How long before you were fully automated?

[TSGannGalt]

Quote from MAESTRO:

There are over 1000 black box based hedge funds in the US. There are forums with thousands of members solely dedicated to the automated trading. Our own fund runs over 200 robots. You get an idea.

200 models... All active? If so, how many in archive?

Anyways, nice number. I've got 57 that I actively monitor and 16 that are actually live, as of this writing. (and I'm solo)

My point is, there no limit to the number of systems out there. The number will keep on growing because the market changes. If the market didn't change, the # of models would be static too.

I'm going to skip through all the little ET discussion patterns and get right to the end of all this... and say that it comes down to what Robustness is and whether it exists or not. Majority of ET developers tend to talk about robustness via Walk-forward or out-sample data, but that really doesn't define whether the models itself is robust. Most "believe" in the concept that robustness is possible without truly understanding what it is.

Only my 2 cents,

.... and I stopped thinking about developing a "Robust System"

 

[propseeker]

thanks for the links, have actually been a fan of most of those sites for years. for the most part though, those are primarily quant based forums... the beautiful minds who come up with all the risk and valuation models but who wouldn't know how to pound a bid if they were looking at it... nor should they, as they have entire desks dedicated to that, and entire other ones just for the IT.

with the exception of relatively few guys, fdax the super spy being one of them, the number of roll your own independent auto guys are pretty damn few. the number actually talking about it even less. a forum of thousands of automated traders would be pretty cool though.

 

[travis]

I've also gone through those links and some of them are called "forums" but are not real forums while others are not about trading (at least not entirely). The biggest automated trading forum that I know of is still this section of the Elite Trader forum, and it only has a few hundred active members.

Maybe this is good, that it isn't so popular. I don't think it would be good if they taught it in universities - it would make things harder. But can you tell me why is it not being taught in universities? I could think of a reason - in Italy we say "the knowledgeable do, the ignorant teach". But this doesn't really explain why it isn't being taught in universities. It maybe explains why a lot of these courses and systems that are being sold, do not work. If they worked, the guy wouldn't need or want to sell them. What do you think? Is this the biggest AT forum on the web, why is AT not being taught in school (and if it were would it make things harder), do most of these courses, books, systems being sold work?

 

[bespoke]

Quote from travis:

I've also gone through those links and some of them are called "forums" but are not real forums while others are not about trading (at least not entirely). The biggest automated trading forum that I know of is still this section of the Elite Trader forum, and it only has a few hundred active members.

Maybe this is good, that it isn't so popular. I don't think it would be good if they taught it in universities - it would make things harder. But can you tell me why is it not being taught in universities? I could think of a reason - in Italy we say "the knowledgeable do, the ignorant teach". But this doesn't really explain why it isn't being taught in universities. It maybe explains why a lot of these courses and systems that are being sold, do not work. If they worked, the guy wouldn't need or want to sell them. What do you think? Is this the biggest AT forum on the web, why is AT not being taught in school (and if it were would it make things harder), do most of these courses, books, systems being sold work?

And what exactly would they teach in a lesson in regards to automation?

 

[travis]

Quote from bespoke:

And what exactly would they teach in a lesson in regards to automation?

Well, they should start from the hardest part - programming. So some students would give up immediately and not waste their time. Then those who make it can easily get through configuring all the software needed. Then, for last, you can teach about the markets - the costs (commissions, slippage, other), the various financial instruments and technical analysis. Four years would be plenty of time. But maybe I am wrong and it is already being taught in universities. Let me know.

But I was also thinking... traders making money would not care to become teachers and would not be allowed to become teachers. The academics despise money, so they would never allow it as a subject, nor learn about it. Maybe I could modify the previous quote as "those who want to do, learn and do, and those who want to learn, learn and teach". I don't know much about this, I am just guessing. Let's hear from others.

 

[Euler]

Quote from bespoke:

And what exactly would they teach in a lesson in regards to automation?

I think I agree with your sentiment here -- teaching "automated trading", as an academic subject, would be very difficult, as ATS's are systems using disparate components, and the value added often depends upon uniqueness. Some automated traders may use pure math, some may use NN's, etc. but the real value only comes with appropriate application and the fact that not too many others are doing the exact same thing.

The danger of a lack of uniqueness was demonstrated in August of 2007, when many value-based algo funds suddenly realized that they were using virtually the same strategies, and on margin. :eek:

If your goal is to become an automated trader whose focus is on math, you're probably better off going to a top PhD program in math. Same story with CS, stat, or mathematical finance. Then you'd still have high tech/defense/government/academia to fall back on, and beyond that, maybe Rentec will even interview you

 

[propseeker]

i don't know where they teach the hands on details anymore, but there was a pretty sophisticated program at upenn sponsored by lehman a few years back. they had a full order book sim and contests pitting bot against bot. look up PLAT for more info of what's been done in the academic space. i do know some schools still have pretty contemporary programs (not on the level of PLAT), giving grad students access to good exch mkt data and encouraging analysis... upenn, loyola in chicago and baruch in ny to name a few.

ultimately though, this is a game where a pretty broad skill set and many thousands of hours are needed to even begin to play. not too many people fit that profile, and teacher or no teacher you have to be a bulldog to want to figure things out on your own.

 

[bespoke]

I believe writing the automation is the easiest step in the process of having an automated strategy. The hardest part is sitting in front of the screen for months or years watching the markets trying to find an edge. I don't believe that can be taught in school.

Risk management, analysis of trading statistics, and how to avoid the pitfalls of over optimization can all be read in just a few chapters in a lot of trading books and could probably be taught in half a semester.

I have no idea what Maestro is running and what level of mathematics it requires so I can't say what level of schooling you would need. But the strategies I run require no math, no indicators, and could be spotted by ANYONE who looks at the market long enough. When I first started trying to find strategies I made them so complex using anything and everything I learned in school that I thought made me smart. But instead all I had to do was watch the market and it was right there in front of my face.

Ok, I may have said too much....

Btw, I voted < 1 year

 

[powerfade]

Quote from RiceRocket:
No one can have their computer trade for them and make money. This is a bullshit question.

I noticed you posted in the 'short the US dollar' thread and someone had to explain the instrument you were claiming you're short. It seems you didn't understand the DX and the fact that a short DX trade is in large part a trade on the Euro as well.

Now you're here with the stupidest comment I have heard in the past year or so.

Tell me. When you come into a forum with a bunch of successful automated traders and belch out the comment that automated trading doesn't work, does it make you feel like you're a part of a group? Is it better, in your mind, to be hated than ignored?

 

[propseeker]

Quote from bespoke:

I believe writing the automation is the easiest step in the process of having an automated strategy...

for model-centric low-frequency strats, sure, maybe, but for execution-centric strats it's nowhere near trivial... and at the end of the day, once you scale all those low-freq's... it all ends up being execution-centric.