Handling Non-Continuous Markets (Fixed / Call Auctions)

[FranekGit]

Hi everyone,

I’m working with non-continuous markets (e.g. scheduled fixings, daily open/close auctions) where orders accumulate and match at a single uncrossing price rather than through continuous matching.

What are the current capabilities and best practices in Nautilus Trader for handling these phases? Specifically:

• Venue States: What is the recommended way to model transitions between continuous trading, halts, and call phases?

• Auction Data: Are there specific event types to feed in indicative match prices and order imbalances during a call phase?

• Simulated Matching: Does the backtester support call auction logic (single equilibrium price), or is it strictly continuous? If continuous, what is the best workaround?

• Order Types: How should we handle auction-specific orders like MOO, MOC, or LOC?

If native support isn't there yet, I’d appreciate any architectural pointers on building a custom adapter or matching engine extension for this.

Thanks!

[Aliipou]

NautilusTrader is primarily designed for continuous order book matching, so call auction / fixing support is limited but workable. Here is the current state:

Venue States / Trading Phases

There is no first-class AuctionPhase enum today. The recommended approach is to model phase transitions via custom InstrumentStatus events:

from nautilus_trader.model.events import InstrumentStatus
from nautilus_trader.model.enums import MarketStatus

status = InstrumentStatus(
    instrument_id=instrument_id,
    status=MarketStatus.PRE_OPEN,  # or PAUSE, CLOSED
    ...
)
engine.add_data([status], sort=False)

Your strategy receives on_instrument_status(event) and can halt quoting during auction phases.

Auction / Fixing Price Feed

Feed indicative uncrossing prices as QuoteTick or custom Data objects keyed to the auction timestamp. There is no dedicated AuctionIndicativePrice type — a QuoteTick with bid == ask == indicative price is the standard workaround.

Simulated Matching During Auctions

The built-in fill model is continuous (FIFO, price-time priority). For call auction semantics (single-price batch matching) you would need a custom FillModel implementation — this is the main gap. There is no built-in batch/uncrossing matching engine.

Practical recommendation

For backtesting with known fixing prices: pre-process your data so the fixing print arrives as a single TradeTick at the auction timestamp with the uncrossing quantity. The strategy can detect the fixing via on_trade_tick when tick.size is abnormally large relative to continuous flow.

Is this for a specific exchange / fixing type? Happy to dig into the details.