Hacker News Reader

Comment by ukoki | Hacker News Reader

Possibly. If you integrate with a trading platform you could always let other users subscribe to a strategy without knowing what the strategy is (eg they only know it's gained X% vs the S&P 500 in the last few months, the author's profile and that 5000 other users have 5MM invested in this strategy). However I have a feeling that a) the money from commissions and b) the feeling of being at the top of a list will compel a lot of people to share their strategies - people blog about their stock picks constantly. Also I'd love to see some left-field data about these companies eg: Google search volume, Wikipedia edits/day, mentions on Twitter, press releases, sentiment analysis of news articles etc...

View on HN

It's funny that you mention it because, as Rob said, we've been playing with the idea of adding a competitive edge, but the commissions is a cool idea. We were also considering a kind of achievements approach as well as a leaderboard but want to try and keep the idea of it as a tool as opposed to a toy.

Love the data idea, in fact surplus of data is already a bit of an issue; we're running out of screen real-estate.

of course - the alternative is to not share anything. With the data you have on your users, once you can isolate users who consistently come up with strategies that consistently make money after their creation date - well, you might as well set up a hedge fund on the side and give the product away for free.

Hah, yeah, unfortunately the legal team at the incubator we're in suggests we might not survive too long as a company if we go down that route :)

Your data is going to be insanely valuable if you can get good analysts working on your platform.

I was wondering if you could elaborate as to why they feel this way? I would love to know that the top X % of traders are willing to play with my money in a transparent manner.

haha fair enough - thinking about it there's all kinds of nefarious ways you could use the data. Looking at your product a bit more, this might be complicating things a little and it's going to hammer your database but I'd love the ability to fit variables to a strategy - eg I want a strategy like "IF X day moving sample average for GOOG is better than Y day moving sample average for GOOG buy ELSE sell" But I don't really know the best X and Y, so I'd like to leave those as variables and have the software find the optimum values based on historical data.

we've all been salivating about the potential for an 'Optimise' button. It would be really cool, it has a few wrinkles though, e.g. there might not be a unique maximum solution, so how do you choose the best one? Once there are a few variable involved it may be a job for a genetic algo, but it would have to be in the background. All food for thought ;-)

Here's how I'd do it: Don't let the variable be the stock itself because otherwise your search space will be too big... limit the number of stocks you can optimise on and constrict it to variables like "moving average duration". That way you can load the stock data onto the client side and do the processing there (presumably you're already doing this for your graphs anyway) - kick it off with a few generations of pretty loose genetic algorithm then when you get near your time limit switch to simulated annealing with the best handful of results to narrow down the answer as fast as possible. Keep it quick, say 2 seconds tops. Then what you can do is give the client a confidence rating so you can say something like "Found the following result with 89% confidence. Click here to perform a longer search for higher accuracy". Work out your confidence rating based on how bumpy the ride was.

Reply on news.ycombinator.com