Where do I hire a programmer to create or help me set up a system for analysis?

Quote from Stoxtrader:

Once you mention Excel and VBA, any programmer is going to assume that the complexity level is low. Generally speaking, competent programmers try to avoid VBA. If your project is not much work, VBA should be OK. If the project is already half-complete, you might stick with VBA. Otherwise consider agreeing to whatever language the software engineer is comfortable with. C, C++, Java, C#, Python, etc.

Feel free to PM me the specs, or post the details to this thread.
More...

Thank you.

I do think the complexity is low and can be done with VBA from what I can tell, but it may present limitations in the future so it is probably a better idea to choose another platform or language.

The guy I hired had no prior projects on Guru, but I took a leap of faith since he presented a very impressive work plan and even added some ideas I had not thought of. Also, I found his credentials on the internet and he seemed very legitimate in addition to doing it for a very low price. I do not think he in any way tried to con me (I still have not paid for anything), but it was simply too much for him.

I already wrote some details in my prior post, but I may post more in private.

Thanks.
 
I have used elance several times with success.

You can post a description of the project and receive bids from all over the world. The amount you offer to pay for a project is held in escrow by elance and if the work is not satisfactory it will be returned to you. Not satisfactory means the programmer did not complete the spec as written. If he wrote the program and followed your spec and it still doesn't work, that is your problem (bad spec).

The problem you are going to have is to write a description of the project so that a programmer can complete the effort quickly. My guess is you cannot communicate what you want to do; or it will not work even though you think it might. You cannot expect the programmer to do anything more than write to your specification....

If you want to hide some of the details applied to trading then describe the problem as a digital signal processing application and provide enough data so that it does not look like ES 1 minute time series.

At this point you have a 'feeling' your approach will work and you want to hire a programmer to not only program to the spec but make it work even if the spec is bad all for a fixed bid. This is what some would call a 'bad trip'.

Good luck,
 
Quote from Laissez Faire:


...
(A) I have a friend who is taking a PHD in finance and is an aspiring programmer, using both VBA, MATLAB and C#, but I`m not sure if he is able to undertake the task and I would prefer to let someone I know can handle it do it from the start. I will read through all the links you guys sent me. Thanks.
...
(B) I already bought MATLAB on a student license
...
(C) It is a matter of statistical analysis and modeling of intraday price data. Automating the process of collecting daily metrics of the intraday price gyrations, volatility analysis, coding each day according to specific parameters and integrating it in a framework where it is possible to do historical queries and predictions based on this information. I expext that once completed, the model will have predictive value,
...
(D) live price data and use this as input throughout the trading day.
More...

A. I was glad to see point A as there is a huge interest in the financial sector among college students some of whom are quite capable if not yet competent.

A previous poster made an important point that these engagements tend to be ongoing relationships as new ideas beget new ideas rather than home repair projects. There are some preconditions to success:
  1. You need to have the time and interest in day-to-day management. Students need alot of guidance.
  2. You need to be able to judge progress. I have seen this happen to other people both ways: A. Investor pays for years to someone who can't do the job and who isn't passionate about learning, and B. investor has competent talent but keeps postponing delivery to add scope until he loses faith
  3. You need to be able to realistically analyze the return, costs, timeliness, and risks of the project. Even if labor is open-source, it is still a business venture that can fail.
If you don't have these preconditions, consider getting the help of a project manager. At least, nail down #3 so everyone doesn't end-up predictably disappointed.

B. Matlab is a great product. It has some really great advanced parallel computing capabilities, but I would recommend starting with Scilab (free) or one of the other open-source work-alikes until you are through the proof-of-concept phase.

I have seen VBA in Excel used as a front-end to enormous system, but when people see VBA in Excel, there is a Bayesian inference that this is a modest, one-man job. Unfortunately, this often reflects more on the knowledge of the speaker than on the task.

C. EOD statistical modeling of intraday data. Both RiskMetrics and kdb fit your description. Although I do believe that a small team of really smart, focused people can do an amazing job on a hard problem, the tail-end of UI, reports, extensible rules, workflows, and support can easily consume a battalion.

If you have a modest <$10M budget, then I would recommend spending a couple of years sitting with a few students to mutually learn how to do financial programming. This should greatly improve the ROI on the project and help you to better define the scope of the project.

D. Real-time data. The challenges of collecting and processing realtime data has been discussed on several threads. There are two things that I want to say relative to this project:
  1. Given the stage of this project, I would recommend downloading the market data historically using a service such as NANEX since this will allow you to validate your ideas across many symbols without incurring market data fees and with a clear migration path to realtime.
  2. Executing statistical analysis realtime for trading purposes is a completely different challenge from generating EOD market metrics for use in an intraday system. For example, LMAX or StreamBase. If you are building one, don't expect to get the other.
For IT projects (integration with existing software) run at companies where pulling-the-plug has career consequences, the success rate for adhoc projects (costs already budgeted) is 74% for small teams, 58% for medium-sized teams, and 40% for large teams. There is no survey for personally funded projects that have significant technological and mathematical risk being run by micro-teams in ones spare time using dispersed, multitasking, contingent talent.

You wouldn't build a house without blue-prints.
I would recommend breakout the books, with the mutual help of a college student, and treat this more like a hobby than installing flooring. If you really have the money and want to treat this like a start-up, then figure-out your management structure and write-out a business plan. You may have a billion dollar idea, but success comes from realizing a plan.

Steven
 
For Laissez Faire and anyone...

I am very interested in quantitative trading system implementation. I have over 15 years experience in high end systems engineering. Would love to hear from anyone needing assistance.

PM sent.
 
This thread makes me glad I can do my own programming ..... I would not envy the task of hiring a programmer, so many potential pitfalls.
 
If you want something done right, do it yourself....

Having to rely on a programmer to do basic research is a major impediment to trading success.
 
Quote from Lornz:

If you want something done right, do it yourself....

Having to rely on a programmer to do basic research is a major impediment to trading success.
More...

absolutely true. if you cant program how can you do any research. if you cant research how can you trade (besides slowly losing money clicking a mouse).
 
some of us have mechanical systems, once we put it on we know what we are going to do, but don't know how to tell a computer to do it. I looked at C# and it was over my head.

Perhaps a better approach for me would be to simply pay somebody to teach me how to program.
 
Quote from Lornz:

If you want something done right, do it yourself....

Having to rely on a programmer to do basic research is a major impediment to trading success.
More...

At the 13th of May 2011 you were fully automated and did not know what to do with all your spare time.

http://www.elitetrader.com/vb/showt...55&perpage=6&highlight=automated&pagenumber=1

Then, at the end of 2011 you were tired of discretionary day trading and wanted to automate your day trading. You were planning to start learning programming in 2012. You also told me you considered hiring a programmer. How could that be if you were fully automated already?

You also told me you never read any books on trading, only to create a huge list of books that were a must read for any trader at another forum shortly after.

Not to mention that you have another handle at this forum as well.

Please keep your bullshit lies and fantasies off this thread. It seems to me that you have nothing of value to add, except slandering other philosophies and in the process implying how much you know, while the fact is that you know shit.
 
Dear Steven Davis,

Thank you very much for your thorough reply.

What you are suggesting seems to be way beyond the scope of what I had planned for this little project of mine, but I very much appreciate the links and the resources you provided. I will look into it tomorrow and meditate on it.

I will probably contact my PHD friend and consider a collaboration with him. He is a smart guy, but I`m naturally distrustful of people and I fear that he might not pull it off. A bad habit of mine. It would probably be the best solution if he could pull it off though.

Please let me get back to you tomorrow.

Thanks,

Laissez Faire

Quote from Steven.Davis:

A. I was glad to see point A as there is a huge interest in the financial sector among college students some of whom are quite capable if not yet competent.

A previous poster made an important point that these engagements tend to be ongoing relationships as new ideas beget new ideas rather than home repair projects. There are some preconditions to success:
  1. You need to have the time and interest in day-to-day management. Students need alot of guidance.
  2. You need to be able to judge progress. I have seen this happen to other people both ways: A. Investor pays for years to someone who can't do the job and who isn't passionate about learning, and B. investor has competent talent but keeps postponing delivery to add scope until he loses faith
  3. You need to be able to realistically analyze the return, costs, timeliness, and risks of the project. Even if labor is open-source, it is still a business venture that can fail.
If you don't have these preconditions, consider getting the help of a project manager. At least, nail down #3 so everyone doesn't end-up predictably disappointed.

B. Matlab is a great product. It has some really great advanced parallel computing capabilities, but I would recommend starting with Scilab (free) or one of the other open-source work-alikes until you are through the proof-of-concept phase.

I have seen VBA in Excel used as a front-end to enormous system, but when people see VBA in Excel, there is a Bayesian inference that this is a modest, one-man job. Unfortunately, this often reflects more on the knowledge of the speaker than on the task.

C. EOD statistical modeling of intraday data. Both RiskMetrics and kdb fit your description. Although I do believe that a small team of really smart, focused people can do an amazing job on a hard problem, the tail-end of UI, reports, extensible rules, workflows, and support can easily consume a battalion.

If you have a modest <$10M budget, then I would recommend spending a couple of years sitting with a few students to mutually learn how to do financial programming. This should greatly improve the ROI on the project and help you to better define the scope of the project.

D. Real-time data. The challenges of collecting and processing realtime data has been discussed on several threads. There are two things that I want to say relative to this project:
  1. Given the stage of this project, I would recommend downloading the market data historically using a service such as NANEX since this will allow you to validate your ideas across many symbols without incurring market data fees and with a clear migration path to realtime.
  2. Executing statistical analysis realtime for trading purposes is a completely different challenge from generating EOD market metrics for use in an intraday system. For example, LMAX or StreamBase. If you are building one, don't expect to get the other.
For IT projects (integration with existing software) run at companies where pulling-the-plug has career consequences, the success rate for adhoc projects (costs already budgeted) is 74% for small teams, 58% for medium-sized teams, and 40% for large teams. There is no survey for personally funded projects that have significant technological and mathematical risk being run by micro-teams in ones spare time using dispersed, multitasking, contingent talent.

You wouldn't build a house without blue-prints.
I would recommend breakout the books, with the mutual help of a college student, and treat this more like a hobby than installing flooring. If you really have the money and want to treat this like a start-up, then figure-out your management structure and write-out a business plan. You may have a billion dollar idea, but success comes from realizing a plan.

Steven
More...
 
You must log in or register to reply here.
Back
Top