Quantopian Shutdown Migration 2020

Hi Joshua, 

the issue is caused by the following line

ret[symbol] = (self.price[symbol][-1] - self.price[symbol][116]) / self.price[symbol][116]

because there was at least one symbol with len(self.price[symbol]) < 116. 

So you could add the statement 

if len(self.price[symbol]) > 116:
  # your code here

But there are still a lot of bugs in the algorithm and other error messages will appear. 

I recommend to complete the bootcamp and study the documentation. 

A few comments in front: 

- use Symbol objects instead of strings. I know pandas automatically convert it to strings since it cannot handle custom object types but you can convert it back via self.Symbol(symbol_string). 

- always check if data is available before entering a position via SetHoldings: 

if data.ContainsKey(symbol) and data[symbol] is not None:
  self.SetHoldings(symbol, weight)

- avoid too many History() calls since they are very expensive

- AddEquity() should be located within Initialize()

- we can reduce the method calc_return() to a single line by doing: 

returns = history.unstack().close.transpose().iloc[[-116, -1], :].pct_change().dropna().reset_index(drop=True).transpose().sort_values(by=0,ascending=False).rename(columns={0:"returns"})

As other before me have said, Alphalens is the most important feature Quantopian had and I think QC should implement something similar, sooner or later. 

I started about a 1 1/2 year ago learning python and algorithmic trading. I used to split my time between Quantopian and Quantconnect. Mainly Quantopian mainly because there was a rich history of lectures and deep dives into subjects. I am disappointed that Quantopian decided to take down their lectures before closing. On the list of my many projects, I had wanted to transfer all of the lectures into Quantconnect as a method to perfect my programming skills. 

In the future, I have to use skylight to ensure I have all of my code locally as a backup. 

Last, Quantconnect I hope that your platform is embraced and utilized by developers, authors, and researchers. If anything were to happen, please give us a fair warning to save our work.  

Hello,

Has anyone managed to convert the "Quality Companies in an Uptrend" algorithm or any variant of it please?

Regards,

Russell

Hi - thanks for existing! Coming over to your platform from Quantopian

I had a Quantopian Algorithm where I used both technical and fundamental data as factors for a Random Forest.  I spent some time trying to move it to QC but it didn't seem like there was a straightforward way of doing it.  I remember that Fundamentals data could be used to filter stocks, and started trying to store information about the fundamentals while filtering in order to use it in the next step of the process.

I gave up because it felt too much like I was using the QC for a purpose other than it was intended for, but now that Quantopian has shut down it may be worth it for me to revisit.

Any help would be appreciated

Russell Cox 

Check out the version in the newest forum posts 

Excited to dig into QC and LEAN. One thing I really liked about zipline that I'd love to see supported: realistic continuous futures contracts. For most strategies you need some degree of continuous futures data, but are numerous problems using a single continuous series stitching back as much as 30 years. Zipline did a better job: on any backtested trading day it would give you a series based on what a continuous series would like if it were stitched together at that point in history. Notably the front month contract at that point in time will be untouched. This allows for much more realistic fills, volatility, etc. in the backtest.

Something zipline didn't do well: multiple currencies. It would be great to trade futures in different markets/currencies and the backtester understand the exhange rates allow us to properly size positions based on portfolio equity in the primary currency of the theoretical account.

Hi Jon,

We are working on adding continuous futures, so continuous futures will be available for use in algorithms in the future.

Best,
Shile Wen

Vladimir 

Server Status

Can I edit or delete comments?
Where can I find more information about the QuantConnect Community Forum?

Hi Vladimir,

Forum posts are only able to be edited for a few minutes after they are published.

We are currently in the process of adding more documentation for the forum.

The Status page is available here.

Best,
Derek Melchin

Derek Melchin

I just want to thank you for helping me translate my algorithm into a QuantConnect API
in this thread, but found that this discussion is closed.
Can you unlock it because I have questions about this algorithm?
Is there any explanation for the rules of discussion?

Derek Melchin

Is there a way to see more than 5 of my last posts desirably all?

Not at this time Vladimir, thank you for the suggestion.

Is there anything in the QuantConnect similar to the Quantopian

AverageDollarVolume pipeline factor?

    AverageDollarVolume(mask = QTradableStocksUS(), window_length = 63)

Hi Vladimir,

There is nothing similar in QC at the moment, so I suggest creating a dictionary of SimpleMovingAverage objects keyed by Symbol and update the SMA with the .DollarVolume field inside a coarse function. Furthermore, I suggest opening a new thread about this particular request.

Best,
Shile Wen 

Hi, after some browsing on algorithms/videos on QC I've noticed that a ticker needs to be defined at the beginning of an algorithm. Is there a way to loop through the entire universe of available stocks? Something like:

for every_stock in SP500:

       fetch and compute data

SP500.rank('custom factor')

Hi Big Mak,

Tickers don't need to be defined at the start, I suggest using Universe Selection to create a Universe of stocks that we can filter on. Please see these docs for reference.

Furthermore that statistic is ~1600 stocks at a given time, and QC/Lean can handle 2000 stocks chosen on a monthly basis.

Best,
Shile Wen

Hi Alyson,

I suggest checking our our BootCamp and Udemy course to get familiar with our API. Furthermore, we provide documentation on migrating code from Quantopian to QuantConnect here.

Best,
Shile Wen