Vlad has been posting live signals of his <A HREF="http://www.spb.addr.com/index.html" target="_blank">intraday system</A> since October 2000 and has been under the scrutiny of misc.invest.futures on Usenet. His EOD system has been issuing live signals since March 2001. His intraday system with 12 months of live signals has made 627 S&P points, or over $31k profit with a single emini contract. That is over 380% annual return with 2x margin and no losing months. My bad on the calculation- I was calculating his EOD system results rather than his intraday system. His intraday system had 5 down months out of 13, with profits of 576.80 points or $28.8k -a 332% annualized gain on 2x margin. The EOD system, issuing live signals since March 2001, is up 326.8 points. Not bad for 8 months. Gotta go eat turkey but I just wanted to be accurate.
dottom, I am in complete agreement with you, I was attempting to illustrate the very same argument in the post to which you replied. Market "intuition" is a learned skill set culled from experience, not an arcane "sixth sense" emerging from the deep recesses of the mind. It includes pattern recognition and a general understanding of market dynamics that intersect to produce a "gut feel" for price action. But, like anything else, a better understanding of what it is that you "see" in the tape or the chart or the pattern etc, a better understanding of the variables that you heretofore chose intuitively, and a those you chose to ignore, will lead to an even greater set of reliable market skills and decision making. And a better trading outcome. In short, isn't it better to have a greater insight into why you're making those particular trading choices than to grope around in the dark with only vague notions from your "gut" to lead the way? And I couldn't care less if you've been successful leading with your "gut", a dissection of this remarkable "gut" of yours can only bring to light more market 'truths." FIREHAWK
This board is getting deeper by the day. Great discussion. The future of trading, really. I'm a purely descretionary trader and getting more so by the day but I feel that if programmers can beat the world chess champ then everythings up for grabs.
Sabena: Deep Blue computer is better than a human brain in chess. In chess, there're 10^120 variations per game. [ check: http://www.mssm.org/math/vol2/issue1/blue.htm ] How many up and down stock price moves - by, let's say 20 cents - are in a single day on average? Maybe 300. BTW: it's easy to count, you just have to create a point&figure chart with a single box=20c and reversal=1. Number of possible scenarios = 2^300, which is less than 10^120. Computer can make so many statistics about the general nature of price behavior. It doesn't even have to know all possible price scenarios, all possible point&figure charts. Very very good system <b>can</b> be better than any human over the long term, since trading is nothing more than numbers. DT-waw
Candle : If I was a God, I would want to be honored by cute blondes making visits to heaven to meet me in person... I most definitely would not want to be honored by 60 year old grannies singing worse than a baboon's mating call. Neo : Man I just adore British humour..... By the way, "I wish to register a complaint, you sold me a dead parrot!" Alex Cleese
I see where you're trying to go with this DT, but I think the movement of stock prices is a much bigger universe than chess. Chess is a finite system. There is a specific goal and specific utility function that can be applied to every move. The system that affects price movement is much larger and I don't know if it will ever be fully defined. You have many, many factors affecting price movement (from people's emotions to the movement of interest rates to the actions of terrorists, etc.). Price movement is a result of the intersection of these numerous different vectors, but you cannot determine the individual vectors from the intersection alone. Although I a strongly believe in TA, I don't think all data is encapsulated in price activity itself. Take Level II quotes & "order flow" for example. Discretionary traders use this information in addition to the price chart to trade. An advanced trading model would also make use of these inputs (but the difficulty is in building the model!). That is not to say that you cannot create a profitable system based only on price & volume. Most discretionary and system traders do this very thing, e.g. look at the price chart. There is definitely some degree of predictability with just price data alone. I just wanted to point out that you cannot get all available information from the price & volume alone, hence the chess comparison is not quite accurate.
One way to begin would be not to attempt to define the entire market "product set", although that would be a very very interesting endeavor, but to model a particular successful trader(s). Make explicit what it is that he does and in what particular circumstances he does it, and, perhaps even more importantly, in what particular market circumstances he doesn't do it. Obviously this includes shedding light on all his implicit assumptions and market "feelings", this is assurredly the most challenging and largest part of the equation (particularly the more "discretionary" he is). Formulate all this into a rigorous algorithm that defines the essence of his trading "intuition". Just a thought, but I can imagine formulating a distinct product set for each successful trader, then programming them all into smart machine to run them simultaneously against the market. Oh yeah baby! :~D Very very interesting notions, also to see the similarities between each trader's approach, might just learn something more about market behaviour. FIREHAWK
FIREHAWK: but to model a particular successful trader(s) You can break down the various elements of the decision making process a trader goes through before initiating a trade. I've heard discretionary traders say they look for specific "setups" then "pull the trigger" when they "get the right feeling". The specific setup would be a good place to start. If the discretionary trader could model this, they could let the computer analyze all potential setups for them, and they would only have to concentrate on the "right feeling". Would this not bring up more potential opportunities? Tony Oz uses a very specific set of "setups" via his scanner. Once his scanner identifies a potential opportunity he analyzes each opportunity using his discretion. Is Tony not using a systems-based approach for the first part of his decision-making process? Don Bright has a specific set of things that he looks for in setting up a potential pair trade. I'm sure this valuation is done by computer which generates a list of potential opportunities. Don would use his discretion and select the best pairs based on some specific criteria. If you break down each step into its component decision making processes, you can begin to model how you trade. If the "setup" can be modeled, then the criteria to "pull the trigger" is the next step. After that, address criteria used do to exit/add/take partial profits. Surely everyone sees it is better to model than not to model? If you could, in theory, model 100% of your decision making criteria, you could backtest to statistically determine which criteria are good and which are not (assuming the historical data for that input is available). I don't understand why some feel scientific examination is of zero benefit? If the argument is "I don't think you could ever model my discretionary decision making process, therefore the system will not be as successful as me" that does not say anything about systems trading, only that the trader cannot model the system.
Since very few of the more successful traders actually do "screening" or "stock picking" - the set up "model" is a bit different. Most of our traders simply trade a few specific stocks, become "surrogate specialists" in those stocks and use the outside market conditions to help in their decision making process. The pairs are looked at in the same way, with the exception that we will pre-load certain pairs into the "scanner" and have the alerts set to flash when reaching a certain pairing price. Over the last couple of decades we have seen very little continued success with trading via the basics of "stock picking." Looking for the most up %, most down %, mov pr breaks, and all of that is ok, but just ok....since you may find stocks that fit your parameters, but you may not know how they are trading. This is a real problem, since the most important thing about trading a stock intraday is how it, all alone, is acting. Understanding the players involved (short seller, institutional buyer with, or without limits, etc.) is so important even after you find a particular issue. We suggest that new people trade 100 share lots for a week or so, just to get a feel for how the darn thing trades. Every stock has a personality, and it is the trader's job to get to know it. IMO....Good Trading.
Ego. You're saying to a successful trader that they aren't what's important, it's the rules they follow. A real blow to their self-worth. Everybody uses a system, they just don't know it. Just because it seems complex doesn't mean it isn't a system. To say you're not using a system is to say that you do something different on every trade (random). Every trader most likely has a "mental" base system, and then several "if-then" options added to it that make up 99% of their "discretionary" trading. If this were true then the methodology can be modeled and given to a machine to perform and backtest.