please code up an algo and add it as a backtest so we can interact with your code.

Okay here it is.

Any idea why it doesn't work?

Seriously? No one had this problem before? Everybody here manually types in the symbols they want to get data for?

I find that hard to believe.

I'd start with this and keep enhancing it:

# Put any initialization logic here.  The context object will be passed to  
# the other methods in your algorithm.  
import datetime  
import pytz  
import numpy as np



def initialize(context):  
    symbol('AAPL') # seed data dict  
def handle_data( context , data ):  
    for tkr in data:  
        pricenow = data[tkr].price  

Thank you Dan!

Let me try it.

Dan,

So just to clarify,

symbol('AAPL') # seed data dict

This part can't become:

symbol( some_tkr )

Is that right? Because that's what I'm trying to not have to type in manually.

Thanks!!

# Try this:  
def initialize(context):  
    # I should declare my symbols here.  
    # Side effect: data-dict gets filled with prices and other data.  
    symbols('AAPL','IBM','SPY')  
def handle_data( context , data ):  
    for tkr in data:  
        print(tkr.symbol)  
        print(data[tkr].price)  

Dan, thanks for the repeated responses!

But I think you are not getting my question. I need to NOT type in the symbols manually into the 'symbol' or 'symbols' object, like you did in your answer. I need to pass them in as a parameter.

# like this:  
def initialize( context ):  
    # generate a list of symbols here, then  
    symbols( this_tkr, that_tkr )  
    # but it gives me an error saying this_tkr is not a string argument, which it is.

What I need to do is look at the prices of multiple stocks, say, every day at the same time. How do you do this?

Hi Ning,

You can't have a string representing the symbol you want because all the symbols must be known before starting the backtest.

But you could have something like this:

def initialize(context):  
    context.stock1 = symbol('AAPL')  
    context.stock2 = symbol('IBM')  
    context.listOfStocks = symbols('GS','C','SPY')  
def handle_data( context , data ):  
    for tkr in data:  # AAPL,IBM,GS,C,SPY  
        print(tkr.symbol)  
        print(data[tkr].price)  

If you want a random bunch of stocks you can use set_universe() instead:

    set_universe(universe.DollarVolumeUniverse(98, 99))  

Hope that helps

Thanks for the response James!

I'm not using a random selection of stocks.

The symbols() function can only take 255 arguments, that's the problem here. I wouldn't mind manually typing it in if I have to.

But that just seems odd that in a library for running trading algorithms it would be so hard to load more than a few stock prices.

There must be something I'm missing here, right?

If I'm doing more than 20 or so I just use a text editor with regular expression search/replace to build the symbols() list and paste it into the algo. I wasn't actually aware of the 255 cap. I think if you use set_universe() then the most you will ever get is about 400 (??) depending on how many stocks are listed on the exchange at that point in time. I think Quantopian's reason was due to the long running-time for algos using so many symbols. Perhaps someone from Q can confirm :)

Thanks James!

I think it makes sense.

I hope there is a way to look at over 400 stocks on every iteration with Quantopian. Maybe there is, and someone with the knowledge can help us out!

To chime in here,

1. The stocks need to explicitly called in the backtest - you can't pass a variable into "symbol". This is because we scan the algorithm for code errors and malicious intent before we run it on our servers.

2. You can trade up to 200 securities per bar of the backtest. This cap exists for performance reasons so that we can maintain a good backtesting experience - otherwise it's a lot of data! A couple months ago we increased the cap from 100 to 200 securities and I imagine as we continue to grow, we'll raise the cap again.

I hope this helps clear up any lingering confusion!

Thanks for the clarification Alisa!

So you said, you can trade 200 securities, but can I load the prices for more than 200?

After so many years, the bug still exists, quite disappointing

when it came to futures, you have to type all the symbols manually, very tired

As you noticed, the 'future_symbol' method works a bit different in the IDE vs the notebook environment. The symbol argument must be a string, and not a variable, in the IDE. This isn't really a bug. Just a difference. Not a way to work around this behavior.