Algorithm Framework
Universe Selection
Universe Selection
Introduction
The Universe Section Model creates Universe objects which select the assets for your strategy. We have identified three types of universes that cover most people's requirements, and built helper classes to make their implementation easier. Creating Universe Selection Models from scratch isn't simple so we recommend you use one of the helper universes we've provided.
To set a Universe Selection Model you should use the AddUniverseSelectionself.AddUniverseSelection method. This should be done from your algorithm Initialize()def Initialize() method:
public override void Initialize()
{
UniverseSettings.Resolution = Resolution.Minute;
var symbols = new [] { QuantConnect.Symbol.Create("SPY", SecurityType.Equity, Market.USA) };
AddUniverseSelection( new ManualUniverseSelectionModel(symbols) );
}
def Initialize(self): # Set requested data resolution
self.UniverseSettings.Resolution = Resolution.Minute
symbols = [ Symbol.Create("SPY", SecurityType.Equity, Market.USA) ]
self.AddUniverseSelection( ManualUniverseSelectionModel(symbols) )
As with all LEAN algorithms; the resolution of the assets added to the universe is configured by the UniverseSettings.Resolutionself.UniverseSettings.Resolution property.
There are 3 types of universes:
- Manual Universes - Universes which use a fixed, static set of assets.
- Fundamental Universes - Universes based on coarse price or fundamental data.
- Scheduled Universes - Universes triggering on regular intervals.
Manual Universe Selection
Manual universe selection chooses a static, fixed set of assets to trade. This is most useful when selecting a set of currencies, or a basket of ETF stocks. You can also add assets with the traditional AddSecurity API.
The ManualUniverseSelectionModel class is initialized with an array of Symbol objects at the UniverseSettings.Resolution data resolution. Symbol objects can be created with the Symbol.Create method.
public override void Initialize()
{
// Set requested data resolution
UniverseSettings.Resolution = Resolution.Minute;
AddUniverseSelection(
new ManualUniverseSelectionModel(
QuantConnect.Symbol.Create("SPY", SecurityType.Equity, Market.USA)
));
}
def Initialize(self):
self.UniverseSettings.Resolution = Resolution.Minute
symbols = [ Symbol.Create("SPY", SecurityType.Equity, Market.USA) ]
self.AddUniverseSelection(ManualUniverseSelectionModel(symbols))
Fundamental Universe Selection
Framework Universe Selection models can use the same function based selection mechanics as other algorithms. QuantConnect provides two helper methods for these universes which handle the framework requirements: CoarseFundamentalUniverseSelectionModel and the FineFundamentalUniverseSelectionModel.
To define a fundamental Universe Selection model you need to create an instance of the class and set with the AddUniverseSelection method:
// Setting Universe Model in QCAlgorithm
public override void Initialize()
{
AddUniverseSelection(new FineFundamentalUniverseSelectionModel(SelectCoarse, SelectFine));
}
IEnumerable<Symbol> SelectCoarse(IEnumerable<CoarseFundamental> coarse)
{
var tickers = new[] { "AAPL", "AIG", "IBM" };
return tickers.Select(x =>
QuantConnect.Symbol.Create(x, SecurityType.Equity, Market.USA)
);
}
IEnumerable<Symbol> SelectFine(IEnumerable<FineFundamental> fine) {
return fine.Select(f => f.Symbol);
}
# Setting Universe Model in QCAlgorithm
def Initialize(self):
self.AddUniverseSelection(
FineFundamentalUniverseSelectionModel(self.SelectCoarse, self.SelectFine)
)
def SelectCoarse(self, coarse):
tickers = ["AAPL", "AIG", "IBM"]
return [Symbol.Create(x, SecurityType.Equity, Market.USA) for x in tickers]
def SelectFine(self, fine):
return [f.Symbol for f in fine]
The fundamental universes perform the same filtering as the traditional algorithm explained in the Universe section. The Coarse selection function is passed a list of CoarseFundamental objects should return a list of Symbol objects. The Fine selection function is passed a subset of FineFundamental objects generated from coarse selection results, and should return a list of Symbol objects. See the Universe section for more information on these filtering functions.
Scheduled Universe Selection
Scheduled universes allow you to perform universe selection at fixed, regular intervals. In live trading this might be applied to fetching tickers from Dropbox, or performing analysis on historical data and choosing resulting symbols. The class for creating scheduled universes is called ScheduledUniverseSelectionModel.
public ScheduledUniverseSelectionModel(
DateRule dateRule,
TimeRule timeRule,
Func<DateTime, IEnumerable<Symbol>> selector,
UniverseSettings settings = null,
ISecurityInitializer initializer = null
)
ScheduledUniverseSelectionModel(dateRule, timeRule, selector, universeSettings=null, securityInitializer=null)
The universe selection helper works in the same way as the Scheduled Event API, requiring a DateRule, a TimeRule to set the callback times, and a function to execute which returns a list of Symbol objects.
// Selection will run on mon/tues/thurs at 00:00/06:00/12:00/18:00
AddUniverseSelection(new ScheduledUniverseSelectionModel(
DateRules.Every(DayOfWeek.Monday, DayOfWeek.Tuesday, DayOfWeek.Thursday),
TimeRules.Every(TimeSpan.FromHours(6)),
SelectSymbols // selection function in algorithm.
));
// Create selection function which returns symbol objects.
IEnumerable<Symbol> SelectSymbols(DateTime dateTime)
{
return new[]
{
Symbol.Create("SPY", SecurityType.Equity, Market.USA),
Symbol.Create("AAPL", SecurityType.Equity, Market.USA),
Symbol.Create("IBM", SecurityType.Equity, Market.USA)
}
}
# Selection will run on mon/tues/thurs at 00:00/06:00/12:00/18:00
self.AddUniverseSelection(ScheduledUniverseSelectionModel(
self.DateRules.Every(DayOfWeek.Monday, DayOfWeek.Tuesday, DayOfWeek.Thursday),
self.TimeRules.Every(timedelta(hours = 12)),
self.SelectSymbols # selection function in algorithm.
))
# Create selection function which returns symbol objects.
def SelectSymbols(self, dateTime):
symbols = []
symbols.append(Symbol.Create('SPY', SecurityType.Equity, Market.USA))
symbols.append(Symbol.Create('SPY', SecurityType.Equity, Market.USA))
return symbols
Creating Universe Models
Universe Models must implement a IUniverseSelectionModel interface. It has one method, CreateUniverses(QCAlgorithm algorithm). The algorithm object is passed into the method to give you access to the QuantConnect API and it should return an array of Universe objects.
// Algorithm framework model that defines the universes to be used by an algorithm
interface IUniverseSelectionModel
{
// Creates the universes for this algorithm, called once after IAlgorithm.Initialize
IEnumerable<Universe> CreateUniverses(QCAlgorithmFramework algorithm);
}
Generally you should be able to extend one of the universes described above so if you ever find yourself needing to do something which doesn't fit into the categories above, please let us know and we'll create a new foundational type of universe model.
Configuring Securities
To configure securities in a universe you should use the SetSecurityInitializer() method. Call this from your Initialize method and set a ISecurityInitializer class, or use the functional implementation demonstrated below for simple requests. This feature is described in detail in the Configuring Universe Securities section.
//Most common request; requesting raw prices for universe securities. SetSecurityInitializer(x => x.SetDataNormalizationMode(DataNormalizationMode.Raw));
# Most common request; requesting raw prices for universe securities. self.SetSecurityInitializer(lambda x: x.SetDataNormalizationMode(DataNormalizationMode.Raw))
You can also see our Tutorials and Videos. You can also get in touch with us via Chat.
