Python model interop: subclassable SlippageModel base, algorithm-level SetSlippageModel, None framework models, callable security seeder

Algorithm.CSharp/AlgorithmSlippageModelRegressionAlgorithm.cs

@@ -0,0 +1,150 @@
+/*
+ * QUANTCONNECT.COM - Democratizing Finance, Empowering Individuals.
+ * Lean Algorithmic Trading Engine v2.0. Copyright 2014 QuantConnect Corporation.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+*/
+
+using System.Collections.Generic;
+using QuantConnect.Data;
+using QuantConnect.Interfaces;
+using QuantConnect.Orders;
+using QuantConnect.Orders.Slippage;
+using QuantConnect.Securities;
+
+namespace QuantConnect.Algorithm.CSharp
+{
+    /// <summary>
+    /// Regression algorithm asserting that the algorithm-level <see cref="QCAlgorithm.SetSlippageModel(ISlippageModel)"/>
+    /// applies a custom <see cref="SlippageModel"/> subclass to all securities,
+    /// with per-security models set afterwards taking precedence
+    /// </summary>
+    public class AlgorithmSlippageModelRegressionAlgorithm : QCAlgorithm, IRegressionAlgorithmDefinition
+    {
+        private CustomSlippageModel _slippageModel;
+        private Symbol _spy;
+        private Symbol _ibm;
+
+        public override void Initialize()
+        {
+            SetStartDate(2013, 10, 07);
+            SetEndDate(2013, 10, 11);
+            SetCash(100000);
+
+            SetSecurityInitializer(new BrokerageModelSecurityInitializer(BrokerageModel, new FuncSecuritySeeder(GetLastKnownPrices)));
+
+            _slippageModel = new CustomSlippageModel();
+            SetSlippageModel(_slippageModel);
+
+            _spy = AddEquity("SPY", Resolution.Minute).Symbol;
+            var ibm = AddEquity("IBM", Resolution.Minute);
+            // per-security models set after the algorithm-level model take precedence for that security
+            ibm.SetSlippageModel(NullSlippageModel.Instance);
+            _ibm = ibm.Symbol;
+        }
+
+        public override void OnData(Slice slice)
+        {
+            if (!Portfolio.Invested)
+            {
+                SetHoldings(_spy, 0.5m);
+                SetHoldings(_ibm, 0.5m);
+            }
+        }
+
+        public override void OnEndOfAlgorithm()
+        {
+            if (Securities[_spy].SlippageModel != _slippageModel)
+            {
+                throw new RegressionTestException("Expected SPY to use the algorithm-level slippage model");
+            }
+            if (Securities[_ibm].SlippageModel != NullSlippageModel.Instance)
+            {
+                throw new RegressionTestException("Expected the per-security slippage model to take precedence for IBM");
+            }
+            if (_slippageModel.CallCount == 0)
+            {
+                throw new RegressionTestException("Expected the algorithm-level slippage model to have been used");
+            }
+        }
+
+        private class CustomSlippageModel : SlippageModel
+        {
+            public int CallCount { get; private set; }
+
+            public override decimal GetSlippageApproximation(Security asset, Order order)
+            {
+                CallCount++;
+                return 0.05m;
+            }
+        }
+
+        /// <summary>
+        /// This is used by the regression test system to indicate if the open source Lean repository has the required data to run this algorithm.
+        /// </summary>
+        public bool CanRunLocally { get; } = true;
+
+        /// <summary>
+        /// This is used by the regression test system to indicate which languages this algorithm is written in.
+        /// </summary>
+        public List<Language> Languages { get; } = new() { Language.CSharp, Language.Python };
+
+        /// <summary>
+        /// Data Points count of all timeslices of algorithm
+        /// </summary>
+        public long DataPoints => 7843;
+
+        /// <summary>
+        /// Data Points count of the algorithm history
+        /// </summary>
+        public int AlgorithmHistoryDataPoints => 20;
+
+        /// <summary>
+        /// Final status of the algorithm
+        /// </summary>
+        public AlgorithmStatus AlgorithmStatus => AlgorithmStatus.Completed;
+
+        /// <summary>
+        /// This is used by the regression test system to indicate what the expected statistics are from running the algorithm
+        /// </summary>
+        public Dictionary<string, string> ExpectedStatistics => new Dictionary<string, string>
+        {
+            {"Total Orders", "2"},
+            {"Average Win", "0%"},
+            {"Average Loss", "0%"},
+            {"Compounding Annual Return", "343.438%"},
+            {"Drawdown", "2.100%"},
+            {"Expectancy", "0"},
+            {"Start Equity", "100000"},
+            {"End Equity", "101922.49"},
+            {"Net Profit", "1.922%"},
+            {"Sharpe Ratio", "10.891"},
+            {"Sortino Ratio", "0"},
+            {"Probabilistic Sharpe Ratio", "66.279%"},
+            {"Loss Rate", "0%"},
+            {"Win Rate", "0%"},
+            {"Profit-Loss Ratio", "0"},
+            {"Alpha", "0.565"},
+            {"Beta", "0.993"},
+            {"Annual Standard Deviation", "0.232"},
+            {"Annual Variance", "0.054"},
+            {"Information Ratio", "7.794"},
+            {"Tracking Error", "0.071"},
+            {"Treynor Ratio", "2.545"},
+            {"Total Fees", "$3.55"},
+            {"Estimated Strategy Capacity", "$16000000.00"},
+            {"Lowest Capacity Asset", "IBM R735QTJ8XC9X"},
+            {"Portfolio Turnover", "19.93%"},
+            {"Drawdown Recovery", "3"},
+            {"OrderListHash", "c4766cde15ad208b5f6c12c6a0af59b9"}
+        };
+    }
+}

Algorithm.Python/AlgorithmSlippageModelRegressionAlgorithm.py

@@ -0,0 +1,71 @@
+# QUANTCONNECT.COM - Democratizing Finance, Empowering Individuals.
+# Lean Algorithmic Trading Engine v2.0. Copyright 2014 QuantConnect Corporation.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from AlgorithmImports import *
+
+### <summary>
+### Regression algorithm asserting that the algorithm-level set_slippage_model() applies a custom
+### SlippageModel subclass to all securities, with per-security models set afterwards taking precedence.
+### It also asserts that BrokerageModelSecurityInitializer accepts a plain callable as security seeder
+### and that None is accepted by the framework model setters as the null model.
+### </summary>
+class AlgorithmSlippageModelRegressionAlgorithm(QCAlgorithm):
+
+    def initialize(self):
+        self.set_start_date(2013, 10, 7)
+        self.set_end_date(2013, 10, 11)
+        self.set_cash(100000)
+
+        # the security seeder can be a plain callable, it gets wrapped into a FuncSecuritySeeder
+        self.set_security_initializer(BrokerageModelSecurityInitializer(self.brokerage_model, self.get_last_known_prices))
+
+        # None is accepted by the framework model setters as the null model
+        self.set_risk_management(None)
+        self.set_execution(None)
+
+        self._slippage_model = CustomSlippageModel()
+        self.set_slippage_model(self._slippage_model)
+
+        self._spy = self.add_equity("SPY", Resolution.MINUTE).symbol
+        ibm = self.add_equity("IBM", Resolution.MINUTE)
+        # per-security models set after the algorithm-level model take precedence for that security
+        ibm.set_slippage_model(NullSlippageModel.INSTANCE)
+        self._ibm = ibm.symbol
+
+    def on_data(self, data):
+        if not self.portfolio.invested:
+            self.set_holdings(self._spy, 0.5)
+            self.set_holdings(self._ibm, 0.5)
+
+    def on_end_of_algorithm(self):
+        if not isinstance(self.securities[self._ibm].slippage_model, NullSlippageModel):
+            raise AssertionError("Expected the per-security slippage model to take precedence for IBM")
+        if self._slippage_model.call_count == 0:
+            raise AssertionError("Expected the algorithm-level slippage model to have been used")
+        if not isinstance(self.risk_management, NullRiskManagementModel):
+            raise AssertionError("Expected set_risk_management(None) to set the null risk management model")
+        if not isinstance(self.execution, NullExecutionModel):
+            raise AssertionError("Expected set_execution(None) to set the null execution model")
+
+### <summary>
+### Custom slippage model derived from the C# SlippageModel base class.
+### The ISlippageModel interface cannot be used as a Python base class.
+### </summary>
+class CustomSlippageModel(SlippageModel):
+    def __init__(self):
+        super().__init__()
+        self.call_count = 0
+
+    def get_slippage_approximation(self, asset, order):
+        self.call_count += 1
+        return 0.05

Algorithm/QCAlgorithm.Framework.Python.cs

@@ -32,6 +32,13 @@ public partial class QCAlgorithm
         [DocumentationAttribute(AlgorithmFramework)]
         public void SetAlpha(PyObject alpha)
         {
+            // None is accepted as the null model, e.g. to disable a model a template set up
+            if (alpha is null || alpha.IsNone())
+            {
+                SetAlpha(new NullAlphaModel());
+                return;
+            }
+
             Alpha = PythonUtil.CreateInstanceOrWrapper<IAlphaModel>(
                 alpha,
                 py => new AlphaModelPythonWrapper(py)
@@ -60,6 +67,13 @@ public void AddAlpha(PyObject alpha)
         [DocumentationAttribute(TradingAndOrders)]
         public void SetExecution(PyObject execution)
         {
+            // None is accepted as the null model, e.g. to disable a model a template set up
+            if (execution is null || execution.IsNone())
+            {
+                SetExecution(new NullExecutionModel());
+                return;
+            }
+
             Execution = PythonUtil.CreateInstanceOrWrapper<IExecutionModel>(
                 execution,
                 py => new ExecutionModelPythonWrapper(py)
@@ -74,6 +88,13 @@ public void SetExecution(PyObject execution)
         [DocumentationAttribute(TradingAndOrders)]
         public void SetPortfolioConstruction(PyObject portfolioConstruction)
         {
+            // None is accepted as the null model, e.g. to disable a model a template set up
+            if (portfolioConstruction is null || portfolioConstruction.IsNone())
+            {
+                SetPortfolioConstruction(new NullPortfolioConstructionModel());
+                return;
+            }
+
             PortfolioConstruction = PythonUtil.CreateInstanceOrWrapper<IPortfolioConstructionModel>(
                 portfolioConstruction,
                 py => new PortfolioConstructionModelPythonWrapper(py)
@@ -88,6 +109,13 @@ public void SetPortfolioConstruction(PyObject portfolioConstruction)
         [DocumentationAttribute(Universes)]
         public void SetUniverseSelection(PyObject universeSelection)
         {
+            // None is accepted as the null model, e.g. to disable a model a template set up
+            if (universeSelection is null || universeSelection.IsNone())
+            {
+                SetUniverseSelection(new NullUniverseSelectionModel());
+                return;
+            }
+
             UniverseSelection = PythonUtil.CreateInstanceOrWrapper<IUniverseSelectionModel>(
                 universeSelection,
                 py => new UniverseSelectionModelPythonWrapper(py)
@@ -117,6 +145,15 @@ public void AddUniverseSelection(PyObject universeSelection)
         [DocumentationAttribute(TradingAndOrders)]
         public void SetRiskManagement(PyObject riskManagement)
         {
+            // None is accepted as the null model, e.g. to disable a model a template set up.
+            // Without this, set_risk_management(None) failed with
+            // "IRiskManagementModel must be fully implemented. Please implement these missing methods on NoneType: ManageRisk"
+            if (riskManagement is null || riskManagement.IsNone())
+            {
+                SetRiskManagement(new NullRiskManagementModel());
+                return;
+            }
+
             RiskManagement = PythonUtil.CreateInstanceOrWrapper<IRiskManagementModel>(
                 riskManagement,
                 py => new RiskManagementModelPythonWrapper(py)

Algorithm/QCAlgorithm.Python.cs

@@ -32,6 +32,7 @@
 using QuantConnect.Util;
 using QuantConnect.Interfaces;
 using QuantConnect.Orders;
+using QuantConnect.Orders.Slippage;
 using QuantConnect.Commands;
 using QuantConnect.Api;
 
@@ -1437,6 +1438,23 @@ public void SetRiskFreeInterestRateModel(PyObject model)
             SetRiskFreeInterestRateModel(riskFreeInterestRateModel);
         }
 
+        /// <summary>
+        /// Sets the slippage model for all securities in the algorithm, including securities added afterwards,
+        /// for instance through universe selection
+        /// </summary>
+        /// <remarks>Individual securities can override this model by calling
+        /// <see cref="Security.SetSlippageModel(PyObject)"/> after this method, e.g. from
+        /// <see cref="OnSecuritiesChanged(SecurityChanges)"/> for securities added by universe selection</remarks>
+        /// <param name="slippageModel">The slippage model to use</param>
+        [DocumentationAttribute(Modeling)]
+        public void SetSlippageModel(PyObject slippageModel)
+        {
+            SetSlippageModel(PythonUtil.CreateInstanceOrWrapper<ISlippageModel>(
+                slippageModel,
+                py => new SlippageModelPythonWrapper(py)
+            ));
+        }
+
         /// <summary>
         /// Sets the security initializer function, used to initialize/configure securities after creation
         /// </summary>