A variety of system development threads have popped up in the past, but they always ended up in mechanical vs. discretionary trading methods and all the mumbo jumbo that comes with it. Discretionary prop traders vs. aspiring system developers. A few quants come in and talk about their experience. Then you find out most of the quants that spoke up were just crunching numbers for arbitrage-type advantageous, and not the swing/day-trading approaches that many here trade/test/aspire to. Then Don Bright jumps in and says he's seen 'em all and hasn't seen a profitable mechanical system yet. He'll back it with his money if someone proves it to him. Then he says 'but he has his own guys looking at 100% mechanical systems' and they 'are having some success'. And always in there is discussion on over-optimizing and a claim that all mechanical systems are doomed to fail due to nature of changing markets and/or curve fitting. If we can do away with all of the noise and deal with real signal, I'd be more than happy to contribute. There are a few other places like www.wealth-lab.com and a couple Yahoo mailing lists where the signal level is higher. I'd love to be proven wrong though and get a real good thread started on system development.
great. i dont want to bother you, but i do not have the ability to do this kind of testing. could you test your method on ES/NQ with a stop on 1.0/1.5 and target on .75/1.0 ? how would that change you test-results.... thanks trading
The problem with a strictly mechanical system is that it can't see the "color of the market" (as Grubman, my favorite Telecom analyst said). I don't understand how Mr. Bright can make a claim about no mechanical systems being profitable -- Larry Williams himself has come up with a number of systems that he used in the past to make a lot of money. I believe the best advantage to a mechanical system would simply be the ability for an autotrading program to react much faster to the market than a human can -- but even that speed is still slightly nullified by the time it takes your order to get to exchange, the delay of your datastream, etc. If a human can do it in 3 seconds and a computer can do it in .01 seconds, and there is a 3 second delay to getting an order to the exchange, then the computer is no longer 30 times faster, but only twice as fast. aphie
Aphie, Actually, what I like in systems is that I can get comfortable with their parameters, strengths and weaknesses, prospects, etc and then at some point I can delegate their trading to brokers and/or computers. That takes care of two significant threats most of us share: real-time distractions (got to answer the phone, take my daughter to the doctor, etc) and real-time "I've got a better idea" viruses (aka lack of strict discipline, because if we're smart enough to develop a good system we're usually not the type who follows directions exactly.) As far as Don Bright et al are concerned, if they just take a serious look at Futures Truth they may change their minds. Or not
Yannis, So do you believe a "blackbox" could be designed to automatically trade throughout the day, totally remove the human component and remain consistently profitable? I'm not saying I don't believe this, I am just asking your opinion. This would solve the "emotional" aspects of trading (unless it has a down day and you unplug it or format the hard-drive). aphie
Aphie, Yes! If we just look at its day-to day trading, absolutely! The question is, how profitable can it be, how often does one have to interfere and adjust parameters, how to control risk, e.g., malfunction, etc. Again, just review the literature, e.g., Futures Truth, and see what's available out there, easily mechanizable via a broker (and increasingly so via computers.) Let your imagination fly a bit, on the question, like, if these things are available to everybody for a few $thousand, what proprietary stuff may be in their own computers to trade the big money? FT is the tiny tip of the iceberg, imo. That's why I am a fervent believer in the existence of the Holy Grail, if we just define it right. And I can feel it in my bones that the successful, big traders don't want me to believe that anything like that is even possible... (Now, let the arrows and spitballs fly... ) "What is the HG" may be another good new thread.
It wouldn't be a "blackbox" if you developed the trading rules and just gave the system to a broker to trade. You get the convenience of supervised execution via a broker and the built-in discipline that goes with it. There are brokers that will execute your orders automatically if you give them a plugin for various trading platforms. They use an automatic execution platform with some level of monitoring/supervision. Obviously platform reliability and execution are key issues here, but blackbox systems (whether you purchased a true blackbox or designed the system yourself and handed it to the broker - i.e. it's only a blackbox to the broker, not to you) are a reality.
what gives blackbox systems a bad name are stupid ads in stocks * commodities and other magazines. purely automated trading is possible with consistent gains on ES and NQ (I find NQ more nerve wrecking, but that's just my opinion). but no one has to be so strict about it, it's possible to mix generated signals and discretionnary trading. just a hint... Back to WarEagle's comment : maybe it is pointless (hehe!) to target 1 point scalps, at least in a systematic way. Just like him, we never succeeded in that way. I would not say it's not possible. But obviously there are many easier ways to trade this. tntneo
TNT and others, I will scalp for .50 points at the very onset so I can accumulate points that will allow me to play larger moves. Whatever I have to do to survive, I will do it. Between a cubicle / Dilbert life and staring at the screen all day, I want to stare at the screen all day -- and if I ever get to the level where I'm trading 20 contracts, then life will be grand. I finally figured out how to get esignals DDE to work in Visual Basic as well. I wrote a small program to give me a "bid to ask" moving ratio on the e-minis. Programming this stuff is a lot of fun, but my main objective is to design an interface that will allow me to trade without screwing up. A couple times I put a limit order in on the simulator, forgot to cancel it, went to a meeting and came back and I was holding 10 contracts in my account and a net loss for that trade of $3,500. If that happened with real money, I'd be #(**%(#@$**#$*#@ aphie
Well I've been scalping es for the last 2 days and now it all makes sense. You say, "1 or 2 ticks stops? That's ridiculous! Insanely tight. Now me, I use a nice respectable 4.5 stop. That's the proper way." Oh yeah? So then tell me, what do you do when the market has moved against you by 4 pts? I'll tell you what you do. You sit there and say, "If it moves against me 1 tick I'll stay, but 2 ticks and I'm out." So the scalper simply starts off where you end up. I am convinced the problem regardless of stops or targets is r to r. So scalping simply forces you to at least get your minimum r to r at 1:1 simply due to the mathmatical fact that the minimum profit you can ever take regardless of your system is 1 tick. And most traders are doomed because they take small profits. So the scalper is simply less doomed. Easy? Hard? Just depends on you. I'll tell you this, when you trade for 6 handles you see all kinds of good scalps you would take if you were a scalper. And when you scalp, you sit there and say, "Man, that would be a good place to get in if you used 2 pt stops."