Is fully automated trading a fantasy? Your opinions here
- Thread starter ybfjax
- Start date
all the really good systems people out there, the ones with C manuals under their pillows and 5000+ trade forward tests will understand this comment:
leave systems alone.
trade discretionary, please!
leave systems alone.
trade discretionary, please!
The article says 70% of the work is done by computers. I wonder what that means? It goes on say that they can tweak variables at any time, and that they are constantly developing new models and algorithms.
So maybe for a time period it is automated while running, but it implies if left alone, the algorithm would eventually faill and lose money. Interesting article in what was said and not said.
MarkBrown admitted this in the first or second reply to my original post on page 1. You can't just push a button and leave the system completely unattended forever.
Another poster said that most people simply don't have BOTH the programming skill AND the trading skills to be able to complete such a robust system.
In the article, http://www.bizjournals.com/sanjose/moneycenter/story.html?id=3498 it mentioned that the TradeBot currently trades between 50-100 million shares a day. And it was that way since 2002. The bot started in 1999 with a starting balance of $25k. Then the article also mentions twice that the TradeBot can have $50k days (on the 50-100 mil volume shares). No exact annual figures were given, but "double-digit profit margins" were mentioned.
Assuming that meant double-digit millions ($10+ mil), to go from $25,000 to $10,000,000 in 3-7 years? Hell, just $1 mil?
Let's see what kind of percentages that would require:
$25k >> $1 million
3 years
110.8% compounded monthly
102.4% compounded weekly
5 years
106.4% compounded monthly
101.5% compounded weekly
7 years
104.5% compounded monthly
101.02% compounded weekly
$25k >> $5 million
3 years
115.9% compounded monthly
103.5% compounded weekly
5 years
109.25% compounded monthly
102.1% compounded weekly
7 years
106.51% compounded monthly
101.5% compounded weekly
$25k >> $10 million
3 years
118.1% compounded monthly
103.92% compounded weekly
5 years
110.5% compounded monthly
102.34% compounded weekly
7 years
107.4% compounded monthly
101.66% compounded weekly
$25k >> $20 million
3 years
120.41% compounded monthly
104.4% compounded weekly
5 years
111.8% compounded monthly
102.6% compounded weekly
7 years
108.3% compounded monthly
101.86% compounded weekly
This is not fully automated, but not bad for 70% automation. For them to reach such gains so quickly, the drawdown appears very low.
This other article, http://www.elitetrader.com/vb/attachment.php?s=&postid=1263772 , was quite a fascinating story. I wonder what his starting balance was?
It would appear that this system, Aquarius, which was originally designed for exchanges to make them more efficient, was built for extremely large volumes with rapid longs/shorts and quick liquidations (scalping). So I'm guessing multi-million dollar starting balance? Oh well, their $90+ million that they rake in every year is incredible.
if i had a secret to tell - it would be that no matter discretionary or mechanical trader. the "KEY" to wealth is two things in trading.
1.) LOW and i mean like close to $1.00 transaction cost for futures.
2.) LOW and i mean like 1% drawdowns allow huge profits and fast recovery.
3.) the other things are frigging mind boggling - we set down and addressed every single issue of data, locations, redundancy, latency, more what if's than there are what if's, hardware, software, clearing services for lowest cost transactions, multi location accounts for anoniminity, can we spell paranoia?, our own server farm, data filtering, order server with types of orders that the exchanges don't allow - rest on our server awaiting execution, systems that signal a buy or sell order and only place the arrows on the chart when confirmation of the trade being filled comes back, dealing with leveraging split fills, downed data feeds, acts of god, you name it, on and on. perfection no near it compared to anything else there is yes very close. add to this i have been active building off old research new models which are as sound as the markets are untouched by the exchanges. that is to say if they will quit messing around with the tick size, trade increments and value of tick - i could be well off if they would stay stable for a few years.
as is i am struggling to make my copter "not chopper" payments of 138,000.00 a month - nor do i have a mega yacht to land it on yet. but i do enjoy my new z06 vette that is more dangerous than a loaded gun and my wife hers.
fellows if your going to buy a sports car do yourself a favor and let you wife pick the color and get her one too. you will be a happy man that way!
http://i19.tinypic.com/2j64ygg.jpg
Mark' chart art!
http://www.elitetrader.com/vb/showthread.php?threadid=89844
Mark Brown
exchanges have programs for professional traders that slash commissions cost to as few as 10c per side on heavy volume. there are professionals that trade in automated mode more than 10.000 contracts, per month with just 5 lot size. here is a example. on average it trades around 200 times per day.
Wow. Obviously you basically are your own brokerage AND data-feed supplier. And obviously the programming seems to go beyond a one-man show. How much does this cost? If you could break down the totals to give us a rough idea, that would be interesting.
Do you think that a one-person programmer could accomplish this with retail methods (outsourcing the datafeed, brokerage, etc. like tradestation.com DTN profitx etc. for example)?
But that is just the exchange part of it....what about the broker? You mean to tell me that large professionals (defined as > 3000 rt per month?) are getting $1 rt? They must be their own introducing brokerage, or closely associated with one.
Yeah. My teacher's method is designed to either breakeven or profit 85-90% of the time. I didn't see how people could start trading for a living with such low balances unless their returns were large and/or they quickly compounded their returns. I have seen that (at least discretionary, but obviously also for automation) it is very possible but the winners must be plenty and ridden long and hard. It kind of reminds me of this quote here:
More from Teresa Lo
Wouldn't it make sense to rent it or do one of thoe pre-paid executive charter programs for use on a case by case basis? Or do you fly it yourself?
"Yeah. My teacher's method is designed to either breakeven or profit 85-90% of the time."
We typically don't care about how much winning trades you are making.
What is important is if your system has underlying edge, how profitable it is considering yearly returns/DD , how stable it is and how much space you have on average trade net profit not to be caught by what I call the "bid/ask spread necessary long term losing bia"( I don't know if it means something in english ).
We typically don't care about how much winning trades you are making.
What is important is if your system has underlying edge, how profitable it is considering yearly returns/DD , how stable it is and how much space you have on average trade net profit not to be caught by what I call the "bid/ask spread necessary long term losing bia"( I don't know if it means something in english ).