I 100% agree with this statement. See my journal here to see me "eating my own dog food" with my live trading account. https://www.elitetrader.com/et/threads/adventures-in-automation.329052/
Actually the reason I said what I said was because exactly the person is going to come back with this. It's not worth millions yet. If you do this for two years, then it's worth more than 10 million. Sorry @fan27, no way around the hard work you need to do
Though, I have to say, it's probably worth ~2 million. In order for anyone with resources to reproduce it, the cost will be at least 2 million.
Well, the hedge fund I spoke with yesterday who has over 50 billion under management is potentially willing to license IP via backtest results alone. We are talking about two things here. One is research automation, which my product offers, and two, strategies I am currently trading generated via my product based on how I am using my product. Obviously, the latter is useful to market the former. I am in the process of creating very detailed videos of what my product does and how to use it. It should be very clear to users if the product will offer value or not.
Be careful. Big companies are experts in stealing your IP before you sell it to them. Lots of stories of Apple doing it. I reviewed your video earlier and your architecture was pretty straightforward to understand. I could reproduce 80% of it, and spend maybe 6 months getting the rest of it if I were motivated to do so. If it's TwoSigma, I'd be even more careful since that is their model though they are a bunch of nerds and might be on the up and up. Edit: Ah, I understand now. You've got them willing to license your source code. Not buy it outright. For what it's worth, I've sold to TwoSigma before and they are on the up and up, but they will drop you in a second if someone internal creates a competing project (then they come back in two years as happened to me!) With your product, I'd be afraid of someone just copying the ideas which is what is really important here IMO.
What I have is not ground breaking technology, though I think it is tuned rather nicely for algorithmic trading. I worked for an SDK company for thirteen years and the decision for our potential customers was buy vs build. The trick is to have your price point attractive enough for the potential customer to choose buy but not too low. Let's say you could reproduce what I have in six months...what is six months of your time worth?
If you've got a ATM that spits out free money, think bigger. In the end, you've got an embarrassment of riches in where to go, there's no way to know the optimal path so just do what you want to do but know your deal breakers. For buy vs build, you also have to look at the core competency of your customer. If you're selling them something core to their business, they're eventually going to buy you out or replace you. Will they build their own version of Windows? Not for common desktop use, but I've seen custom OSes used before in server rooms. If it's something that is core to their business, it is not just possible, but likely. What I sell to TS is not even close to core though they use it every day, yet some nerd decided he wanted to stop using it because I didn't want to implement their pet feature and convinced the company to go his way for two years until he got bored and they came back to me. This was after a few years of 5 figure licensing. The buy vs build decision in my case is not even a no-brainer, it's just obvious. But even then, they wanted to replace it... At the hedge fund I worked at, they used a third party OMS and one project I was consulting on was to replace it. The thing about replacing a OMS is you don't need all the fancy functionality, only the core stuff. And that's what I did in something like two months. Now to be fair, they continued to use the OMS in conjunction with my project but their days are numbered. Multi-million dollar licensing going to zero, if it hasn't already (I think it has, not sure.) So it's not strictly buy vs build. Optical scanning software (I think that was your company) is a commodity. Auto-trading platforms are also a commodity and I wouldn't be surprised if you're the first one in a new class of soon-to-be-commoditized platforms.
One model that I have been asked to follow, by the way, is a co-development model. That is where they pay me for the code, they can develop on it as well and I can pull any of their features for free. But they are not willing to pay enough to make that possible. This may work for you.
PS: Don't solve the problems I'm telling you about until you come across them. I'm just telling you my experience.
Good points. I think I am going to start with a yearly subscription model instead of a perpetual licensing model. I would rather have fewer higher dollar customers and have more time to develop the tech for my own trading which is primarily what I built the product for.