Implementing custom alpha model

To implement a custom Alpha model in QuantConnect, you can follow these steps:

1. Define your Alpha model class: Your Alpha model class should extend the AlphaModel class. This class must implement the Update method, which receives a Slice object and returns an array of Insight objects. Extensions should also implement the OnSecuritiesChanged method to track security changes in the universe. Here is an example in Python:

class MyAlphaModel(AlphaModel):
    def Update(self, algorithm, data):
        insights = []
        # Your logic here
        return insights

(Source: Key Concepts)

1. Add your Alpha model to your algorithm: In the Initialize method of your algorithm, call the AddAlpha method and pass an instance of your Alpha model.

self.AddAlpha(MyAlphaModel())

(Source: Key Concepts)

1. Assign a unique name to your Alpha model: To ensure your Alpha model is compatible with all Portfolio Construction models, assign a unique name to your Alpha model. By default, the class-type name is used as the Alpha model name.

class RsiAlphaModel(AlphaModel):
    Name = "RsiAlphaModel"

(Source: Key Concepts)

1. Generate Insights: The Alpha model predicts market trends and signals the best moments to trade. These signals, or Insight objects, contain the Direction, Magnitude, and Confidence of a market prediction and the suggested portfolio Weight. You should generate insights on the set of assets provided by the Universe Selection model and only generate them when your predictions change.

2. Backtest and Live Trade: Once you have implemented your Alpha model and added it to your algorithm, you can backtest your algorithm using QuantConnect's backtesting framework. If the backtest results are satisfactory, you can then deploy your algorithm for live trading.

Remember, developing an Alpha model requires a good understanding of the financial markets and the specific assets you are trading. You should also be familiar with the Python programming language and the QuantConnect API.

Hi, I keep getting this error 

Runtime Error: Types deriving from 'AlphaModel' must implement the 'IEnumerable<Insight> Update(QCAlgorithm, Slice) method. in AlphaModel.cs:line 50 (Open Stack Trace)

despite having this def in the class:

    def update(self, algorithm, data) -> List[Insight]:
    	insights = []
    	...
    	return insights

Been trying with the samples such as NullAlphaModel but can't work. Can anyone advise?

I narrowed down and find that if my main.py adds an Equity like 

self.add_equity("MSFT", self.set_reso).Symbol

then the custom alphamodel fails to run. 

Found the true issue, for some reason, cannot use this in main

self.alpha = SomeAlphaModel()

Hi Kelvin

Indeed, since self.alpha is already a callable method in the QCAlgorithm class, we cannot override it with other objects, which will break the algorithm. We suggest you might use self._alpha instead.

Best
Louis