You are making a negative statement (logically speaking) and to prove you wrong, it is enough to show just 1 such system's existence. I have 2 completely different systems in testphase right now, and manually they are giving very good results. Only reason I haven't automated them because I can't program, but looking for a programmer...
You're right, scaling up is absolutely a problem, and "forever" is generally measured in days or weeks these days (in my history, anyway)... Don't get me wrong, I'm all for my traders working hard to improve their results, and I even use an automated program for my opening only orders....but by automating something that has worked for 20 years (for me anyway, probably much longer), I have the ability to "scale up" with more stocks, layers, etc. But, this automated program has no guarantees because it requires homework, modifications, and manual review every morning for news, gov't reports, beta's, earnings, and all the rest of the things that we have to go through before the opening daily. (No argument on my end, it's just that I consider trading to be "simple" - not "easy" but simple enough that most programs don't add much to the mix in the long run. All the best, Don
Please keep us informed of your progress...as they say "you just never know"....My comment above about the simplicity of trading is what I'm really trying to convey... why complicate it....that's all.... I have reviewed hundreds of systems over the years...and there is generally some aspect that is being overlooked or misunderstood... Good luck with your venture... Don
I can't prove a negative, but I'm willing to review tax returns... so I'm just speaking from personal experience.... My friend, Blair Hull (Hull Trading, market wizards, etc.) had some of the best option trading software in the world, and put people in most of the trading pits worldwide, but even with all that he had to use the human factor, albeit connected by hand held computers. I don't want to upset anyone, but stock brokers and engineers (even software engineers) are the hardest groups to teach trading at this level....however, the engineers are usually much smarter and "come around" much sooner, LOL. All the best, Don
I guess Jim Simons at Renaissance must be trading manually his $6 billion dollar fund earning 50% a year just because Don can't grasp the fact that systems make money.
They're good alright, especially those astrophysicists....LOL. I don't think I would call this an "automated system" since it takes 60 researchers to make it work....but, I agree their results are excellent to date....Trading groups make money, computers help, no question about that. --------------------------------------------------- Its $5 billion Medallion Fund has averaged 35% annual returns Scientifically based investment strategy For over two decades, Renaissance has been at the forefront of research in mathematics and economic analysis. Renaissance employs more than 60 top scientific specialists, including mathematicians, physicists, astrophysicists and statisticians, who review market data to find statistical relationships that predict the price movements of commodities, currencies and stocks. -------------------------------------------------- Don
It usually takes a thread about 10 pages to process the "usual" responses. Your persistance in defining and redefining the questions is very important. The fund you are working with has also given you a great deal of information which to consider. Primarily, they are telling you that they do not have the capability to work on the problem. Too bad. You have drawn some credible people into the thread and it valuable to see how the array of participants built the consensus you now articulate. Now there is a possibility that your attention will shift to the place where the pay dirt resides. In your realm it is extremely important to make and keep a log of failure and to profit from knowing where things do not work. On the other hand it is important to recognize that other's failures may not be proof that something does not work. The fund you work with is an example of a particular staff that has not been able to function in the realm of the problem. To eliminate their "fishing" you will have to construct the scope and bounds of the problem solution. So far you have taken one step in that direction. That is, you have stepped away from particular limited non solutions to making a first effort at giving the problem a wholistic solution consideration. You are very unfamiliar with what is required to have or begin to study and design an approach that is wholistic. Shooting the messenger is ordinarily what is done at this point. Suppose, for a moment considering what a person would do to inform you. How would someone who is where you want to be suggest to you as to how to get to the place that is familiar and of high utility. For ten pages you can notice that this is not happening and no one is serving as a guide for you. To get to where you need to be is not done with the two questions that you are asking. (I just skipped the consideration of the one Q you added to your quest irrationally). You need to move across the line in the sand from discussing failure and to go to begin to discuss success. This means that your last question has to be addressed differently as well. It is more than a case of leaving "edges" behind (even acrary's fine example centering around mutuals) and going to partial market considerations (trends). You must go to the roots of a whole consideration of what is going on; how it works; and why it works. Commonly, this is done in most endeavors by measurement. The financial industry has failed to do this by putting forth an off the mark effort to build money making systems where they put raw data into a box and design the box to make money. It is not an easy task to explain why this approach doesn't generally work. Choosing the goal to make money defeats the process of designing the box. Don Bright is eloquent in his descriptions of this industry wide failure as he sees it. I commented elsewhere, recently, on some measurement inquiry. As did acrary to you, I gave an example. By focusing on the end goal and by using raw data as the starting point, the what, how and why the MARKET works, gets left out of the design of the box. What you may design will come from very different considerations than those with which now, you are familiar. Occasionally, individuals get to the point where they are capable of making observations. Observations can lead to classifications and descriptive formulations of phenomena. By knowing how to know, a person can go far in eliminating the spurious. My views are based upon delving into the matters at hand and, over time, reaching conclusions based upon the truths of how the market works. In public, the assessor level, the beginner level, and some intertmediate level stuff is at play. But, so far, not much beyond that. One exception is the YM/ES leading/lagging relationship that is now being used operationally. I regard this as intermediate plus functionality. To design and build what you want to supply takes the coding of several (many many in fact) items arranged in an array that involves seven processes which feed each other in an arrangement of links and nodes. I gave two examples of nodes in a post recently: sentiment and pace; six levels of one and five levels of the other. All eleven signals of their presence or absence (think Boolean) serve to "gate" other information to other nodes. All of this is secondary in nature and the primary considerations are more complex. Your second question is really the "gating" question in disguise. I threw it away since it is the "too late" version of thinking about your problem. I engaged with Scientist a couple of years back. The movement of the market operating point was the idea that I introduced. It is the basis for proving that macro analysis and decision making is off the mark. I used square cells to make my point and he countered with diamonds. We both agreed that or "rolling" the matrix forward, there were advantages of diamonds. Think of how steel culverts of various lengths are "rolled" from strip steel. The main point he came to understand was that the operating point of the market "migrates" and does not jump around all over the map. Further, the migration occurs as a consequence of all other avenues being blocked. (A logic construct that is not common in code writing). Here, you have your thread. And you have your questions. You are going to have to make a "sea change" in thinking to be able to begin to address the opportunity that you are sitting upon. My path which is just crossing yours for a moment in time is to give away all that I have spent my time observing, measuring, formulating, designing, packaging and successfully operating. I reduced to paper a meachanical version of what you are looking for at the capitalization levels of your modus well before eminis were invented. So last week I gave away my original copies (think of it as a wall of charts that resemble connecting up the flight charts from Florida to Maine) to show just the functional nodes and links in enough detail for coding. I also recorded the camtasias that are required to update things to the current levels of electronic data availability. In them you can see the immediate future of the market annotated (20 five minute bars ahead) and the market, in time filling the ranges as the discussion came down to the detail of tick charts and the DOM. What I feel is the answer to the problem is knowing all the coefficients of all the functions on seven sequential levels of consideration. If three differing periods of regression are needed to define sentiment; I noted each regression duration and the exacting gating of the boolean values through to the primary comparitor (one of many) that provided the NOW decision result required at that moment. The camtasias show the operations at the tick pair level for about two hours where the annotaed future moved into the present as set up about twenty bars in advance. This means that trading takes place by just giving up the spread at turning points in the market price. Obviously this is not a price only based system. My observation of your dialogue is that you are tending to deal with price primarily. The ratio of market sentiment cyclic change to price cyclic change is four to one it turns out. The ratio of price acceleration/deceleration to sentiment cyclic change is one to two, respectively. I expect to have most of my stuff public within 2 to 6 months as shareware to those committed to helping others and solving community problems with their newfound increased performance. Taker types will not be involved. This stuff works on all fractals.
Hi Edgar, It includes trendlines and intraday adjustments which is better done manually (more efficiently and accurately).
Don there are automatic trading system constantly in place, otherwise markets would not be as efficient as they are. that does not mean that there is opportunity for the average guy, but saying it does not exist is alomst naive. just look at index and futures, i doubt they are kept in line by discretionaries only. different exchanges trading the same instruments, different exchanges in different currencies, option markets you name it ... there is constantly arbitrage happening, it is just not accessible with our normal fee/speed/knowhow setup. yes, the best single market pattern recognition is in your head, but just try for once to watch five 125.000 stock pairs and trade five hundred of these twice a day ... our mind is perfect on selective, yet fails on mass data ... strange, would not have expected this line of argument by someone with your reputation ... yet, peace