Universes
Futures
Introduction
A Futures universe lets you select a basket of contracts for a single Future. LEAN models Future subscriptions as a universe of Future contracts. A Future universe is similar to an Option universe, except Future contracts don't have a strike price, so the universe filter primarily focuses on the contract expiration date.
Create Universes
To add a universe of Future contracts, in the Initialize
initialize
method, call the AddFuture
add_future
method. This method returns an Future
object, which contains the continuous contract Symbol
symbol
. The continuous contract Symbol
symbol
is the key to access the contracts in the FutureChain that LEAN passes to the OnData
on_data
method. The continuous contract Mapped
mapped
property references the current contract in the continuous contract series. When you create the Future subscription, save a reference to the Future
object so you can use it later in your algorithm.
public class BasicFutureAlgorithm : QCAlgorithm { private Future _future; public override void Initialize() { UniverseSettings.Asynchronous = true; _future = AddFuture(Futures.Currencies.BTC, extendedMarketHours: true, dataMappingMode: DataMappingMode.LastTradingDay, dataNormalizationMode: DataNormalizationMode.BackwardsRatio, contractDepthOffset: 0); _future.SetFilter(0, 62); } public override void OnData(Slice data) { if (Portfolio.Invested) { return; } data.Bars.TryGetValue(_future.Symbol, out var continuousTradeBar); data.Bars.TryGetValue(_future.Mapped, out var mappedTradeBar); MarketOrder(_future.Mapped, 1); } // Track events when security changes its ticker, allowing the algorithm to adapt to these changes. public override void OnSymbolChangedEvents(SymbolChangedEvents symbolChangedEvents) { foreach (var (symbol, changedEvent) in symbolChangedEvents) { var oldSymbol = changedEvent.OldSymbol; var newSymbol = changedEvent.NewSymbol; var quantity = Portfolio[oldSymbol].Quantity; // Rolling over: to liquidate any position of the old mapped contract and switch to the newly mapped contract var tag = $"Rollover - Symbol changed at {Time}: {oldSymbol} -> {newSymbol}"; Liquidate(oldSymbol, tag: tag); if (quantity != 0) MarketOrder(newSymbol, quantity, tag: tag); } } }
class BasicFutureAlgorithm(QCAlgorithm): def initialize(self): self.universe_settings.asynchronous = True self._future = self.add_future(Futures.Currencies.BTC, extended_market_hours=True, data_mapping_mode=DataMappingMode.LAST_TRADING_DAY, data_normalization_mode=DataNormalizationMode.BACKWARDS_RATIO, contract_depth_offset=0) self._future.set_filter(0,62) def on_data(self, data): if self.portfolio.invested: return continuous_trade_bar = data.bars.get(self._future.symbol) mapped_trade_bar = data.bars.get(self._future.mapped) self.market_order(self._future.mapped, 1) # Track events when security changes its ticker allowing algorithm to adapt to these changes. def on_symbol_changed_events(self, symbol_changed_events): for symbol, changed_event in symbol_changed_events.items(): old_symbol = changed_event.old_symbol new_symbol = changed_event.new_symbol quantity = self.portfolio[old_symbol].quantity # Rolling over: to liquidate any position of the old mapped contract and switch to the newly mapped contract tag = f"Rollover - Symbol changed at {self.time}: {old_symbol} -> {new_symbol}" self.liquidate(old_symbol, tag=tag) if quantity: self.market_order(new_symbol, quantity, tag=tag)
The following table describes the AddFuture
add_future
method arguments:
Argument: |
Argument: |
Argument: |
Argument: |
Argument: |
Argument: |
Argument: |
Argument: |
Argument: |
Continous Contracts
By default, LEAN only subscribes to the continuous Future contract. A continuous Future contract represents a series of separate contracts stitched together to form a continuous price. If you need a lot of historical data to warm up an indicator, apply the indicator to the continuous contract price series. The Future
object has a Symbol
symbol
property and a Mapped
mapped
property. The price of the Symbol
symbol
property is the adjusted price of the continuous contract. The price of the Mapped
mapped
property is the raw price of the currently selected contract in the continuous contract series.
// Get the adjusted price of the continuous contract. var adjustedPrice = Securities[_future.Symbol].Price; // Get the raw price of the currently selected contract in the continuous contract series. var rawPrice = Securities[_future.Mapped].Price;
# Get the adjusted price of the continuous contract. adjusted_price = self.securities[self._future.symbol].price # Get the raw price of the currently selected contract in the continuous contract series. raw_price = self.securities[self._future.mapped].price
The continuous Futures contract isn't a tradable security. You must place orders for a specific Futures contract. To access the currently selected contract in the continuous contract series, use the Mapped
mapped
property of the Future
object.
// Place a market order for the currently selected contract in the continuous contract series. MarkerOrder(_future.Mapped, 1);
# Place a market order for the currently selected contract in the continuous contract series. self.market_order(self._future.mapped, 1)
To configure how LEAN identifies the current Future contract in the continuous series and how it forms the adjusted price between each contract, provide dataMappingMode
data_mapping_mode
, dataNormalizationMode
data_normalization_mode
, and contractDepthOffset
contract_depth_offset
arguments to the AddFuture
add_future
method. The Future
object that the AddFuture
add_future
method returns contains a Mapped
mapped
property that references the current contract in the continuous contract series. As the contracts roll over, the Mapped
mapped
property references the next contract in the series and you receive a SymbolChangedEvent
object in the OnData
on_data
method. The SymbolChangedEvent
references the old contract Symbol
and the new contract Symbol
. You can use SymbolChangedEvents
to roll over contracts.
public class BasicFutureAlgorithm : QCAlgorithm { // Track when the continuous contract switches from one contract to the next. public override void OnSymbolChangedEvents(SymbolChangedEvents symbolChangedEvents) { foreach (var (symbol, changedEvent) in symbolChangedEvents) { var oldSymbol = changedEvent.OldSymbol; var newSymbol = changedEvent.NewSymbol; var quantity = Portfolio[oldSymbol].Quantity; // Rolling over: To liquidate the old mapped contract and switch to the new mapped contract. var tag = $"Rollover - Symbol changed at {Time}: {oldSymbol} -> {newSymbol}"; Liquidate(oldSymbol, tag: tag); if (quantity != 0) MarketOrder(newSymbol, quantity, tag: tag); } } }
class BasicFutureAlgorithm(QCAlgorithm): # Track when the continuous contract switches from one contract to the next. def on_symbol_changed_events(self, symbol_changed_events): for symbol, changed_event in symbol_changed_events.items(): old_symbol = changed_event.old_symbol new_symbol = changed_event.new_symbol quantity = self.portfolio[old_symbol].quantity # Rolling over: To liquidate the old mapped contract and switch to the new mapped contract. tag = f"Rollover - Symbol changed at {self.time}: {old_symbol} -> {new_symbol}" self.liquidate(old_symbol, tag=tag) if quantity: self.market_order(new_symbol, quantity, tag=tag)
In backtesting, the SymbolChangedEvent
occurs at midnight Eastern Time (ET). In live trading, the live data for continuous contract mapping arrives at 6/7 AM ET, so that's when it occurs.
Data Normalization Modes
The dataNormalizationMode
data_normalization_mode
argument defines how the price series of two contracts are stitched together when the contract rollovers occur. The following DataNormalizatoinMode
enumeration members are available for continuous contracts:
We use the entire Futures history to adjust historical prices. This process ensures you get the same adjusted prices, regardless of the backtest end date.
Data Mapping Modes
The dataMappingMode
data_mapping_mode
argument defines when contract rollovers occur. The DataMappingMode
enumeration has the following members:
Futures.Indices.VIX
(VX) doesn't support continous contract rolling with DataMappingMode.OpenInterest
DataMappingMode.OPEN_INTEREST
and DataMappingMode.OpenInterestAnnuak
DataMappingMode.OPEN_INTEREST_ANNUAL
.
Contract Depth Offsets
The contractDepthOffset
contract_depth_offset
argument defines which contract to use. 0 is the front month contract, 1 is the following back month contract, and 2 is the second back month contract.
Filter Contracts
By default, LEAN doesn't add any contracts to the FuturesChain it passes to the OnData
on_data
method.
To set a contract filter, in the Initialize
initialize
method, call the SetFilter
set_filter
method of the Future
object. The following table describes the available filter techniques:
SetFilter(int minExpiry, int maxExpiry) set_filter(minExpiry: int, maxExpiry: int) Selects the contracts that expire within the range you set. This filter runs asynchronously by default. |
SetFilter(Func<FutureFilterUniverse, FutureFilterUniverse> universeFunc) set_filter(universeFunc: Callable[[FutureFilterUniverse], FutureFilterUniverse]) Selects the contracts that a function selects. |
# Select the contracts that expire within 182 days. self._future.set_filter(0, 182) # Select the front month contract. self._future.set_filter(lambda future_filter_universe: future_filter_universe.front_month())
// Select the contracts that expire within 182 days. _future.SetFilter(0, 182); // Select the front month contract. _future.SetFilter(futureFilterUniverse => futureFilterUniverse.FrontMonth());
The following table describes the filter methods of the FutureFilterUniverse
class:
StandardsOnly() standards_only() Selects standard contracts |
IncludeWeeklys() include_weeklys() Selects non-standard weekly contracts |
WeeklysOnly() weeklys_only() Selects weekly contracts |
FrontMonth() front_month() Selects the front month contract |
BackMonths() back_months() Selects the non-front month contracts |
BackMonth() back_month() Selects the back month contracts |
Expiration(TimeSpan minExpiry, TimeSpan maxExpiry) expiration(min_expiry: timedelta, max_expiry: timedelta) Selects contracts that expire within a range of dates relative to the current day |
Expiration(int minExpiryDays, int maxExpiryDays) expiration(min_expiry_days: int, max_expiry_days: int) Selects contracts that expire within a range of dates relative to the current day |
Contracts(IEnumerable<Symbol> contracts) contracts(contracts: List[Symbol]) Selects a list of contracts |
Contracts(Func<IEnumerable<Symbol>, IEnumerable< Symbol>> contractSelector) contracts(contractSelector: Callable[[List[Symbol]], List[Symbol]]) Selects contracts that a selector function selects |
The preceding methods return an FutureFilterUniverse
, so you can chain the methods together.
// Select the front month standard contracts _future.SetFilter(futureFilterUniverse => futureFilterUniverse.StandardsOnly().FrontMonth());
# Select the front month standard contracts self._future.set_filter(lambda future_filter_universe: future_filter_universe.standards_only().front_month())
You can also define an isolated filter method.
// In Initialize _future.SetFilter(Selector); private FutureFilterUniverse Selector(FutureFilterUniverse futureFilterUniverse) { return futureFilterUniverse.StandardsOnly().FrontMonth(); }
# In Initialize self._future.set_filter(self._contract_selector) def _contract_selector(self, future_filter_universe: Callable[[FutureFilterUniverse], FutureFilterUniverse]) -> FutureFilterUniverse: return future_filter_universe.standards_only().front_month()
Navigate Futures Chains
FuturesChain
objects represent an entire chain of contracts for a single underlying Future. They have the following properties:
To get the FuturesChain
, index the FuturesChains
futures_chains
property of the Slice
with the continuous contract Symbol
.
public override void OnData(Slice slice) { if (slice.FuturesChains.TryGetValue(_symbol, out var chain)) { // Example: Select the contract with the greatest open interest var contract = chain.OrderBy(x => x.OpenInterest).Last(); } }
def on_data(self, slice: Slice) -> None: chain = slice.futures_chains.get(self._symbol) if chain: # Example: Select the contract with the greatest open interest contract = sorted(chain, key=lambda contract: contract.open_interest, reverse=True)[0]
You can also loop through the FuturesChains
futures_chains
property to get each FuturesChain
.
public override void OnData(Slice slice) { foreach (var kvp in slice.FuturesChains) { var continuousContractSymbol = kvp.Key; var chain = kvp.Value; } } public void OnData(FuturesChains futuresChains) { foreach (var kvp in futuresChains) { var continuousContractSymbol = kvp.Key; var chain = kvp.Value; } }
def on_data(self, slice: Slice) -> None: for continuous_contract_symbol, chain in slice.futures_chains.items(): pass
Examples
The following examples demonstrate some common Futures universes.
Example 1: Contracts Expiring On Quarter End
If you run strategies with a quarterly trading cycle, you can filter Future contracts that expire only on quarter ends to reduce the need of rollovers.
public class QuarterlyFuturesAlgorithm : QCAlgorithm { public override void Initialize() { var future = AddFuture("ES"); future.SetFilter(futureFilterUniverse => futureFilterUniverse.ExpirationCycle(new int[] { 3, 6, 9, 12 })); } }
class QuarterlyFuturesAlgorithm(QCAlgorithm): def initialize(self): future = self.add_future("ES") future.set_filter(lambda future_filter_universe: future_filter_universe.expiration_cycle([3, 6, 9, 12]))
Example 2: Contracts Expiring Within One Week
The trading volume of Futures reaches its maximum during the last week of the trading period. The following algorithm selects contracts that expire within a week:
public class WeeklyFuturesAlgorithm : QCAlgorithm { public override void Initialize() { var future = AddFuture("ES"); future.SetFilter(futureFilterUniverse => futureFilterUniverse.Expiration(0, 7)); } }
class WeeklyFuturesAlgorithm(QCAlgorithm): def initialize(self): future = self.add_future("ES") future.set_filter(lambda future_filter_universe: future_filter_universe.expiration(0, 7))
Other Examples
For more examples, see the following algorithms: