Quantopian Live Shut Down

Jared Broad are there plans to add talib to the python libraries?

Lol, QC is about to take off.

Also a Quantopian transplant.

I just find it incredible that they'd up and do away with live-trading out of nowhere like this. There's so much talk over there of 'community' and 'research' and (...), but I think the truth is that they're giving up on what they set out to accomplish - which was to build a place where the open exchange of knowledge and ideas could allow your average investor to join the algorithmic trading revolution, so that it was no longer limited to those with hedge-fund type bankrolls.

It's hard for me not to think that this is more of a "take the algos/money and run" move on their part. So bizarre and disappointing!

Oh well, I'm glad QC and IBuildPy are available. If there's anyone with experience porting Quantopian algos to QC, don't hesitate to post. :)

Hi Kern.

About the porting, I am familiar with both platforms, but all of my development experience on Quantconnect is with C#. I've been successfully porting my better strategies to Quantconnect.

Hi Jonathan Gomez QuantConnect has talib! Feel free to try it.

As soon to be ex-quantopian live trader, I already feel welcome here, I truly wonder if Quantconnect is going to integrate Robin Hood too, that would make a lot of things easier, since I manage 2 portfolios on RH and moving all to other trader would be complicated.

Hi Jared,

Thanks for the reply.

I would like to say, sinec the QuantOpian just have the Python platform, I think the QuantOpian user are also favorite on Python and prefer looking for altenative of Python platform. If you could set more man power on improving your Python platform, maybe you could win many of the QuantOpian user.

I am a programmer and I know C/C++, Java etc. But turely to say, I find Python maybe the most suitable language for handling financial datas and Python has a lot of built dunctions like talib. If I am wrong, please correct me.

I just try to test the QuantConnect Research with link of Jared. But it takes very long and at the end I got error that the web side is not reachable. I am not sure if this has something to do with the fire wall in my company. But I have no problem by using the Resaech by QuantOpian and other platform.

Brokerage integrations are hard but we can do it if we can get some help. First thing is as a company we need official permission from RobinHood. You as RH account holders would be the best people to ask them for this -- we've already sent in an application for API access a number of times.

Jared,

I went over pretty much all the documentation and most of the tutorials. Overall, the capabilities are really rather impressive. I've started playing around with some algorithms and the only frustration so far is the lack of an online debugger. I ended up cloning the project and debugging in Visual Studio, but online debugging--especially considering the lack of offline data--certainly would be a welcome addition to the genuinely great work you and your team have done.

Taking you up on the offer to give some pointers, I'm, again, trying to somewhat mimic the pipeline capabilities from Quantopian. One of the algorithms I'm looking at is a day-trading algorithm. I know having a huge universe to track every day would be very computationally expensive. However, what I'm trying to do is down-select a handful of stocks each day to track for the duration of the day. Say, for example, I am looking to start with the 500 top dollar-volume stocks and, at market open, select only 10 to track for the rest of the day--e.g. based on gap at market open. Can you point me in the direction of what would be an efficient way of doing so?

Many thanks.

Thanks for the kind words Sari Louis 

Re Debugger -- loud and clear -- we're thinking of the best way to solve this.

Coarse universe selection is done pre-market open like you've explained; but the only difference is there's (currently) no way to track 10 after the market opens; so you would need to do that scanning after the market is open. Like this psuedo-code below,

# untested, something like this *should* work.

# initialize
self.AddUniverse(self.Universe.DollarVolume.Top(500));

self.Schedule.On(self.DateRules.EveryDay("SPY"),
self.TimeRules.AfterMarketOpen(self.spy, 1),
Action(self.gappers))

self.Schedule.On(self.DateRules.EveryDay("SPY"),
self.TimeRules.BeforeMarketClose(self.spy, 1),
Action(self.saveclose))

# ondata
def OnData(self, slice):
# save off slice.
self.__bars = slice.Bars;

# open minute gap handling
def gappers(self):
self.__gaps = { find gaps from bars -- difference between __close and __bars }
# find gap stock
for stock in gaps
# do something.

# save close prices EOD.
def saveclose(self):
self.__close = __bars;

I second the debugger. Been trying to translate quantopian into Quantconnect and keep getting errors. Problem is without a helper to show me what line gives a problem I am sort of lost where to fix it. Making the debugging process easier would probably help a lot of people as far as converting their algos. There's a lot of subtle differences.

Thanks for the reply, Jared Broad. That's pretty much what I'm doing, but the backtest is timing out because it keeps tracking and pumping data into all 500 stocks--I think. Maybe I can manage the tracked securities manually. I'll look into that now, but if you have other pointers with respect to the performance for something like that, please let me know.

P.S: I'm using C#, not Python, if that matters.

Understood, if possible please make an example algorithm which should be working and share it with us in support. We'll dig into any slowness and either increase timeouts or fix the root cause.

Thanks again for the reply, Jared Broad. I created the algo and I tried using the "Support" tab, but it said I can't send an email to support without upgrading--happy to do that, obviously, once I know I can get the algo to work. So I shared it with you. Not sure if that's what you meant or if there's someone else I should be sharing it with. Thanks.

Regarding Robinhood integration:

If anyone reading this would like and easy form letter to use in asking Robinhood to work with QuantConnect, please feel free to drop the following in a support request at https://support.robinhood.com/hc/en-us/requests/new

Just deployed to QC live an 80% recreation of one of my live Quantopian algorithms.

Quantopian tended to have nonstop "Infrastructure problems" disconnecting your algorithm every other day. The last time I tried QC live, I couldn't even connect to IB for the first week due to a QC infrastructure bug. This time the process was much smoother without any issues. Here's to hoping successful and stable live trading on QuantConnect :).

:) Hope we can do better! 

Interactive API servers have had alot of issues the last few months -- I understand why QP shutdown live trading -- it takes a lot of time and energy to maintain, And when their API goes down there's only so much we can do too -- and so to be safe we disconnect the algorithm after 15 minutes. 

We are going to make a work around for this though as some people are not using premarket data etc. and so would rather it attempt reconnections right up until 930am than be forced to login manually every day.

If the algorithm does disconnect we've had some luck asking IB to move your account to a different API server. Sometimes they are pinned to a buggy server which is rebooting outside of IB's assigned times.

I just sent my letter to RH support but they are notoriously slow so we will see if they can get their act together before sept.

I got 27k on RH that I would really rather not move to IB. My algo trades frequently and this would be a huge loss of profiablility. Jared, you should consider using RH's unofficial API as a workaround until RH gets it's act together, it's not like they restrict it, it's out in the open for anyone to use.

Hi Jared Broad (and, I'm staritng to gather, Alexandre Catarino :)),

Still having trouble with the backtest timing out per the message above. I had shared the algorithm with you, but also posted it here:

Any help would be appreciated.