Off the shelf software vs a mix of open source?

Discussion in 'Automated Trading' started by ChainsawDR, Aug 18, 2019.

  1. I can't help you, but I can empathize with your situation. Is 18 months your total exposure to the markets in general? If so, I can tell you that 18 months isn't that long a time.

    I hired my first team of programmers 10 years ago and was thrilled with the results. A very fast and top notch delivery from a start-up company who wanted to prove themselves. Like you, I reasoned that my time was better spent working and focusing on the markets and having programmers take care of the coding since I'm not a programmer.

    After that first delivery and subsequent deliveries - I had a less than stellar experience with a new programmer on that team. In hindsight, I should have switched to someone else a lot sooner. A lot of time spent debugging, trying to communicate what I want, waiting for deliveries, find errors, comunicate, rinse and repeat. Then, I had a new idea and had to wait to get that implemented/tested. Then, maybe I discovered that this new idea wasn't that useful after all. So, a lot of money AND TIME was spent developing my stuff.

    I don't know how much time it takes to become a proficient coder who can do this stuff on his own, but in hindsight, maybe that's what I should have done. Not sure to be honest. The point is that being reliant on someone else is a very frustrating situation to be with.

    In the end, I've spent a lot of time and money developing something I'm now satisfied with, although it still have a few things which are lacking (mostly related to speed/performance). I'll deal with that in version 10.0. If I were to do it all again, it's not unlikely that it would have been a better choice to at least learn the basics of programming and also be more critical in terms of who I hire (might be better to pay up once instead of going with someone who's cheaper).
     
    #11     Aug 21, 2019
  2. CharlesS

    CharlesS

    Would you be able to trade your algos manually, if you had the time available during market hours ?

    If not, is it due to algos with mathematical calculations that just can't be done mentally, or due to difficulty of executing multi-condition algos in short time frames, or both ?

    I'm having both of these problems, have a sfwr dev background, but don't want to delve into a coding project at the moment, though I expect to.
     
    Last edited: Aug 21, 2019
    #12     Aug 21, 2019
  3. Thanks for sharing laissez faire. Agree that paying more upfront might be the better path (I went for a cheaper option that promised 12 days for a first milestone and it’s only now being delivered after 50 days - lesson learned!). For clarity, I’m definitely going to invest in learning the skills to code algo rules myself and won’t outsource that, I’ve just been trying to avoid some of the infrastructure work (an example would be, if I place a market order for X stocks in Y symbols, I don’t want to code the logic that listens for the confirmation from the broker API or handles exception messages. I’m hoping there’s either software that does this already or an SDE can own this).

    Hi Charles, no I wouldn’t be able to trade the rules manually. It’s a bit of both. There’s a lot of mathematical rules for calculating the entry and exit, with the exit calculations being continually updated based on how each stock and the wider market is behaving (with stops handled outside of the market rather than leaving them on the exchange to be swept up by market makers). I’m aiming to spread the portfolio investments over 20 positions (5% each) so staying on top of so many (and the moving exits) would be too difficult to do manually.

    From the feedback on the thread it seems like the best option would be to invest upfront in a good developer who can setup the infrastructure while I invest time in learning the coding skills to research and apply the core algorithm (avoiding the cost of exploring ideas mentioned above -if ideas don’t pan out at least the skills learned to explore them wouldn’t be wasted). Laissez faire I presume by the fact you’re on v9 that the approach of outsourcing the infrastructure has ultimately panned out for you (otherwise you wouldn’t have iterated it this far)?

    Thanks again for the assistance (don’t have any friends in the algo trading space to talk things through so this has been really really helpful)
     
    #13     Aug 21, 2019
  4. I guess. But if I were to do it again - I would have paid more initially and paired up with someone who really knows what they're doing. Easier said than done, I think. Also, maybe I'm paranoid, but I'm a bit wary of sharing my stuff with other people - even programmers. Fortunately for me, most people I've worked with are lost on my ideas anyway and don't really understand markets. The ideal would probably be a collaboration with someone who knows both trading and the markets.

    Anyway, seeing that you're working 60 hours per week, I wonder what progress you realistically can be able to make anyway? Unless you're able to work on this during your working hours?

    In the end, programming is a tool, but it's no substitute for market knowledge. Understanding how the market works and moves is a long process. No way around it, I think...
     
    #14     Aug 21, 2019