Quantifying mental "systems" is pretty tough, but I tend to agree that we can certainly "model" by example...and especially model what "doesn't work"... Never add to losers (except in xxx case), Never take home a loser (except in xxx case), Never buy on bid, never sell on offer (except when closing or MM'ing)....well, as they say, the market will always be smarter than we are, so were better off trying to respond to it rather than trying to anticipate it. I usually just say this: "show me your sheets" and if you have more money at the end of the year than at the beginning, you have a good system (at least for now). Keep the egos out of it, we need the best equipment, costs, strategies, and even some "luck" (as much as I hate to admit it)...to keep plugging away year after year. Good traders seem to be very adaptable to willing to make trading changes quickly.
All: I want to thank those who commented on this topic. Unfortunately, I have not learned a whole lot by the various comments other than each person's subjective opinions about the possible usefullness of day trading systems. I was hoping to use this message thread as a location for people to contribute real mathematical formula and setups to be used in developing an effective day trading system. If people are willing, I think we could all benefit from those types of posts. Thanks, TAHOE
Well, it's kinda tough to ask people to post their trade secret... but let me start off with my limited experience in system development... I can talk about what does NOT work, so that people can cut down research time by many fold. 1) Canned indicators don't work... most of them are just derived from price action and always lag... they can provide some guide lines to prevent you from on the wrong side of the market, but no good to give entry/exit signals. 2) Daytrading system on stocks are tough to develop, due to the uncertainty in slippage.. thus most people try futures, especially eminis. 3) Complicated stuff doesn't work.. if you cannot explain your system in 60 seconds.. it's possible you are over optimizing. 4) back test enough data... for daytrading systems, probably 1.5 years of data is needed.... that of course, depend on the frequency of your trades. So in conclusion... you can forget about testing gazillion combinations of those textbook indicators and hoping to find that holy grail..... Go look at 1000 intraday charts and find price action set ups you think is a winner, code it, back test it, forward test it, and maybe with enough work, you find something.
In the interest of getting this thread back to the original topic I will post the basics of my favorite system. if you go to wealth-lab.com you'll see that several of the developers there are focusing on band-style systems. The basic idea: a. Take the 10 day MA of a security. b. Draw lines 1.5 STD on either side of the MA c. Buy when the price of a stock hits the lower band. d. Short when the price hits the upper band e. Set a 10% profit (or a profit of 1.5 ATR). f. Most of these systems actually work better with no stops set. Just keep position sizes small. I know this is controversial but you can backtest these things for yourself. On Nasdaq 100 stocks these systems (and there are many variations of the above system on wealthlab) perform very well. About 70-80% profitable trades. The trick is all in the money management and increasing size as your equity increases. Several innovative approaches developed on wealth lab once the first of these systems was posted. Among the developments were: a. replacing Bollinger Bands with Keltner Bands (bands based on ATR rather than standard deviations). This increases the number of signals without greatly reducing the probabilities. b. Replacing the simple moving average with a moving average based on John Ehler's work in Mesa Adaptive Moving Averages (MAMA). These moving averages adapt much faster to vertical movements in the price of the stock. The results are amazing. I wish I understood all the math but I just got the book and I'm working on it. On Nasdaq 100 stocks over the past year the odds were 83% that a trade would be successful and the results are very profitable. c. Creating the bollinger bands around the spread between a stock and QQQ. When that spread hits its upper bollinger band (divergence between the stock and QQQ) and the price of the stock hits its lower bollinger band, BUY. Again, the probability for success is very high. A similar approach was developed using VIX instead of QQQ. Good results on Dow 30 stocks. These systems have been banged on and backtested by some of the best developers I've seen. All very creative. My own efforts in this group have been inconsequential by comparison. I believe several of the members of this discussion group have been trading these systems successfully. Including me.
documented. Just go to the top 10 chartscripts section. click on any chartscript. You will get a description + the code + a link to historical performance on Nasd 100, ISDEX, and Dow 30. Also you can test on your own lists of stocks or individual stocks. Enjoy!
>I want to thank those who commented on this topic. Unfortunately, I have not learned a whole lot by the various comments other than each person's subjective opinions about the possible usefullness of day trading systems. I was hoping to use this message thread as a location for people to contribute real mathematical formula and setups to be used in developing an effective day trading system. If people are willing, I think we could all benefit from those types of posts. I'll stand by my original response...I have yet to see a pure mathematical or analytical "system" that works consistently, so that may be why you have not seen anything posted. Although somewhat subjective, my "opinion" is based on quite a number of years in the industry and hundreds of "systems." But, I will always make room for those who want to put into practice their vision (and, yes, we have "operator assisted" programming that has been very successful --- but we will not see those numbers until that game is over).
http://www.oddballsystems.com/ Mark Brown fully disclosed this system in Active Trader Dec. 2000 issue. Allegedly used by a 100 million dollar hedge fund from 1992 - 1997. System uses the intermarket dependency between the NYSE advancing issues and the S&P futures. The system isn't a pure "daytrading" system, however it does trade intraday.
I stand by my original response : there are many profitable systems, from scalping to long term core trading. It is safe to say no system work all the time too and they are ways to deal with it. There is no question you can model a trader's process. There is nothing paranormal about someone trading. Yes, just like in any line of work, you will find traders so good, they are on another level of trading. These ones are hard and possibly impossible to model. but hey.. there are not that many and you don't need to be like them to make money. Instead of searching for the holy grail, the perfect solution and 100% accuracy method of trading, go for reasonable risk reward. This is a game of probabilities. And no it is not random (if it was, please explain me how a human would be better, if that's your opinion, than a machine at guessing a random output.. I just mention that because I remember some threads about it. people claiming it was random and still thinking they can trade in that case and not a machine.. that's logical!). Everytime the systems topic come out, it is always the same story : discretionnary against system traders. This show little understanding of what this is about. Any trader with a method, I respect. And you know what ? I have modeled several of them too. Oh yeah.. and do you know what they are good at ? not following their method. Again, there are mechanical systems (not only system trading, but automated on top of that!) which make money. Not all the time.. when the system does not work anymore stop using it ! I mean, if a discretionnary trader's method does not work anymore I don't think he/she will trade forever either. So why do you want system traders to do that ? System trading is not about having something which works all the time with all kind of markets and forever. It is about MAKING MONEY ! This argument could be compared to the debate man vs machine in the industrial world. All these workers were against the machines (of course) pretending people are better. The question is not who is better though. Sure some craftmen are so much better than their machine counter part. So what ?! I am sure your car was only built by people, right ? I know who won industrial argument. It is the same here. The debate is at the wrong place. It is not about people better machines. it is about machines helping people do much more. For example, instead of trading (badly) one market, I can trade multiple markets because the machine do it for me (well). Light bulb anyone ? tntneo
The S&P relationship is primary, and so very basic to any intraday entry/exit strategy that we sometimes forget to mention it (except of course during "newbie" training). It is absolutely essential that you keep a "spoo" chart with FV premium/discount window next to the tick chart. This is the "window" to the market (and we even use the audio squawk box to give us a few second lead time). And, don't forget that the market had quite a run during the time period mentioned (not to take anything away from the results), but even the most basic of trading "systems" that were biased somewhat by the "buy first" strategy did pretty well.