance exposure before gate closure forces them into the TSO's balancing mechanism.
Position in the cascade: Intraday sits between day-ahead (welfare-maximizing auction, ~13:00 D-1) and balancing (TSO-administered, real-time). It is the last market-based opportunity to trade — the participant's final chance to avoid imbalance prices.
Two Trading Mechanisms
European intraday markets use two distinct mechanisms, increasingly in parallel:
Continuous Trading (SIDC / XBID)
- Order book matching: first-come, first-served
- Pay-as-bid (not uniform price)
- Cross-border via shared order book (implicit capacity allocation)
- Opens at 15:00 D-1; runs until gate closure
- Quarter-hourly, half-hourly, hourly, and block products
- No single clearing price — every trade at its own price
- Liquidity concentrated near delivery
Intraday Auctions (IDAs)
- Discrete auctions at fixed times (e.g., 15:00, 22:00 D-1; 10:00 D)
- Uniform clearing price (welfare maximizing, like DA)
- Cross-border capacity allocated implicitly in each auction
- Better price discovery (aggregates information)
- Quarter-hourly and hourly products
- Three pan-European IDAs under SIDC 2.0 (IDA1, IDA2, IDA3)
- Complementary to continuous — not a substitute
Why Both? The Economic Logic
Each mechanism serves a different need:
| Property | Continuous | Auction |
|---|
| Price discovery | Weak — fragmented, bilateral prices | Strong — single price aggregates all information |
| Immediacy | Instant execution when matched | Must wait until auction time |
| Cross-border efficiency | Sequential capacity allocation (first-come) | Simultaneous optimization (like DA) |
| Liquidity aggregation | Spread over hours; thin early, thick late | Concentrated at auction moment |
| Strategic behaviour | Sniping, latency advantages possible | Sealed-bid; no timing games |
SIDC 2.0 design choice: The complementary model (auctions + continuous running between them) attempts to combine the price discovery of auctions with the flexibility of continuous trading. IDAs are scheduled at key decision points — after DA results, before evening ramp, and on delivery day — to "reset" the price signal and re-optimize cross-border flows.
Gate Closure: The Critical Boundary
Gate closure (GCT) is the moment after which you can no longer trade your position — any remaining imbalance becomes the TSO's problem, settled at the imbalance price. The GCT varies by country and is a fundamental design parameter:
| Market | Local GCT | Cross-border GCT (SIDC) | Product granularity |
|---|
| Germany (EPEX) | D, H-5 min (local) | D, H-60 min (cross-border) | 15-min, 30-min, 60-min |
| France (EPEX) | D, H-5 min | D, H-60 min | 30-min, 60-min |
| Nordics (Nord Pool) | D, H-60 min | D, H-60 min | 60-min (15-min being introduced) |
| Spain (OMIE) | D, H-25 min | D, H-60 min | 60-min |
| UK (N2EX, Nordpool) | D, H-60 min | N/A (not in SIDC) | 30-min settlement periods |
The GCT asymmetry matters for optimization: Cross-border SIDC gate closure (H-60) is earlier than local gate closure (H-5 in DE). This means in the final hour, you can only trade locally — cross-border arbitrage opportunities are locked. An optimizer must account for this: cross-border positions should be closed out before H-60, while local fine-tuning continues until H-5.
Price Formation: Pay-as-Bid vs. the DA Benchmark
Continuous intraday is pay-as-bid — there is no single clearing price. This creates several consequences:
- No transparent reference price: The "intraday price" is typically reported as volume-weighted average (VWAP), ID1 (first trade), ID3 (last 3 hours), or IDFull (full session) — each tells a different story
- Information rent: A participant with better forecasts captures the spread between their valuation and the counterparty's. Unlike DA where surplus is split uniformly, intraday rewards private information directly
- Price dispersion: The same product (e.g., DE hour 14) may trade at 50 €/MWh early in the session and 120 €/MWh near gate closure after a wind forecast drops. Both are valid transaction prices
- Benchmark risk: If your imbalance settlement references the DA price (as in some schemes), any intraday trade at a worse price is a cost, not a hedge
Example: Wind portfolio adjustment
14:00 D-1: DA position = sold 200 MW for hour 14 based on 210 MW wind forecast. DA price: 45 €/MWh.
22:00 D-1: Updated forecast = 180 MW. Need to buy back 20 MW. ID price available: 52 €/MWh. Cost: 20 × (52−45) = 140 €.
10:00 D (delivery day): Forecast drops further to 160 MW. Buy back 20 MW more. ID price now: 85 €/MWh. Cost: 20 × (85−45) = 800 €.
Residual imbalance: Actual generation = 155 MW. Short 5 MW settled at imbalance price: 130 €/MWh. Cost: 5 × (130−45) = 425 €.
Key insight: Earlier trading (at 52 €) was cheaper than later (85 €), but you didn't yet know you'd need it. The optimization tradeoff is forecast accuracy vs. intraday price trajectory — trading too early exposes you to forecast error, too late exposes you to illiquidity premium.
Intraday Products: Granularity Matters
Intraday markets trade at finer granularity than day-ahead in most zones:
| Product | Duration | Key Use |
|---|
| Hourly | 60 min | Bulk position adjustment; legacy product |
| Half-hourly | 30 min | UK settlement period matching; ramp capture |
| Quarter-hourly | 15 min | Matches imbalance settlement period (DE); captures intra-hour ramps (solar noon, wind ramps) |
| Block products | Multiple hours | Baseload/peakload position changes; less common in ID |
Why 15-minute matters: Germany's imbalance settlement period is 15 minutes. If you only trade hourly in intraday, you can still be exposed to within-hour profile risk. A flat 100 MW hourly sale versus a solar ramp (120 MW in Q1, 80 MW in Q4) creates a 15-min imbalance even if the hourly net is zero. Quarter-hourly trading eliminates this exposure.
Cross-Border: The Shared Order Book (SIDC)
The Single Intraday Coupling (SIDC, formerly XBID) connects local order books across borders with implicit capacity allocation:
- Shared Order Book (SOB): Orders visible to all participants across coupled zones. A German seller can be matched with a French buyer if capacity exists.
- Capacity Management Module (CMM): Tracks available cross-zonal capacity in real-time. Each cross-border match consumes capacity atomically — no over-allocation.
- Shipping Agent: When a cross-border match occurs, automatically creates the nomination to the relevant TSOs.
Key limitation: capacity is allocated first-come, first-served. Unlike DA's welfare-maximizing allocation, the first trader to match across a border "uses up" capacity that might have been more valuable to a later trader. This is why intraday auctions were introduced — to periodically re-optimize the allocation.
Optimization Strategies for Intraday
| Strategy | Actor | Mechanism | Risk |
|---|
| Forecast-driven rebalancing | RES portfolio | Buy/sell as forecasts update; trade volume = forecast delta | Trading too early (forecast reverts) or too late (illiquidity) |
| Spread trading | Storage / flex | Buy low-price hours, sell high-price hours within intraday | Pay-as-bid means less transparent spreads; execution risk |
| Ramping capture | Flexible generation | Sell quarter-hourly products during steep ramp periods (sunset, wind drops) | Ramp events may not materialize; VWAP vs. marginal price confusion |
| Cross-border arbitrage | Trading desks | Exploit price differentials between zones via SIDC SOB | Capacity may be exhausted; GCT for cross-border earlier than local |
| Imbalance price speculation | BRPs with flex | Deliberately leave position open if expected imbalance price is favorable | Imbalance prices are volatile and asymmetric; regulatory scrutiny |
The Fundamental Optimization Problem
For an intraday optimizer managing a portfolio, the decision at each moment is:
At time t, for delivery hour h:
Expected cost of trading now = (ID price at t) × (volume to adjust)
Expected cost of waiting = E[ID price at t+Δ] × (volume) + P(no liquidity) × E[imbalance cost]
Trade if: marginal cost of immediacy < expected cost of waiting
This is a
stochastic optimal stopping problem — related to optimal execution in financial markets (Almgren-Chriss framework adapted for energy). The key inputs are:
- Forecast uncertainty trajectory (how much will the forecast change?)
- Intraday price model (mean-reversion? momentum? jump diffusion?)
- Liquidity model (bid-ask spread as function of time-to-delivery)
- Imbalance price distribution (asymmetric, fat-tailed)
Intraday vs. Day-Ahead: Structural Differences
| Dimension | Day-Ahead | Intraday |
|---|
| Pricing | Uniform (marginal) | Pay-as-bid (continuous) / Uniform (IDAs) |
| Objective | Maximize total welfare | Bilateral matching (continuous) / Welfare (auctions) |
| Information | All orders revealed simultaneously at gate closure | Orders arrive sequentially; asymmetric information rewarded |
| Cross-border | Welfare-optimal allocation (EUPHEMIA) | First-come-first-served (continuous) / Optimal (IDAs) |
| Volumes | ~80-90% of total traded volume | ~10-20%, but growing rapidly with RES penetration |
| Function | Physical scheduling, investment signal | Position adjustment, risk management |
Your tangible win: You now understand the full journey from DA schedule to gate closure — the two ID mechanisms (continuous + auctions), why both exist, how cross-border works differently in each, and the fundamental optimization tradeoff (trade now vs. wait for information). Combined with Lesson 2, you can trace how a position cascades from DA through ID to balancing.
Primary Source
For the operational design of SIDC and the shared order book:
ENTSO-E / NEMO Committee, Single Intraday Coupling (SIDC) — Operational Design. Available at: nemo-committee.eu/sidc
For the academic treatment of intraday price dynamics and trading strategies:
Kiesel, R. & Paraschiv, F. (2017). "Econometric analysis of 15-minute intraday electricity prices." Energy Economics, 64, 77–90. — Foundational paper on quarter-hourly price formation.
Narajewski, M. & Ziel, F. (2020). "Econometric modelling and forecasting of intraday electricity prices." Journal of Commodity Markets, 19. — State-of-the-art forecasting approaches for continuous ID.
Retrieval Practice
Why does the SIDC continuous market produce worse cross-border allocation than the day-ahead auction?
- The SIDC algorithm uses a simpler optimization without network constraints or PTDFs
- Capacity is allocated first-come-first-served rather than to the highest-value trades
- Cross-border intraday trades are limited to a fixed percentage of the ATC value
- Continuous trading operates on hourly products only while DA uses quarter-hourly
Retrieval Practice
A German wind portfolio discovers at 08:00 on delivery day that its forecast for hour 14 has dropped. Why might it prefer to trade quarter-hourly products rather than hourly?
- Quarter-hourly products have lower trading fees on EPEX SPOT than hourly products
- The cross-border gate closure for quarter-hourly products is later than for hourly
- The wind drop may affect specific 15-min periods within the hour, matching the imbalance settlement granularity
- Quarter-hourly products are exempt from the REMIT reporting obligations that hourly face
Retrieval Practice
What economic role do the intraday auctions (IDAs) serve that continuous trading alone cannot provide?
- They allow participants to submit block orders which continuous trading does not support
- They aggregate information into a single price and re-optimize cross-border capacity allocation
- They provide a legally binding nomination to the TSO which continuous trades do not create
- They enable trading of products with delivery periods shorter than fifteen minutes
Retrieval Practice
The optimal intraday trading problem for a RES portfolio is best characterized as:
- A linear program minimizing transaction costs subject to position limits and ramp constraints
- A game-theoretic Nash equilibrium between the portfolio and other market participants
- A stochastic optimal stopping problem balancing immediacy costs against forecast uncertainty
- A convex optimization maximizing expected profit using historical intraday price patterns
💡 Want me to go deeper on intraday price modelling, the mechanics of SIDC 2.0 auctions, how to formulate the optimal execution problem mathematically, or how specific assets (storage, demand response) optimise in intraday? Just ask.
Rechtliches: Impressum