My teacher only charged me a one-time fee and we talk on the phone often, to this day, and I purchased it 6-7 months ago. I'm one of his only students who actually keeps in touch as often as I do. He didn't sell his method to alot of people. In his experience he saw a lot of system vendors not even trading their own system. They just wrote something that tested half-way decent in tradestation or some other platform on old data and then just packaged it and sold it. Or if a system was profitable, it required a large starting balances $250k+ and had to handle large drawdowns at times. you should expect 10-30% annually. He said if there really was a fully automated system that could consistently print money (profitable no less than month-after-month, and usually week after week), it would NOT be sold or leased. (remember, for smaller starting balances). But again, what are the starting balances? What are the percentage gains? Because $5-10k I believe is what most of us start at when trading futures. Some may put in $20k initial. Unless you inherit your fortune, most of us had to start from rags or at least a more humble beginning. Hence the "low" amounts. I don't see why 5-10% weekly is not possible. Especially with compounding. Even without compounding, With a 5k acct, $500 (10%) would require 50 ticks net x $10 per tick. Well, a little over 50 ticks to take out the commissions (say $5 rt). The overage would depend on the volume of trades. This is true. But, if we could afford it, most of us would probably fly (or own) our own jets vs having to fly commercial . Most of us own a car and toliet paper Yes, but again what kinds of drawdowns and balances are we looking at? I can't imagine if it is 20% monthly for real that the drawdown would be very large. Is this being sold to the general public? Also, there has to be some sort of cut-off point to where you could only trade, say 10,000 lots per day or xx% of the average total volume in a particular market. In other words, there would be a saturation point for each market traded. of course, once you're acct is built up to that saturation point, well, that's definetly enough to live off of. ---- This thread is actually quite interesting. I appreciate all of the insights. I think the bottom line is, in order to make a living with fully automated systems ($3k per month for a single person enough?), you must first have a substantial amount of money to start with, say $100k. So then the little guy would first have to hustle and grind (either manual trading or some other forms of income) to build up to that minimum first.
This idea that you cant make more money with an automated system using a small account is completely unsubstantiated. It seems nothing more than the discretionary crowd trying to make themselves feel better.
Many people hate the idea of automation working at all. "My mentor told me it doesn't work" is such a waste of time. People that spout nonsense like this because it fits their agenda should not be trusted. I'll say it again, there are many successful traders and managers using fully functional automated systems that make consistent money. You can reduce the risks mentioned above through trading multiple markets and multiple systems, resulting in a very smooth equity curve. Nothing "prints money," that's not the idea at all. It's simply quantifying discretionary trading methods into a system that reduces user error and emotions. Those who "optimize a system for 600% in 1 year and lose 90% the following year" simply don't know anything about system development. And yes, when done properly, it works.
I can envisage a automated system to work, but to what extend? this require two or 3 component, first is the ordering / decision engine that link up to the data provider / broker, this is where all the live rules are kept. (concern with speed and accuracy) second is the libarian engine that file the data into a database (concern with effective database both read and write) thirdly is the datamining / neuro-network / genetic algorithm engine that constantly review the live rules in the first engine, search for new algorithm and back test the rules, if it is successful it submit it to the first engine. (optimisation, heuristics) this way you should have a complete system that is more or less hand free. the only thing you would have to be carful of is that dont over fit the model. the monitoring function of the system owner becoming not a trading monitoring but model fitting monitoring. the key thing is then how much this will make, if the anual reutrn is less then 15% you might as well put your money into properties, if the annual return is around 30 - 40% then the system is worthwhile, which I really doubt if it is acheivable.
I think the % of return is the last barrier that I need to commence on the development. may be it is off topic, any free data provider out there?
*Sigh* There's a lot of generalisation and "yes it does work if it's done right" type comments but very little to back it up. And as to the comment I don't know anything about system development - I have worked in several banks in the algorithmic trading sections - and NOT ONE OF THEM uses algorithmic trading for speculative investing. Most of the algo-trading is used to automatically hedge client positions, to fill large orders through a series of small automatic trades and for arbitrage. If you want a sure fire way to make money then it's arb, the algorithms are (relatively) simple to code. However I don't believe any home user could get a trade feed quick enough to take advantage before the numerous banking arb systems have already done so. Gone are the days when you could see a 20-30 point discrepancy on cross-currencies before the banks would jump in. I would love to let a system do all of my trading - and I've developed a few systems that make money - but the fact is I can out-perform all of them with my discretionary trading. I thought about automating my thought processes but it would be an extremely difficult thing to do as I take many things into consideration - not just price/volume - but I look at other charts to see what's happening there. It all comes down to changing market conditions, all of the systems could make a profit most of the time - however there were periods where profitable setups in one period resulted in losses in the other, it's very difficult to code around that to differentiate the two - what you typically end up with is a system which avoids those difficult situations and then it hardly trades at all. I think anyone posting here saying it is possible blah blah - probably has their own auto-system which hasn't been back tested properly. You need to run it on data you've not optimised it for to see how it performs, this is particularly true for any system which uses indicators such as moving averages/RSI etc.. (which from my experience will fail eventually). Run your system on a different instrument (e.g. if it's for cable run it on Eur.Usd) - if it fails to make a profit - that's a big red warning light flashing.....
You are contradicting yourself. That is exactly the advantage of a blackbox>>>emotionless trading. Thus when there is a huge rally or meltdown, it supposed to trade it the same way... An evidence for their existence is IB's Trading Olympiad. All the systems there are fully programed. Check out their returns....