Is fully automated trading a fantasy? Your opinions here

[MarkBrown]

Quote from Trader KGB:

Let's take the low end of that range, scalable 20% monthly and 4 years. Starting with $100k that's $630M gross compounded monthly. The world's smartest quants working for the world's biggest hedge funds can't even come close to the figures you're quoting. Something is exponentially off or you'll soon be the first trillionaire..

just wondering who do you think the worlds smartest quants are? i am wondering if there is a list somewhere?

do you think any of them are on elite trader talking to regular people? how many of them i wonder are high school dropouts? how many have multiple phds i wonder? i wonder how many own there very own 3.8 million dollar copter and make 138,000 a month payments on it? just got me thinking about a world i have no clue about.

if you could day trade a emini sp for 2500 margin - on that margin do you think anyone could make 2500 a month?

but back to the original questions. some guys mentor wants to keep him dependent on coming back to the guru no doubt for lessons that cost money or to trade through a firm where he shares in commissions some way.

so he tells the new trader you cant trade mechanically its impossible. i agree the newbie cant trade mechanically and make money. i can but he can't i agree with his guru.

 

[Wayne Gibbous]

Quote from traderdragon2:

And now for a short break.

Wall Street pic

Money aways makes women horny!

 

[Trader KGB]

Quote from markbrown:

just wondering who do you think the worlds smartest quants are? i am wondering if there is a list somewhere?

Can't say I do. Though I have a strong inclination that many of the aforementioned quants work for the funds behind these figures:

http://www.hedgeindex.com/hedgeindex/en/hedgoverview.aspx?cy=USD

..whose returns are a far cry from the figures that sigsegvboogman mentioned. I'm not doubting the possibility of double-digit to triple-digit monthly returns for an ATS in certain scenarios, but scalable 20% average monthly returns since 2000 would control all of the money on the planet.

 

[sigsegvboogman]

No, not necessarily. My definition of "scalable" I guess is my own definition. One old system we traded for maybe 5 years only performed to a certain optimal peak cash level that the market could absorb. One of our main systems now can take maybe 100x the cash over the system mentioned above and maybe more. There is a limit to what you can do with this one, but the ceiling is much much higher. So I guess that is why I called it scalable. You can't keep reinvesting your cash to make more cash with regards to what we are doing. We have an average cash balance and run with that because you feel its optimal. The cash needs change as the market changes, but the algorithms stay the same for the most part.

 

[Maverickz]

Quote from Trader KGB:

but scalable 20% average monthly returns since 2000 would control all of the money on the planet.

No system, automated or manual, for Forex, Stocks or any other instrument can scale indefinatly. Why? Because if it did that system would eventualy become to large in the given market to get fills. Imagine if you suddenly tried to sell a million shares of a stock or a million contracts of some futures. Who would take the other side of the trade? What would that do to the price?

The ideal automated system would actually only scale to the point that the returns could be pulled to use as spendable income and provide enough of that consistantly for a person to live comfortably on. If it ever needed to scale it should scale by adding additional instruments instead of scaling one or two intruments into infinity.

 

[MarkBrown]

Quote from Maverickz:
No system, automated or manual, for Forex, Stocks or any other instrument can scale indefinatly.

yes you would need separate accounts and be possibly long and short (on same and or different signals) at the same time in the same and or like markets. you could actually cause flight from one financial market to the next depending on how your system trades.

i know a system once which controlled most the professional volume in a market and after two years of stupendous profits it got run into the ground. expect about the time you start killing it an end is imminent - as a lone system. as a group you can survive and do your thing.

mark brown

 

[TrueStory]

It is not necessary to scale into infinity. One need only add markets and new systems to trade more money. It is certainly possible to trade fully automated over many markets and systems and earn fantastic returns, better than Wall Street.

20% plus per month? As was mentioned above, people that claim that are generally shills who don't know what they're talking about. Anyone making 20% a month will be retired and living on a yacht in 2 years.

However, 50-100% per year IS doable using automation...

 

[MarkBrown]

Quote from TrueStory:
Anyone making 20% a month will be retired and living on a yacht in 2 years.

wondering how big a yacht would qualify as being rich? also i think a guy would need one of these on the back of the yacht to fully qualify as being rich. also it would need to be bright corvette z06 yellow.

http://www.mdhelicopters.com/products.php?id=MD_EXPLORER
notice it has no tail rotor

http://www.4yacht.com/main.asp?-vessel_basic_info.asp-&vessels_id=59281&curr_id=7

i mean how many millions does it take to be successful at trading?

so you have to have a phd working for a large financial firm with deep pockets to be a real legit trader?

 

[booking]

I spent a lot of time looking at developing an autoFX system. I found that I could create a system that generated over 600% profit in one year through optimisation. But try it on another year and it would lose anything up to 90%. So you de-tune your system and eventually you end up with one that makes a modest 15-35% profit a year with your available test data.

But this is the whole crux of the problem with auto-systems - you have to undergo long periods when they dont make money and you have to reduce the risk so much that the returns are unspectacular - otherwise you risk losing much of the capital the system is playing with.

And just because the system works on the test data you have - there's no guarantee that it'll work on future data - and do you want to risk your capital on something which unattended may lose your life savings?

I think automation has it's place - particularly with arbitrage - a computer can take advantage of an arb situation far more quickly than a human, but teaching a computer to predict future market moves... in my experience your simple system becomes more and more complicated to prevent those big drawdowns and ends up with so many parameters that they become impossible to maintain.

 

[traderdragon2]

Quote from booking:
So you de-tune your system and eventually you end up with one that makes a modest 15-35% profit a year with your available test data.

A consistent 15-35% a year would beat the vast majority of mutual funds. Nothing to sneeze at.

But this is the whole crux of the problem with auto-systems - you have to undergo long periods when they dont make money

Not true. There are good and bad auto systems, just like discretionary traders.

and you have to reduce the risk so much that the returns are unspectacular - otherwise you risk losing much of the capital the system is playing with.

Again, only true for bad systems.

And just because the system works on the test data you have - there's no guarantee that it'll work on future data - and do you want to risk your capital on something which unattended may lose your life savings?

True for discretionary traders too.
A well written system will die a very slow death and give you plenty of time to turn it off before it starts going damage.

I think automation has it's place - particularly with arbitrage - a computer can take advantage of an arb situation far more quickly than a human, but teaching a computer to predict future market moves... in my experience your simple system becomes more and more complicated to prevent those big drawdowns and ends up with so many parameters that they become impossible to maintain.

I havent experienced that at all.

Robust systems behave well, last a long time, and have edges that fade very slowly over a period of many years.