History and SymbolData

Hi Nicholas,

Short answer is: it depends. :)

Long answer is: would you mind sharing some code? Sharing objects depend on if they are mutable or not, and if they are mutable, how they are mutated, when they are mutated, how they are reused, etc.

Medium answer is: normally, the class can be reused and history calls can happen pretty much anywhere.

Fred

I have the code available here, though currently its very broken as I'm stumbling my way through the quantconnect API. 

The fundamentals are:

1. Scan for stocks where the bollinger bands are inside the keltner channels, add then to the universe
2. Place a trade when the bollinger bands exit the keltner channels

So you can see why I would like to re-use my symboldata class as they essentially re-use the same data.

The code is attached below, which is in the main.py file. I've commented out my symboldata class and coarse selection function so as to be able to post it.

Thanks,

Nick

Hi Nick,

As you say, the code is not complete but now that I understand what you want to do, I would say that YES you can re-use your dictionary of SymbolData objects between Universe Selection and Alpha. Essentially, your Universe Selection will update your SymbolData objects with new universe selections and will update the indies and your Alpha will read your SymbolData objects to provide insights.

I was being careful about maybe you wanting to update your SymbolData objects at multiple places. While there is nothing bad about it, it can create unwanted behaviour if it is done without careful consideration. Here, you won't have “multiple writers” but only one so it's fine.

You're on the right path. Do come back if you need more help later.

Cheers,

Fred

Hi Fred,

Thank you for your advice! I appreciate the time you've taken to have a look and provide some advice. I will indeed. For now, I have one further question. I'm having difficulty with my history request in the SymbolData class, could you suggest how to write this section of code as well as how to apply it to my symbols? 

Sincere regards,

Nick

Sure:

class SymbolData(object):

    # Make sure symbol is a Symbol object and not a string ticker
    def __init__(self, symbol):
        self.symbol = symbol
        self.bollinger = BollingerBands(20, 2, MovingAverageType.Simple)
        self.keltner = KeltnerChannels(20, 1.5, MovingAverageType.Simple)
        
        history = self.History(symbol, 20, Resolution.Daily)
        for bar in history.itertuples():
            tradeBar = TradeBar(bar.time - timedelta(1), symbol, bar.open, bar.high, bar.low, bar.close, bar.volume, timedelta(1))
            self.bollinger.Update(bar.time, bar.close)
            self.keltner.Update(tradeBar)

#    def Scale(self):
#        return (self.bollinger.UpperBand.Current.Value - self.bollinger.LowerBand.Current.Value)
        
#    def is_ready(self):
#        return self.bollinger.IsReady and self.keltner.IsReady

It's getting late and we're not working with compiling backtests so I'm not 100% sure it compiles but it should be close. I could have worked on a complete backtest but I believe you will learn much more by doing it yourself. Come back if you need more help.

Hi Fred, 

Thank you for providing the code. You're right and I'll work through applying it to the symbols myself.

All the best,

Nick

Hi Fred,

I've been able to access the history request in SymbolData by passing in the algorithm parameter so I can call algorithm.History(…..). However this means I need to pass in the algorithm parameter to the coarse selection function in order to initialise the SymbolData class on my symbols. I've been pouring through the forums and documentation trying to work out how to access the algorithm parameter through the coarse selection function, but it seems it only accepts 2 arguements (self, coarse). I've attached the code below. Could you point me in the right direction so I can achieve this? 

Regards,

Nick

Sure Nick.

In CoarseSelection, self is the algo 😊.

So you'll be able to do “SymbolData(self, symbol)”. Yeah, makes a lot of “self”…

You'll need some more edits here and there in your commented out CoarseSelection method but it's heading in the right direction.

Fred

Hi Fred,

Haha it does indeed, thank you for clearing that up.

All the best,

Nick