›

T&C

st157101st157101 Posts: 3
in Quick Question

Three quick questions before I start:
1) What profit share do you offer (I heard 10% on chat with traders - low relative to industry and of course 1/10 of what I get trading my own capital)?
2) What guarantees do you give that CloudQuant staff won't snoop on our strategies?
3) If you trade our strategies, are they netted for correlated PnL of other strategies you trade?

Thanks!

Comments

  • td753764td753764 Posts: 67
    edited May 2018

    Great Questions.

    1) What profit share do you offer?

    Answer: Each algo license is negotiated with the algo creator. The typical terms for a full working algo is 10%.
    Those that receive higher in the industry are responsible for much more than creating and keeping the algo fresh. They typically are responsible for running the algo (trading), risk managing, reporting, and are often licensed individuals whose full-time job is quantitative trading. Our normal offer of 10% of net profits is rather rich considering the split in responsibilities. An algo creator can (and some do) quite literally do this on the side.

    Consider the scope too. We are allocating millions to trading algos. See our record of allocations. This means that receiving 10% of net profits of a $5M to $20M allocation of trading strategy net profits results in (potentially) nice paychecks without risking anything other than your time and effort. You won't have risked any of your own capital or your day job.

    2) What guarantees do you give that CQ staff won't snoop on our strategies?

    Legal Answer: We lay out our responsibilities and your rights in multiple ways in our user agreements. We also cover this in our license agreements when we negotiate for the use of your algo.

    Technical Answer: None of our portfolio managers have access to any algos except their own. The security of the system is built to allow users to access only algos in their own library. We may in the future add the concept of a "team" or "group" but in doing so we will still ensure that no CQ portfolio manager, or any other user, has the ability to browse the intellectual property of another.

    Business Answer: Our business is built on trust and partnership. If we ever "snooped" someone else's algo we would never get any other business and lose what we already have. It is a fundamental conflict of interest that would damage our reputation beyond the possibility of repair.

    3) Are profits netted with other algos?: no. Trading profits, losses, and costs are tracked at the algo level. This means that if your algo is making a profit and all other algos are losing money you will still be paid. And if all algos are making money but yours is in a period of draw down :( then you won't be receiving a profit share.

  • st157101st157101 Posts: 3

    On the "Business answer" to the second question: it's not a conflict of interest if the business goals change. E.g. Stevie Cohen makes you an offer you can't refuse for all the algos. ahem Quantopian.

    One way for algo developers to mitigate this risk is to develop outside CloudQuant and upload their signals. Do you offer an API for pulling in external user data?

    Also, final question: is CloudQuant open source?

  • td753764td753764 Posts: 67

    Business goals do change. However, the legal agreements don't without counterparties agreeing to the changing terms. We make it clear both legally and in our constant posting that the algo is your intellectual property. No Debate. You can reference a lot of talk about intellectual property with CloudQuant on our main site https://info.cloudquant.com/tag/intellectual-property/

    One Way ... do you offer an API for pulling in data? This is a big question that is broken down into a series of sub-questions which I will try to enumerate below:

    • Do you have an API that will allow reading a file with my PKL or configurations? Yes we do. This is one of the ways that people marry machine learning and the CloudQuant backtest and live trading together to ensure that their trading strategies are working well. The data would have to be uploaded to CloudQuant.
    • Do you have an API that will read from an external source outside of the CQ environment? No. We do not allow access to external systems in live trading. This would be a major security breach and risk breach that we, as a regulated entity, will not allow.
    • Do you offer an API that will allow an external system to send you trade signals from outside your system (i.e. FIX 4.x or 5.x)? We have in past taken signals as part of the algo development and backtest workflow. But we really, really, really don't like this for live trading. There are a number of reasons for this including regulatory, risk management, position management, technology-risk, security, and operations.

    Is CloudQuant open source? We have placed a number of strategies and some other utilities into open source. We are likely to continue placing new useful tools and utilities into the open source realm. Our backtesting engine and the trading systems are not open source projects. These are part of our competitive advantage as a trading firm.

  • st157101st157101 Posts: 3

    Understood on business goals, but my algo being my IP and my algo being secret are two different things. In this business, owning an algo is worth little if it is not secret.

    With respect to pulling in data: regardless of the mechanism, is there any way to upload data to CQ intraday and have it read by an algo?

  • ptunneyptunney Posts: 246
    edited June 2018

    You can manually add a file to the user files and read if from there.
    If you had a successful algo and requested funding then you could get direct access to upload a file.
    Whilst it is understandable to be wary of any service where you are using their resources for free and they are promising you privacy, especially in this day and age, at some point you have to have mutual trust.
    Also, realistically, you cannot expect to run a "blind algo" live, there are compliance requirements, the SEC will often want to know why a particular algo behaved in a particular way.
    If you do produce an algo that is profitable and you want to get it funded and take it live then there will always be a requirement for review of code.
    But until then, rest assured, we are not remotely interested in your code, nor do we honestly have time to look at a users code.

Sign In or Register to comment.