Do we need to explicitly add Security to calculate automatic EMA values?

Log Up

Here is my attempt at what you are trying to do in your algorithm by initializing the indicators in OnSecuritiesChanged().

If you are satisfied with my answer, please accept it and don't forget to like it.

Hi Vladimir ,Thank you for the reply. I saw the code you mentioned, but it seems it initializes the indicators when we get the changed securities right?(Please correct me if I am wrong here). What I want to achieve is to filter out top 10 stocks which has highest momentum of 50 EMA above 200 EMA. For that I need to have indicators in CoarseSelectionFunction to sort the securities.

Log Up
In my code, the algorithm selects stocks with the largest dollar volume in CoarseSelectionFunction(), 
then initializes self.ema_f and self.ema_s in OnSecuritiesChanged(), and selects stocks with positive EMA momentum (self.ema_f > self.ema_s) in OnData(). 
What is not in line with your original request to directly call self.EMA(symbol)?

Hi Vladimir, I am attempting to initialize EMA while in CoarseSelectionFunction() itself to prevent unnecessary additions of securities which will be called on OnData(), but it seems the security is not added into algorithm(hence no history for indicator also) while in CoarseSelectionFunction(), needing to call addSecurity() explicitly. Is there any way we can only have securities in algorithm which pass ema_f>ema_s at the time of universe selection itself? 

Vladimir because I don't only want ema_f>ema_s , I want top n which has highest ema_f>ema_s

You can request history for all securities in CoarseSelectionFunction, use this data to calculate EMAs, sort and choose top symbols. 

Direct answer to your question would be that yes, in order for automatic EMA calculation, you need to have security added to the universe, otherwise there will be no data sent to the indicator function. But it wouldn't be optimal processing all stocks minute data only to calculate daily EMAs, that is why I recommend using History API.

Hi Yuri Lopukhov , Thanks for the reply. Now when you confirm it, I got it. I am processing the universe on daily resolution hence was thinking of calculating indicator in CoarseSelectionFunction. By History API can you please tell me if you mean something like below? If yes, then it seems here also “SPY” has been first  added into algorithm to populate indicator right?

def Initialize(self):
    self.AddEquity("SPY", Resolution.Hour)
    # define a 10-period daily RSI indicator with shortcut helper method
    self.rsi = self.RSI("SPY", 10,  MovingAverageType.Simple, Resolution.Daily)
    # initialize the indicator with the daily history close price
    history = self.History(["SPY"], 10, Resolution.Daily)
        for time, row in history.loc["SPY"].iterrows():
            self.rsi.Update(time, row["close"])

Log Up

Here is “top dollar volume top moving average ratio universe” based on self.History() in CoarseSelectionFunction().

If you are satisfied with my answer, please accept it and don't forget to like it.

Hi Log Up,

To sum up:

• Indicators like EMA, RSI, etc, need data, of course.
• Inside the Universe Selection filter function, you can choose to select assets you never selected before. If you do so AND decide to create an indie on them, it will complain about not having data because you are essentially trying to register an indie (with the helpers SMA, EMA, etc, the indie is auto-registered) on a symbol which is not scheduled to receive data. It will only be subscribed later, when your filter function returns.
• So in a nutshell, when you use helper indie functions, you need to pass assets for which your algo is subscribed. But you can do so anywhere in your code, as long as the flow order (symbol subscribed - helper indie) is correct.

Fred

Log Up

-→I want top n which has highest ema_f>ema_s

Here is “Top Dollar Volume Top EMAR universe” which selects symbols exactly as you requested.

self.selectedSymbols = sorted(self.emar, key = lambda x: self.emar[x])[:N_MOM] 

If you are satisfied with my answer, please accept it and don't forget to like it.