Academics have proven many times that markets and TA are random. We've had threads on ET started by people that had a successful automated trading thing going, sometimes for years, that just gave back all the gains and more at some point.
Academics have proven many times that our roads are built all random. We've had threads started by people that had successful cars going, sometimes for years and they just smashed their car and more at some point. The point worth considering, it's not systems so much as ourselves/human nature. We let our guard down, get greedy, sloppy, careless... Then we have brokers and data, computer systems to contend with which may fail. This doesn't relate to an auto system not being profitable, it relates to accidents. We can have a simple system and it may make money. We can build a more complex system and it may make more money but there is more to go wrong unless it is built with lots of safety features.
I think it's just that a lot of people underestimate the difficulty of this problem. However, there are plenty of capacity constrained strategies that one could implement and make a reasonable return on. For example I have a friend who trades for himself doing mostly stat arb strategies with high sharpe (~4) and a rough capacity of $10m clearing through IB. However, I think he pays around $100k per year in IT costs (e.g. proximity hosting, hardware). It also took him about 2 years to get the system fully operational and this is with a doctorate in computer science.
It is probably a lot easier and more profitable to train as an accountant, join a company listed on an exchange and know from an insiders point of view whether or not to invest in that company. Most of one's friends will also be accountants if you get my drift ?
A long time member here, although I don't post much. To this topic I can contribute so here it goes for whatever it's worth: Is it possible to build an automated, profitable trading strategy? This is an oversimplified question (it doesn't actually ask the right kind), but in short, yes, without a shred of doubt. There are a number of algo shops that make a killing at this every day. There are a number of individual traders who make a great living at it as well. Now, what is the barrier to entry? There are at least four that I know of: 1. Money. Not as high as some might think because many algo strats are not very latency sensitive. However, one must be capitalized with at least 12 months of expenses in addition to trading capital. In my experience, 12 months wasn't nearly enough, but I am different, just like everyone else. 2. Experience. This is a very, very big one. In my view one has to be at least a promising discretionary trader (someone who knows how to take money out of the market, but not necessarily keep it) before any automation can be even thought of. One must understand the markets very well, through spending thousands of hours trading them instead of reading books and mining data. That latter point is key: a purely academic/scientific initial approach is fundamentally flawed as it takes out the organic nature of the markets out of the equation and thus prevents one from gaining a complete understanding. However, this approach must by all means be applied, but only after said understanding has been gained. 3. Time. It takes 18-24 months to code a dependable platform. Being able to code this platform implies proficiency in programming and systems overall, in addition to having obtained the above experience. Or, it implies that there is a free access to someone else's proficiency. In my experience, this is an exceedingly rare combination of skills, especially given the low survival rate. By the way, existing tools can be used and there is evidence that they can be used profitably. There are notable disadvantages and it may take an almost equivalent amount of time to get them set up/trained right. 4. Drive and commitment. Hardest and most important of all of the above. Have no illusion, making this work is very, very hard. And once you think you realized just how hard it is, it gets harder still. So, you must really want this. You have to want it so bad that you dream about your system/strategy/whatever at night. You think about it on your commute home and you think about it as soon as you wake up in the morning. And you work all the time. Evenings, mornings, weekends. holidays. Nights. And when all of this doesn't produce the results you expected, you take a hard look at yourself and do it all over again. And to make it yet harder you must balance it all with your life. The actually important things like your family, your parents, your friends. So, is it possible? Sure thing. "It's long way to the top if you wanna rock'n'roll." /Wulfrede
Show me one academic study that claims 'roads are built all random'. Is this what passes for discourse these days?