v3

v3 is a breaking change to the initialisation of the framework that allows the user to have full control over data streams and strategy subscriptions whilst removing the dependency of a Betfair client.

This means an instance can be setup to execute on either individual venues or multiple concurrently, current status:

• Betfair: Completed
• BETDAQ: Completed
• BetConnect: Development
• Polymarket: Roadmap
• Kalshi: Roadmap
• BetDEX: Roadmap
• Smarkets: Roadmap
• Matchbook: Roadmap

Breaking Changes

Live

Historically flumine would create streams based on provided market filters however from v3 this has moved to the following design:

import betfairlightweight
from flumine import Flumine, clients
from flumine.streams.betfairmarketstream import BetfairMarketStream
from betfairlightweight.filters import streaming_market_filter

# Initialize your client
trading = betfairlightweight.APIClient("username")
client = clients.BetfairClient(trading)

# Initialize the framework
framework = Flumine(client)

# Add your strategy to the framework with a stream
framework.add_strategy(
    ExampleStrategy(
        stream=BetfairMarketStream(
            framework,
            market_filter=streaming_market_filter(
                event_type_ids=["7"],
                country_codes=["GB"],
                market_types=["WIN"],
            ),
        )
    )
)

# Start the trading framework
framework.run()

A strategy can therefore subscribe to multiple streams and a stream can be subscribed to multiple strategies, these can also be across VenueTypes.

A single stream with multiple strategies:

# Create stream(s) (market data)
stream = BetfairMarketStream(
    framework,
    market_filter=streaming_market_filter(
        event_type_ids=["7"],
        country_codes=["GB"],
        market_types=["WIN"],
    ),
)

# create strategy and subscribe to stream(s)
strategy = ExampleStrategy(name="one", stream=stream)
framework.add_strategy(strategy)

strategy = ExampleStrategy(name="two", stream=stream)
framework.add_strategy(strategy)

Multiple streams with a single strategy:

# Create stream(s) (market data)
stream_one = BetfairMarketStream(
    framework,
    market_filter=streaming_market_filter(
        event_type_ids=["7"],
        country_codes=["GB"],
        market_types=["WIN"],
    ),
)
stream_two = BetfairMarketStream(
    framework,
    market_filter=streaming_market_filter(
        event_type_ids=["1"],
        country_codes=["GB"],
        market_types=["MATCH_ODDS"],
    ),
)

# create strategy and subscribe to stream(s)
strategy = ExampleStrategy(name="one", streams=[stream_one, stream_two])
framework.add_strategy(strategy)

It is also possible to subscribe or unsubscribe after a strategy has been initialised:

strategy = ExampleStrategy()
strategy.subscribe_to_stream(stream)
strategy.unsubscribe_from_stream(stream)

Simulation

The simulation setup has also changed to match live as well as removing framework, renaming market_filter to file_path and moving listener_kwargs to its own variable.

import betfairlightweight
from flumine import FlumineSimulation, clients
from flumine.streams.betfairhistoricalstream import BetfairHistoricalStream

# Initialize your client
client = clients.SimulatedClient()

# Initialize the framework
framework = FlumineSimulation(client=client)

# Add your strategy to the framework with a stream (this can be a list)
file_path="tests/resources/PRO-1.170258213"
strategy = LowestLayer(
    stream=BetfairHistoricalStream(
        file_path=file_path,
        listener_kwargs={"inplay": True}
    ),
    max_order_exposure=1000,
    max_selection_exposure=105,
    context={"stake": 2},
)
framework.add_strategy(strategy)

# Start the trading framework
framework.run()

Note

There are no other changes to the public / exposed API of flumine in v2->v3.