Journey from investment bank to independent automated trader

[lolatency]

Just wanted to comment on that Tradestation -> C++ DLL -> R bridge I wrote.

I don't recommend doing this. I basically patched RInside, RCpp, and built R.dll, RLapack.dll, RBlas, etc. on Windows. For whatever reason, it's unstable when I take in as a DLL. It works fine as a stand alone console application. I was looking for a quick fix to glue together various components. I don't know if it's synchronization, non-reentrant code, or what, but I don't intend to investigate further and am halting this path of development.

I'd never actually tried interfacing R to anything external before, but the process has been kind of a turn-off. I'm sure people can get it to work correctly, but I'm less interested in making R work correctly than I am just getting lots of results without much development time.

If you happen to have done this before, or have a ready-to-go DCOM bridge or something, speak up. I'm switching over to QuantLib, and possibly ROOT. Hopefully the process won't take me much more than today or tomorrow.


....

As an aside, .. I worked at an options MM back in 2005 that was a small shop that spent a chunk of their time writing their own C++ routines for a lot of what I'm doing. One of the reasons I want out of the institutional world is because of the absurd practices of some places. This particular shop had about 150mil under management, and they never let their non-quant devs see their math libraries. I eventually reverse engineered the damn thing, because they gave us the debug versions of the binaries. I was able to take the symbols, use WinDBG, and such to piece together what they were doing.

There were some real idiots in that place. They shut down in early 2007. (I left almost as soon as I got there, because I sensed the brokenness of the place.)

...

In any case, I'm going through the same predicament of having all of the quantitative pieces together and working smoothly, even though I've prototyped much of what I want in R already. The day job didn't help too much, but I'm betting this long weekend will help me get what I need converted over.

 

[lolatency]

Haha, I just looked at the first post I made in this journal. It happened in April. Assuming I continue to post and stick to this, and take time to develop slowly as I have been, I wonder how long it will be before I go live.

I suppose that is the first thing you posters can mock. Realistically speaking, I don't think I'll have anything running live until ... at best, end of August, ... and at latest, maybe October. Let's see if I can pick up the pace a little bit more.

Goal: Have a first backtest with the new system ready by the end of this weekend. I am tinkering with a few ideas to "correct" my ideas, and won't know if they're fruitful at all until later.

The development cycle for me involves taking a relationship, trading it out with fixed parameters, and then looking at where the system breaks. I usually then pull out all of my books, notes, old code and work around these problems.

Feel free to post tech questions and code to this thread. It helps me stay motivated and keeps my mind on the task.

I'll see if I can come back to this thread with something more concrete by Sunday night -- which, if I can get QuantLib to do what I want, should happen.

 

[asiaprop]

could you give some general examples (I am not interested in your detailed strategy at all) where a retail high frequency trader possesses any edge? I am still curious, never got an answer.

Quote from lolatency:

Haha, I just looked at the first post I made in this journal. It happened in April. Assuming I continue to post and stick to this, and take time to develop slowly as I have been, I wonder how long it will be before I go live.

I suppose that is the first thing you posters can mock. Realistically speaking, I don't think I'll have anything running live until ... at best, end of August, ... and at latest, maybe October. Let's see if I can pick up the pace a little bit more.

Goal: Have a first backtest with the new system ready by the end of this weekend. I am tinkering with a few ideas to "correct" my ideas, and won't know if they're fruitful at all until later.

The development cycle for me involves taking a relationship, trading it out with fixed parameters, and then looking at where the system breaks. I usually then pull out all of my books, notes, old code and work around these problems.

Feel free to post tech questions and code to this thread. It helps me stay motivated and keeps my mind on the task.

I'll see if I can come back to this thread with something more concrete by Sunday night -- which, if I can get QuantLib to do what I want, should happen.

 

[lolatency]

Quote from asiaprop:

could you give some general examples (I am not interested in your detailed strategy at all) where a retail high frequency trader possesses any edge? I am still curious, never got an answer.

Well, I'd rather not say; however, the Weinstein fading and trading thread is one where he clearly is a retail trader with an edge. The Pairs Trading Journal people seem to be doing ok also. Neke also seems to be doing well. They're all retail traders, because their print outs and blotters indicate all retail platforms.

ET has had some successes with retail traders. I'm assuming these people aren't liars, of course.

 

[asiaprop]

not sure whether your reply was supposed to be cynical or not. I specifically asked about "high frequency" strategies. None of those guys runs anything close to high frequency.

I am not sure what your secrecy is really about. I conclude you are either obscure on purpose because you dont yet fully understand your edge yourself or at least are not fully convinced of any edge. Or you are just not the sharing type of guy, which is odd and kind of ironic after you have pretty much banked on others' willingness to share (willingly or unknowingly) by taking others' code/ideas either in a legal or at other times at the very least in a highly "questionable" fashion.

An index options trader who is politely asked to state any perceived edge in trading volatility but refuses to even comment in a general way comes across as either being selfish and arrogant or on the other side clueless.

Anyway, wish you good luck in your secret operation.

Quote from lolatency:

Well, I'd rather not say; however, the Weinstein fading and trading thread is one where he clearly is a retail trader with an edge. The Pairs Trading Journal people seem to be doing ok also. Neke also seems to be doing well. They're all retail traders, because their print outs and blotters indicate all retail platforms.

ET has had some successes with retail traders. I'm assuming these people aren't liars, of course.

 

[chvid]

lolatency - thanks for a good read

 

[propseeker]

Quote from asiaprop:

I am not sure what your secrecy is really about. I conclude you are either obscure on purpose because you dont yet fully understand your edge yourself or at least are not fully convinced of any edge. Or you are just not the sharing type of guy, which is odd and kind of ironic after you have pretty much banked on others' willingness to share (willingly or unknowingly) by taking others' code/ideas either in a legal or at other times at the very least in a highly "questionable" fashion.

An index options trader who is politely asked to state any perceived edge in trading volatility but refuses to even comment in a general way comes across as either being selfish and arrogant or on the other side clueless.

Anyway, wish you good luck in your secret operation.

Quote from asiaprop in this thread http://www.elitetrader.com/vb/showthread.php?s=&threadid=163907&perpage=6&pagenumber=15

Sorry Atticus, but risking that my posts be considered lazy and dubious I also mentioned now several times that I will NOT go into further details. Which options I trade as a function of model output and how I manage the risk exposure is not something I plan to volunteer here. Gomen ne ;-)...

By the way why won't you volunteer to expose to all of us your vol model and maybe you can describe to all of us why you think you have an edge in this market and where that edge exactly lies? [you asked this sarcastically because it was obvious to you the preposterousness of it being asked of you! apparently this does not apply here?!]

maintaining secrecy for obvious reasons deserves respect... being a hypocrite for obvious reasons deserves none.

so, which one were you asiaprop... selfish, arrogant or clueless?

 

[academic]

Quote from propseeker:

maintaining secrecy for obvious reasons deserves respect... being a hypocrite for obvious reasons deserves none.

asiaprop just got shutdown. good job.

 

[asiaprop]

I pointed out numerous times that there is a huge difference between elaborating on very specific strategy ideas and on the other side discussing general aspects of high-frequency trading ideas.

Atticus asked about my very specific setup and I refused to discuss it for obvious reasons and the very same thing I mentioned to lolatency that I am not interested in his specific ideas. I always respect that someone holds on to what works for themselves. There is a difference between a specific setup and the general ideas. And there goes your hypocricy argument.

So, if we cannot even discuss general ideas anymore then what is left? Critizing others because their points were made without a factual base is not something I respect, however, what you do is NOT adding any value whatsoever.(point me to it in case I miss it, I did not read any of your posts vs the time you obviously spent to dig up my posts, thank you for the attention I was hoping for more constructive contributions).

P.S.: Since when should anyone respect another person who proudly stated he stole code and ideas from pretty much every shop he previously worked for, something he would be immediately fired for if found out. Any sales person that steals client lists will get himself/herself into a huge legal dispute even after the fact, what makes the same actions by an IT guy respectable? Just wondering...aside that fact I still appeciate lolatency's refreshing writing style as mentioned!

Quote from propseeker:

maintaining secrecy for obvious reasons deserves respect... being a hypocrite for obvious reasons deserves none.

so, which one were you asiaprop... selfish, arrogant or clueless?

 

[asiaprop]

Why don't you address me directly if you disagree with anything I said?

Re your role model Ernest Chan I am sometimes lost how some of you with IT backgrounds start to salivate just because you read about someone mentioning "correlations, Fourier transforms, matrix inversions" without checking the author's track record.

Quote from academic:

asiaprop just got shutdown. good job.