Patenting a trading methodology

[mattown]

Hey everywhere,

I posted here a looong while back and it's been a crazy year. Still haven't decided whether to pursue a FE degree but I did manage to secure a job all the while with a financial data startup in San Fran...


Anyways I have a really important question for everyone out there, is it possible to patent a new trading methodolgoy strategy? I've created my own quantitative strategy (at least I think it is original, may not be) that basically involves 'favorable situation' matrix. I don't want to give out the details of it but I think it could be highly profitable after the 4 years of back testing that I've done on it. I've been reading Ernie Chan's book about how to start your own quantitative trading business step by step and I've decided to go all in and create a fund based on the strategy.

The strategy utilizes only price history and a reuters knowledge direct data feed. Is it possible to patent or create a trade secret out of this information if the way I manipulate it is unique?

Please advise anyone...

 

[bluechimp]

Well, I'm no expert here, but have you considered not to patent it? I'd guess that in the case of patents, there will come a day when you have to make your intellectual property a public knowledge, at which time you'd lose any kind of competitive advantage you were trying to protect. Firms like Coca-Cola actually never patented their recipe for the drink, and opted to keep it a company secret. I have actually never heard of any fund patenting their trading strategy.

What were you trying to achieve by patenting it?

 

[mattown]

I pretty much just wanted to patent the formula and methodology because I'm paranoid someone will steal it and if that were to happen I'd want to take legal recourse.

I think it also may give me more bargaining chips in in the event I want to work for someone else and implement the strategy I suppose. ..?




​

 

[Andy Nguyen]

Have you verified that your trading methodology is new or hasn't been patented?
It's possible that people are already making based on this methodology so I don't think I would bother with patenting this. If you are so sure about the success of this, you should be busy making money off it. Once it no longer profits, then you come up with something else. My bet is that the usefulness of a patent would out last the profitability of your methodology.
The safest place to keep an idea is in your head.

 

[mattown]

I haven't found anything resembling it yet...but it's only been a week, and there is a vast sea of legal crap and ideas out there as far as I'm concerned...

I just want to weigh what my options are. I'm going to go through with the strategy anyways, but if there are ways I can protect myself then I'll at least explore them.

 

[financeguy]

you won't be able to protect it. even if you do get a patent, someone can do something just ever so slightly different, perhaps even improve upon it, and you won't be able to do anything about it. just go trade on it and don't tell anyone your exact methodology. you're wasting time not getting started on something that actually works.

 

[mattown]

Well I've given up trying to patent this damn thing and I'm just going to start the strategy regardless if anyone has done it. I have a feeling my strategy in many ways isn't unique.

It's actually a pretty simple strategy based on only a reuters fundamental data feed, a big 'ol custom matrix and machine learning algorithms. I don't even know how the machine learning algorithms truly work but they get the results that I need (weights!). I've also proven that the method is bull/bear market independent with the 15 year back test I've done.

I managed to convince enough people (family) to invest an initial amount (think around a quarter million) in the stock program for the NYSE and NASDAQ and I hope to set it up by the end of the summer. If it works I'm just going to continue trading for myself and my family and get the same program set up on exchanges around the world.

I'm also going to start a blog that tracks my progress regarding the fund, I feel it'll be a great way to meet people since over the last couple years it has been difficult to find anyone who takes me seriously regarding this trading work (applied to many jobs, never got any replies, ever). I eventually want to trade for someone else with more capital as I feel there is a capital limit to this strategy but I have faith in my ability to scale it. Not because I'm intelligent per se, but I have faith in my ability to be creative and blend with people who are intelligent and interested in the same work.


I feel with my background this is probably the best way to get my quant career started. By background I mean being 25 with some financial datafeed work experience coming from a relativity weak astrophysics student at a "pretty good but not spectacular" university.

I will eventually want to pursue a MFE for the extra knowledge after I get this set up and hopefully at least prove this time I have a track record of some sort for the MFE admissions people.

Anyways thanks for your input you guys and I'll apprise everyone of my progress regarding the blog and quant business.

 

[Andy Nguyen]

I do hope that you keep Quantnet updated of this project. You can add the link to your blog/site in your signature.
A bit off topic but have you thought about the tax implication? Would you claim any gain/loss on personal tax or under any business entity?

 

[mattown]

Oh I've definitely thought about the tax implications, for sure I'm going to form it under an entity, either an S corp or general llc. The gov can go to hell when it comes to me paying more for their mistakes as far as I'm concerned.

I like S corp because you pay less depending where you form it and with trading that can be anywhere (neveda, delaware , i.e. good places) vs your traditional bricks and motor.

I've found this article in the legal code concerning living in California helpful. Basically to the best of my knowledge it shows if you are just managing external bank accounts and nothing else (which I would be from San Francisco, CA ; ) ) then you won't get dinged from California's broke a$$ when it comes to S corps, which is something like a couple %'s since you can still reside in california and not be a 'intra-state' business, i.e. one subject to california tax regardless..

WAIS Document Retrieval

see (ap) - (2)

The trick really will be setting it up with foreign exchanges, I don't even know where to start for those but hopefully if it does well here in north america first then I can pay someone else to figure it out.

 

[alain]

Before you jump in front of the train, have you paper trade your strategy with life data considering taxes and transaction costs? That's the last step.

Also, you said you backtested the strategy with 15 year worth of data. Is that out-of-sample data? how often do you rebalance your portfolio? what's the drawdown of you strategy? sharpe ratio? which benchmark are you going to use? What is the performance of this benchmark in comparison with a regular buy and hold?

15 years going back seems like a long time so make sure you don't survivorship bias in your tests.

 

[mattown]

Yes it is with real world data.

There are some suvivorship bias issues that exist but I did my best to clean up the data. The portfolio gets rebalanced based on what orders go through and how the market behaves through stop losses and limit orders that range between 0 to 5 days.

I didn't run it on intraday data but I based it solely on end of day data that was available (open, high, low, last). I've based my strategy 'around' this information.

The transaction costs were factored in, taxes I'm not worried about going to let an accountant handle it.

The sharpe ratio for the trailing year is 15 point something, based on the SP 500 for the volatility and 5% as the risk free rate.

Basically my plan is this, learn how to implement my strategy with c++ and an api like interactive brokers. Then 'test' run the strategy for a month, the results are similiar to what I expect then I'm going to switch the model on. I don't have c++ experience yet and built the model with Oracle but it is simple enough that I think I can learn the api/c++ in a month or two.

The model itself also has a kill switch for instances like last october. :D

 

[alain]

Well 15 sharpe ratio is impressive so good luck!!! post your progress here.

 

[Andy Nguyen]

I also hope that you checked with what others have done and discussed in term of paper trading, setting up system, back testing over at forums like elitetrader, traderslaboratory

 

[Subramaniam]

Coming back to the original question on patenting: As bluechimp mentioned, Have you considered "trade secret" protection as opposed to "patenting". Trade secret is less powerful than patenting however it seems more approriate in your case.
You can read more on wikipedia: http://en.wikipedia.org/wiki/Trade_secrets

 

[mattown]

Yeah I'm familiar with trade secrets, that's how Coca-cola is so awesome

I'm going to just go forth with it, if it makes substantial money I'll hire attorneys. I figure I just need to take a 'one-step-at-a-time' approach to this. What good is a trade secret if it doesn't work is what I'm thinking?

I should have the blog up by next week, I want to dedicate it to those who want to do this on their own. Who knows maybe this time I might actually have someone interested in hiring me? heh.

 

[Adam G]

There is no point in patenting or getting a trademark on an algorithm. As other people said, once it is in the wild, people will copy the idea and find some way to get around the patent by changing it slightly. My advice is to get a non-disclosure agreement and have everyone that you pitch the idea to sign the NDA. This is your only real legal recourse. You can get drafted NDA's online for fairly cheap.

 

[sak]

Learn from GS and CAT

Is it not that people patent only systems which can be reverse engineered.

GS never patented the "process" which just recently was stolen/compromised. And I can see the reason. The code/idea changes every couple of days. Patent would just expose entire idea to the planet and it would no longer be novel. People would one way or the other figure out to avoid or circumvent it, and still use it. Remember CAT trying to patent the yellow color. I think they were successful but still people just change the composition to a slight amount, and get the same yellow color without violating the patent.