IP Protection and the ongoing sustainability of the business model
-
I wish this post can be a starting point for debating about some fundamental aspects of a business area like the one in which Quantiacs operates.
-
How do you protect the IP of a strategy's author? How can a quant who spent hours and hours developing a good strategy be sure that you won't look at its code?
-
Is the 10% on the net profits just for 1 year? What happens if you use more than 1 year?
I fear that issues like the ones above(expecially the IP related one) could prevent top notch authors from participating in the competition.
I also have the feat that the business could not be sustainable in the long term. Quantopian, which seemed to have a business model really similar to Quantiacs, has closed down and from rumors, opinions I gathered here and there I.P. protection was the main issue for talented authors.
I think you will have to give ground on this front, maybe by accepting somehow signals generated by cloud systems set up by authors or other systems that can 100% safeguard the I.P.
And the fact that you end paying royalties after one year is also crucial. If I designed a very good, long term strategy why shouldn't I get rewarded long time?
-
-
@magenta-grimer Hello, these are two very good questions.
Let me start from the second. The competition model is a starting point for quants, an incentive to develop and send code. The reward in terms of prize money can be substantial, as if you win you get 1 M USD in allocations and 10% of the profits without any downside risk for you. The payouts are equivalent to an investment of 100k USD, keeping all profits for you, but with all downside risk on your side.
However, nothing prevents your system to get funded for a longer time, even if your system is not among the winners of the contest. The contest is indeed using as a criterion to define the winners only the live period (4 months), which is quite short. In the past we had systems which did not perform well enough for 4 months to get a prize but were (and are still) making great in the long run.
For all these systems we have individual agreements in place with the quants who submitted.
As we write in the general agreement: https://quantiacs.com/termsofuse
at point 3, two models are in place: payouts according to Net New Profits, or a Monthly Management Fee.
So, to answer your question: "If I designed a very good, long term strategy why shouldn't I get rewarded long time?"
In this case you will get rewarded. However, we should work and setup a set of systematic rules to make more clear the criteria we use to decide if a system should get funded or not in the long term beyond the competition boundaries.
-
@magenta-grimer Concerning the first point about IP, there are 3 aspects:
-
in the terms of use: https://quantiacs.com/termsofuse it is stated that the IP of the submitted Private Content stays with the quant (point 10).
-
how we protect your IP:
a) on the web level our developers are trying to understand the potential vulnerabilities in our system and run penetration testings;
b) we log all internal activities, use separation of roles and access privileges inside the company;
c) the servers we use are located in facilities with early warning fire systems, special door locking systems, video-monitored high-security perimeter fencing, entry via electronic access control terminals and modern surveillance cameras for 24/7 monitoring of access routes;
d) our company policy forbids all employees and contractors of Quantiacs to participate in the Quantiacs Contest.
- submitting signals only and not code is certainly the strongest way to be sure that only the quant who developed the code has access to the source code, but it has two evident drawbacks.
-
It is very impractical, as the quant has to send daily the positions. For backtesting this is not a problem, but for the live paper trading, and later for the live trading in reality it can quickly become a nightmare: what should we do if the quant stops sending the positions? Anybody who worked in the industry knows that real trading poses several additional complications respect to backtesting, especially as we are interested in strategies with a large capacity. What should we do if an investor puts 50M USD and the quant stops sending his signals?
-
How can we run stress tests on the strategy, for example running it on artificial data sets to see how the strategy reacts to particular market conditions? Obviously the biggest problem we are facing is overfitting of backtest results, as quants have access to all past data, so we need to run several tests on the code to decide about long-term perspectives (see also previous answer)
-
-
Great reply @support ! It really addresses many of my concerns in a good way.
Thanks
Mine's is just an idea, but maybe for IP protection one can think about putting in charge a third entity, such as Amazon or Google or similar firms with cloud activities and upload the source code there, such an "escrow" system....
-
@magenta-grimer Thank you, we will think about it and see if it is doable
-
@magenta-grimer Another point: you can develop your strategies locally using Quantiacs software and data, see documentation:
https://quantiacs.com/documentation/en/user_guide/local_development.html
In this case you will not share any material.
-
@support Ok, but then to submit to the competition I have to send it to you. For the live evaluation phase it's always that you have the code at your own disposal
-
@magenta-grimer That is correct, but you can for example train a model locally on your machine, and then submit a code which uses the results of the training (for example stored in numerical form) but does not reveal the underlying idea.
