QuantConnect Community

Welcome Serge d'Adesky !

We don't support investor IB accounts at the moment. We'll fast track the feature to get it done in the next 2 weeks -- once its done it'll need a server for every time you "login". If IB does the trade mirroring in their backend you could probably achieve it with a single server; otherwise you'd need a new server for every client; the cheapest of which is $10/mo. If there are more than 50 contact us and we can arrange some special/dedicated hosting for you.

Jared,

Thanks for your response. Let me see if I understand this correctly:

1) It seems that  master accounts are currently not supported. You write that it could be ready in as little as 2 weeks. Does that mean you are already working to implement this or that you could if I - and perhaps a few others - showed enough interest?

2) Once implemented, if IB does the trade mirroring,( which I believe they do) , the limitations on the number of servers required would only  be constrained by the virtual server's memory/cpu, am I correct? In other words, with my master account, I currently send an order using "grouped profiles", where I send one order say for 732 shares of aapl and IB apportions that either proportionally according to preset rules for that "basket" into the following:

U939333   BUY 500 SHARES OF AAPL STOCK LIMIT 123.00

U922444   BUY 10 SHARES OFF AAPL STOCK LIMIT 123.00

U922943   BUY 222 SHARES OF AAPL STOCK LIMIT 123.00

Could I do that with just the one server?  

Right now, I only have about 10 customers trading algorthmically, but I would like to expand on that, and eventually grow it to a few 100.  

I would love to be a part of the QuantConnect community to share in the general knowledge and grow the toolset that can be used to address trading universe. But I need to control my costs. Some of my account are large but others are quite small = with a few holding as little at as $10,000. These are set up on algo's that will not trigger the pattern trading restrictions of too frequent trading and are targeted to return a safe 6-12% a year. That's only $600/$1200 a year in gross profit from which my fees (around $150 / year are deducted) . If I need to charge each cient an additional $10 x 12 = $120 , reducing my client's net return to 3% to 9%. I would like to be able to continue and in fact expand on my customer base of small customers as I think this is an underserved part of the market. 

My alternative of course is to use python with TWS api  and my existing cloud servers. Yes I'll have a lot more bellyaches first establishing the headless tws/python configuration, but I'll be able to run all my users for essentially the monthly cost of my cloud server (around $75-100 / month).  Equally important in my considerations, I am comfortable the rug cannot be pulled out from under my feet from one day to the next as I've just experienced with Quantopian. 

Obviously, you guys need to make a profit. In fact, I would feel much more comfortable knowing that your business model is quite profitable as that ensures that you will be in business for many years to come.  But have you considered - perhaps in your "institutional " package - offering an "unlimited" user accounts but with specified bandwidth and cpu charges that increase proportionately to usage?  My algo's use compartively small memory footprints to come to a buy/sell deciscion - and then send  50-odd identical orders to IB 12 to 16 times per week. 

I look forward to your responses. I hope you are not going to crazy with this Quantopian business but I'm sure Q's loss will be QC's gain.  I'm very admiring of what you have accomplished since your start and wish you the best in the future.

Hey Serge d'Adesky. Yes we're going to add advisor accounts due to the demand.

If IB is doing the mirroring on the backend; then as mentioned you should be able to do it with 1 server. You will likely need to pay a professional data subscription to the exchanges ($90 ish/mo total I think?).

We do make LEAN easily runnable on your own server so you are not locked in and the "rug can't be pulled out from under you". However in setting up your own server, paying for datafeeds + exchange licensing you'll soon see QC is incredibly cheap & only possible because we're pooling resources of thousands of quants together and getting brokerage sponsorship.

Jared,

Hi. Those responses are frankly just what I wanted to hear. Yes you are correct about the cost of data feeds. One can choose to get some and not others (for example I've chosen NASDAQ only rather than all the major exchanges - since I trade primarily nasdaq stocks and can get by the minute data from other subscriptins like stockcharts.com). That's one of the reasons I have to be parsimonious in my other costs. $100 here and $75 there, and it soon starts to add up to real money.

I expect with data usage and server fees with QC I will eventually be paying $100 to $200 a month, but that is because there are lots of clients being served by those accounts, thereby making me a profit. I also swore after the Q fiasco I would not ever be so dependent on one provider again. Of course I had always retained copies of all my code and versions, but the inability to quickly run them elsewhere is a big problem. So I will be taking you up on the suggestion of running a backup copy on my own server. They will not need to be hooked up to a lot of data feeds, since my algos don't do a lot of historic number crunching and really trade a half dozen etf's. 

In all likelihood you've gained a new permanent client and I'll be setting this up this week.Thanks for your prompt responses.

Happy we could help :) If need a hand setting up external server let us know.

Started a thread to directly address this here:

Serge, Jared,

I am in the same boat. Trying to link one algorithms to multiple client accounts in IB. Could you walk me through the set up you used on QC and IB to do that?

Any help will be much appreciated

Hi Sudhir,

Financial advisor support was added in 2018; you can see an example of how to do it here; and some documentation on it here. You will need an IB account starting with "FA".

Largely it's very simple and identical to all other algorithms on QuantConnect. You will need a professional account for this feature.

Thanks for the response Jared. Emilio did update me about this option as well. Not sure it does what I want to do. For eg. I have two client portfolios say both with 40% allocated to SPY and 60% allocated to TLT. I now need to change the allocation for both the accounts to 30% to SPY, 60% to TLT and 10% to GLD. The account groups and allocation functionality that the API's help provide to access to do not seem to provide the ability to rebalance based on percentages? Right?? Or am I missing something basic in my understanding here....

Spoke to IB. They suggested using their model portfolios; but the API's don't seem to provide the ability to modify a model portfolio...

Hi Sudhir,

Our documentation shows that we can rebalance instituational accounts based on percentages by assigning both:

• "PctChange" to self.DefaultOrderProperties.FaMethod
• The desired percentage to self.DefaultOrderProperties.FaPercentage

Best,
Derek Melchin