I want some understanding regarding ownership of IPs in a pod structure. If I bring “x” systematic strategies when I join an HF and develop “y” while working there would x be treated differently than y while leaving the fund? When a PM joins another MM how it works on the IP side? I want some general understanding and know it will depend a lot on how a contract is written.
If someone is not a senior PM then is it easy to negotiate on IP before joining an HF?
| Best Public Markets Gig? ER, MM HF, SM HF, or LO AM? | 13 | 1d | ||
| +42 | How to start a family office (HF rainmaker) | 17 | 12h | |
| +33 | Law to Quant Pivot? | 10 | 1d | |
| +20 | How to destress at Pod HF? | 7 | 8h | |
| +19 | Hedge Fund WLB | 6 | 10h | |
| +19 | HF Final Round Interview | 4 | 3d | |
| +18 | People who work at hedge funds, what made you interested initially? | 13 | 7s | |
| +17 | Most appealing public credit seats out of RX? | 1 | 4d | |
| +16 | good fundamentals, bad chart | 8 | 1d | |
| +16 | Which Funds focus on timing the broad market rather than L/S individual equities? | 10 | 38m |
Career Resources
Get instant access to lessons taught by experienced private equity pros and bulge bracket investment bankers including financial statement modeling, DCF, M&A, LBO, Comps and Excel Modeling.
What do you mean by IP? code or idea?
Code
Most MM do not let you keep your IP. I think best case scenario is that you share IP ownership with the firm, meaning that technically you can take the code to another shop. Firm still retains all rights to the IP even in this case.
In general, quant IP ownership is a little overblown. No reasonable fund will blindly run a piece of software without the original owner or someone who is intimately familiar with the code base, no matter how well it performed previously. If you end up leaving the most likely scenario is that your code will end up untouched on some server and the only people looking at it will be new college grads who are given it as grunt work to try to understand.
Thanks. Got the idea.Probably no point in negotiating much over it and focus on other aspects.
i've never seen actual code allowed to leave...in general...you just go rewrite your code at the new shop (if you wrote it once...you can do it again, usually easier and faster the 2nd time)...and for this reason, people generally keep a copy of their code on a home computer...and often encrypt some portion at the fund so that it only runs with a daily or weekly password
Thanks. It is a better idea to build on my home computer and always keep a copy.
Please read your contract before doing anything and make sure you aren’t violating your agreement. What is being described above would definitely violate agreements I’ve seen.
You should not do anything with your personal computer (and may get sued for that), but there is nothing stopping you from making the code you write on the company servers obscure and hard for someone else to understand.
A few MM PMs out there will keep running code after firing the quants who wrote it. They won't do it blindly but will make sure to understand the alpha and market phenomenon it's capturing.
What is the best way to obscure your code?
there are lots of ways...using obscure variabel names (i, j, k etc...) and using multi dimensional arrays imagine an array with 5 dimensions for storing all sorts of stuff, use lots of sub functions and object inheritance even when not necessary...use a database for storing temp data which could easily be stored in the code memory (in this case, theoretically a different process could also be interacting with the database) so a new developer would be scared to touch /. change it....and indeed..perhaps there is another process that interacts with the DB...and that process is encrypted...And it corrupts the database if a password is not entered into some other system every 4 weeks or whatever...(so if you leave there is a kindof "self-destruct")
essentially...all those "best practices" you learn in programming class.....don't use them if you're the only person you want looking at /touching your code
One-letter variable names are easy to do. You can have parameter values that are "unlocked" only with a manual password that acts as a checksum, which is how cheat codes in video games worked back in the day. File encryption tools may work if you can download them. I've seen PM themselves use compiled native code (without giving analysts the source) and weird pointer arithmetic in C. Of course, you should get a sense of what the environment is like before doing any of this. A few PMs are out to screw you over but most are not.
The issue with code obfuscation is that you're more likely to unknowingly introduce a bug, or even worse, make it 10x more difficult to debug the program six months down the road. I've worked with someone in the past who employed similar techniques. His code was also insanely non-performant, and he would routinely pull all-nighters trying to add new features or conduct new research.
Beyond that, a PM should never let you be in a position where you're the only person looking at or touching code. Now, I haven't worked for a MM, so I don't know if this is realistic in that environment, but I have worked for a small, high-performing quant group, and our #1 priority was avoiding "bus factor of 1" situations. All code was developed on Git branches, and every single line of code went through code review before merging it into master. At no point would we have allowed shenanigans like encryption (obviously, we would encrypt database keys and the like) or intentional obfuscation. It's detrimental to the health of the business. Why give any individual researcher that much leverage by having them be the only person who really understands their strategies? I certainly get why you'd want that kind of moat as an individual contributor, but I would never structure a team to allow that kind of situation.
The reason I called out that our team was high-performing is because I didn't feel that mandatory code review negatively impacted our velocity. In fact, it increased our long-term velocity. It made it easier to help out on a project, because everyone coded with roughly the same idioms. It also gave people more conviction in backtest results.
Using DB is a very good idea and didn't think of it. Thanks
you can convert a python script into a standalone .exe executable and so nobody can see the code inside...this should make it pretty easy to add a once-a-month passkey (the date + some checksum digit) so that your code won't run without you being there. If you are the only one touching the code, this seems pretty easy to do.
Presumably this is where you have the guts of your IP model logic...and the external stuff like data cleaning / manipulation can be visible in standard python scripts
U guys r sociopathic geniuses.
Seriously threads like this get me hard and wishing I had dropped business for CS.
Voluptatem voluptatem dolor at tempore molestiae quisquam et. Suscipit ad quae architecto neque. Vitae pariatur corrupti architecto modi id enim nemo molestiae.
Provident quisquam debitis molestiae expedita. Ut aliquam vero beatae et sapiente asperiores et voluptatem. Minus accusamus et reprehenderit molestiae. Illum quod ratione sed ab vitae possimus omnis.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...