Equity
Legacy Fundamental Universes
Introduction
Warning: This API for Universe Selection was deprecated on November 2023. Please refer to the new Fundamental Universe API.
There are several ways to create an Equities universe. You can select a universe based on CoarseFundamental data
or the constituents of an ETF, and then you can further filter your universe down with corporate fundamentals.
The following sections explain each of these techniques in detail.
Coarse Universe Selection
A coarse universe enables you pick a set of stocks based on their trading volume, price, or whether they have fundamental data. To add a coarse universe, in the Initializeinitialize method, pass a filter function to the AddUniverseadd_universe method. The coarse filter function receives a list of CoarseFundamental objects and must return a list of Symbol objects. The Symbol objects you return from the function are the constituents of the universe and LEAN automatically creates subscriptions for them. Don't call AddEquityadd_equity in the filter function.
public class MyCoarseUniverseAlgorithm : QCAlgorithm
{
public override void Initialize()
{
UniverseSettings.Asynchronous = true;
AddUniverse(CoarseFilterFunction);
}
private IEnumerable<Symbol> CoarseFilterFunction(IEnumerable<CoarseFundamental> coarse)
{
return (from c in coarse
orderby c.DollarVolume descending
select c.Symbol).Take(100);
}
} class MyCoarseUniverseAlgorithm(QCAlgorithm):
def initialize(self) -> None:
self.universe_settings.asynchronous = True
self.add_universe(self.coarse_filter_function)
def coarse_filter_function(self, coarse: List[CoarseFundamental]) -> List[Symbol]:
sorted_by_dollar_volume = sorted(coarse, key=lambda x: x.dollar_volume, reverse=True)
return [c.symbol for c in sorted_by_dollar_volume[:100]]
CoarseFundamental objects have the following attributes:
The total number of stocks in the US Equity Security Master dataset is 30,000 but your coarse filter function won't receive all of these at one time because the US Equity Security Master dataset is free of survivorship bias and some of the securities have delisted over time. The number of securities that are passed into your coarse filter function depends on the date of your algorithm. Currently, there are about 10,000 securities that LEAN passes into your coarse filter function.
Fundamentals Selection
A fundamental universe lets you select stocks based on corporate fundamental data. This data is powered by Morningstar® and includes approximately 8,100 tickers with 900 properties each. Due to the sheer volume of information, fundamental selection is performed on the output of another universe filter. Think of this process as a 2-stage filter. An initial filter function selects a set of stocks and then a fine fundamental filter function selects a subset of those stocks.
To add a fundamental universe, in the Initializeinitialize method, pass two filter functions to the AddUniverseadd_universe method. The first filter function can be a coarse universe filter, dollar volume filter, or an ETF constituents filter. The second filter function receives a list of FineFundamental objects and must return a list of Symbol objects. The list of FineFundamental objects contains a subset of the Symbol objects that the first filter function returned. The Symbol objects you return from the second function are the constituents of the fundamental universe and LEAN automatically creates subscriptions for them. Don't call AddEquityadd_equity in the filter function.
Only 8,100 assets have fundamental data. If your first filter function receives CoarseFundamental data, you should only select assets that have a true value for their HasFundamentalDatahas_fundamental_data property.
public class MyUniverseAlgorithm : QCAlgorithm {
public override void Initialize()
{
UniverseSettings.Asynchronous = true;
AddUniverse(CoarseFilterFunction, FineFundamentalFilterFunction);
}
// filter based on CoarseFundamental
IEnumerable<Symbol> CoarseFilterFunction(IEnumerable<CoarseFundamental> coarse)
{
// In addition to further coarse universe selection, ensure the security has fundamental data
return (from c in coarse
where c.HasFundamentalData
select c.Symbol);
}
// filter based on FineFundamental
public IEnumerable<Symbol> FineFundamentalFilterFunction(IEnumerable<FineFundamental> fine)
{
// Return a list of Symbols
}
}
class MyUniverseAlgorithm(QCAlgorithm):
def initialize(self) -> None:
self.universe_settings.asynchronous = True
self.add_universe(self.coarse_filter_function, self.fine_fundamental_function)
def coarse_filter_function(self, coarse: List[CoarseFundamental]) -> List[Symbol]:
# In addition to further coarse universe selection, ensure the security has fundamental data
return [c.symbol for c in coarse if c.has_fundamental_data]
def fine_fundamental_function(self, fine: List[FineFundamental]) -> List[Symbol]:
# Return a list of Symbols
FineFundamental objects have the following attributes:
Example
The simplest example of accessing the fundamental object would be harnessing the iconic PE ratio for a stock. This is a ratio of the price it commands to the earnings of a stock. The lower the PE ratio for a stock, the more affordable it appears.
// Take the top 50 by dollar volume using coarse
// Then the top 10 by PERatio using fine
UniverseSettings.Asynchronous = true;
AddUniverse(
coarse => {
return (from c in coarse
where c.Price > 10 && c.HasFundamentalData
orderby c.DollarVolume descending
select c.Symbol).Take(50);
},
fine => {
return (from f in fine
orderby f.ValuationRatios.PERatio ascending
select f.Symbol).Take(10);
});
# In Initialize:
self.universe_settings.asynchronous = True
self.add_universe(self.coarse_selection_function, self.fine_selection_function)
def coarse_selection_function(self, coarse: List[CoarseFundamental]) -> List[Symbol]:
sorted_by_dollar_volume = sorted(coarse, key=lambda x: x.dollar_volume, reverse=True)
filtered = [x.symbol for x in sorted_by_dollar_volume if x.has_fundamental_data]
return filtered[:50]
def fine_selection_function(self, fine: List[FineFundamental]) -> List[Symbol]:
sorted_by_pe_ratio = sorted(fine, key=lambda x: x.valuation_ratios.pe_ratio, reverse=False)
return [x.symbol for x in sorted_by_pe_ratio[:10]]
Asset Categories
In addition to valuation ratios, the US Fundamental Data from Morningstar has many other data point attributes, including over 200 different categorization fields for each US stock. Morningstar groups these fields into sectors, industry groups, and industries.
Sectors are large super categories of data. To get the sector of a stock, use the MorningstarSectorCode property.
var tech = fine.Where(x => x.AssetClassification.MorningstarSectorCode == MorningstarSectorCode.Technology);
tech = [x for x in fine if x.asset_classification.morningstar_sector_code == MorningstarSectorCode.TECHNOLOGY]
Industry groups are clusters of related industries that tie together. To get the industry group of a stock, use the MorningstarIndustryGroupCode property.
var ag = fine.Where(x => x.AssetClassification.MorningstarIndustryGroupCode == MorningstarIndustryGroupCode.Agriculture);
ag = [x for x in fine if x.asset_classification.morningstar_industry_group_code == MorningstarIndustryGroupCode.AGRICULTURE]
Industries are the finest level of classification available. They are the individual industries according to the Morningstar classification system. To get the industry of a stock, use the MorningstarIndustryCode.
var coal = fine.Where(x => x.AssetClassification.MorningstarIndustryCode == MorningstarSectorCode.Coal);
coal = [x for x in fine if x.asset_classification.morningstar_industry_code == MorningstarSectorCode.COAL]
Practical Limitations
Like coarse universes, fine universes allow you to select an unlimited universe of assets to analyze. Each asset in the universe consumes approximately 5MB of RAM, so you may quickly run out of memory if your universe filter selects many assets. If you backtest your algorithms in the Algorithm Lab, familiarize yourself with the RAM capacity of your backtesting and live trading nodes. To keep your algorithm fast and efficient, only subscribe to the assets you need.
Live Trading Considerations
The live data for fundamental universe selection arrives at 6/7 AM Eastern Time (ET), so fundamental universe selection runs for live algorithms between 7 and 8 AM ET. This timing allows you to place trades before the market opens. Don't schedule anything for midnight because the universe selection data isn't ready yet.
Examples
The following examples are typical filter functions you may want.
Example 1: Take 500 stocks that are worth more than $10 and have more than $10M daily trading volume
The most common use case is to select a lot of liquid stocks. With a coarse universe filter, this is simple and fast. The following example selects the top most liquid 500 stocks over $10 per share.
IEnumerable<Symbol> CoarseFilterFunction(IEnumerable<CoarseFundamental> coarse)
{
// Linq makes this a piece of cake;
return (from c in coarse
where c.DollarVolume > 10000000 &&
c.Price > 10
orderby c.DollarVolume descending
select c.Symbol).Take(500);
} def coarse_filter_function(self, coarse: List[CoarseFundamental]) -> List[Symbol]:
sorted_by_dollar_volume = sorted(coarse, key=lambda x: x.dollar_volume, reverse=True)
filtered = [ x.symbol for x in sorted_by_dollar_volume
if x.price > 10 and x.dollar_volume > 10000000 ]
return filtered[:500]
Example 2: Take 10 stocks above their 200-Day EMA and have more than $1B daily trading volume
Another common request is to filter the universe by a technical indicator, such as only picking stocks above their 200-day EMA. The CoarseFundamental object has adjusted price and volume information, so you can do any price-related analysis.
ConcurrentDictionary<Symbol, SelectionData>
_stateData = new ConcurrentDictionary<Symbol, SelectionData>();
// Coarse filter function
IEnumerable<Symbol> CoarseFilterFunction(IEnumerable<CoarseFundamental> coarse) {
// Linq makes this a piece of cake;
return (from c in coarse
let avg = _stateData.GetOrAdd(c.Symbol, sym => new SelectionData(200))
where avg.Update(c.EndTime, c.AdjustedPrice)
where c.DollarVolume > 1000000000 &&
c.Price > avg.Ema
orderby c.DollarVolume descending
select c.Symbol).Take(10);
} # setup state storage in initialize method
self.state_data = { }
def coarse_filter_function(self, coarse: List[CoarseFundamental]) -> List[Symbol]:
# We are going to use a dictionary to refer the object that will keep the moving averages
for c in coarse:
if c.symbol not in self.state_data:
self.state_data[c.symbol] = SelectionData(c.symbol, 200)
# Updates the SymbolData object with current EOD price
avg = self.state_data[c.symbol]
avg.update(c.end_time, c.adjusted_price, c.dollar_volume)
# Filter the values of the dict to those above EMA and more than $1B vol.
values = [x for x in self.state_data.values() if x.is_above_ema and x.volume > 1000000000]
# sort by the largest in volume.
values.sort(key=lambda x: x.volume, reverse=True)
# we need to return only the symbol objects
return [ x.symbol for x in values[:10] ]
In this example, the SelectionData class group variables for the universe selection and updates the indicator of each asset. We highly recommend you follow this pattern to keep your algorithm tidy and bug free. The following snippet shows an example implementation of the SelectionData class, but you can make this whatever you need to store your custom universe filters.
class SelectionData(object):
def __init__(self, symbol, period):
self._symbol = symbol
self.ema = ExponentialMovingAverage(period)
self.is_above_ema = False
self.volume = 0
def update(self, time, price, volume):
self.volume = volume
if self.ema.update(time, price):
self.is_above_ema = price > ema // example selection data class
private class SelectionData
{
// variables you need for selection
public readonly ExponentialMovingAverage Ema;
// initialize your variables and indicators.
public SelectionData(int period)
{
Ema = new ExponentialMovingAverage(period);
}
// update your variables and indicators with the latest data.
// you may also want to use the History API here.
public bool Update(DateTime time, decimal value)
{
return Ema.Update(time, value);
}
}
Note that the preceding SelectionData class uses a manual EMA indicator instead of the automatic version. For more information about universes that select assets based on indicators, see Indicator Universes. You need to use a SelectionData class instead of assigning the EMA to the CoarseFundamental object because you can't create custom propertiesattributes on CoarseFundamental objects like you can with Security objects.
Example 3: Take 10 stocks that are the furthest above their 10-day SMA of volume
The process to get the 10-day SMA stock volume is the same process as in Example 2. First, you should define a SelectionData class that performs the averaging. For this example, the following class will serve this purpose:
class SelectionData(object):
def __init__(self, symbol, period):
self._symbol = symbol
self.volume = 0
self.volume_ratio = 0
self._sma = SimpleMovingAverage(period)
def update(self, time, price, volume):
self.volume = volume
if self.sma.update(time, volume):
# get ratio of this volume bar vs previous 10 before it.
self.volume_ratio = volume / self.sma.current.value private class SelectionData
{
public readonly Symbol Symbol;
public readonly SimpleMovingAverage VolumeSma;
public decimal VolumeRatio;
public SelectionData(Symbol symbol, int period)
{
Symbol = symbol;
VolumeSma = new SimpleMovingAverage(period);
}
public bool Update(DateTime time, decimal value)
{
var ready = VolumeSma.Update(time, value);
VolumeRatio = value / VolumeSma;
return ready;
}
}
This class tracks the ratio of today's volume relative to historical volumes. You can use this ratio to select assets that are above their 10-day simple moving average and sort the results by the ones that have had the biggest jump since yesterday.
def coarse_filter_function(self, coarse: List[CoarseFundamental]) -> List[Symbol]:
for c in coarse:
if c.symbol not in self.state_data:
self.state_data[c.symbol] = SelectionData(c.symbol, 10)
avg = self.state_data[c.symbol]
avg.update(c.end_time, c.adjusted_price, c.dollar_volume)
# filter the values of selectionData(sd) above SMA
values = [sd for sd in self.state_data.values() if sd.volume > sd.sma.current.value and sd.volume_ratio > 0]
# sort sd by the largest % jump in volume.
values.sort(key=lambda sd: sd.volume_ratio, reverse=True)
# return the top 10 symbol objects
return [ sd.symbol for sd in values[:10] ]
IEnumerable<Symbol> CoarseFilterFunction(IEnumerable<CoarseFundamental> coarse)
{ return (from c in coarse let avg = _stateData.GetOrAdd(c.Symbol, sym => new SelectionData(10)) where avg.Update(c.EndTime, c.Volume) where c.Volume > avg.VolumeSma orderby avg.VolumeRatio descending select c.Symbol).Take(10); }
Example 4: Take the top 10 "fastest moving" stocks with a 50-Day EMA > 200 Day EMA
You can construct complex universe filters with the SelectionData helper class pattern. To view a full example of this algorithm, see the EmaCrossUniverseSelectionAlgorithmEmaCrossUniverseSelectionAlgorithm in the LEAN GitHub repository or take the related Boot Camp lesson.
You can also see our Videos. You can also get in touch with us via Discord.
Did you find this page helpful?
