At the risk of sounding like an ad for a tampon comercial ........yes I use a one size fits all!!! But first let me explain. My "system" is programed with several parameters that utilize 3 different timeframes! I'm always in the market per say but not necessarily getting filled every day. For the most part I almost always have a signal almost every day. When you get a chance please feel free to read some of the FORECAST in my T-REX Journal. I have posted a set of rules & setups that I look for. I know that I sound like a lunatic ..........maybe I am! but I really believe that I have the closest thing to the holy grail that your going to get. Now I'm not stupid enough to claim that I have created the perfect system but I have a VERY VERY VERY GOOD system that is extremely profitable and down right scary when it comes to FORECASTING price movement. Today for example my system bought near the bottom of the bar in both the NQ & ES. Several times in the last 6 weeks this phenominon has taken place in the FORECAST. However, this is common place for the SWING VIX Indicator that I created. I hope that answers your question some what.
..sorry, I was not aware of your thread. I only post when the meds aren't working. Eventually they find something new that calms me down and I fade away again. I shall check out your Cassandra-like forecasts (you may recall that recent geo-pharmaco-archeological research has proved that the vestal virgins were high on light hydrocarbon fumes). Help a senior citizen, here, please. Your holding period is days? Why are you on that fractal? (FYI, that is a mathemagical concept, not a pharmaceutical). I myself crave the intraday fractal, so many more hits per day (it's a pharma half-life thing). My point is (where was I?) that if your system is shit hot in all time frames, it will make more money in the shortest possible time frame which still makes money with your fixed expenses. I am consoled that your simple system is in fact complex. In all seriousness, is it such that you can code it, or, like most really fab systems, it there an ineffable intuitive element to it which makes it uncodable? For my education, how many parms does it take to make a forecast (equivalently, how many rules are there)? As an empirical theoretician of system development, this is a question of cosmological, nay gynecological, significance to me. Please note that I am not asking anything proprietary. Think of me as Harry on 200mg of lithium carbonate. Best regards. - Mike
Best of timing methods are not possible to backtest . Timing methods which are possible to backtest are usually based on a fixed ratios, which work just slightly better then random . Walter
...could you please elaborate on your post? I would like to understand the error in my thinking about systems and backtesting. If it's a system, it has rules. If it has rules, you can write algorithms to represent them. If you have algorithms, you can backtest. If you are referring to an intuitive system, I can understand that. I haven't had much success coding them up either! Thanks. - Mike
Simplest Timing device is a triangle. Connect highs and lows with trendlines and they will eventually intersect somewhere in the future. Can you program that ? triangle alone is not profitable, so you can add a divergence of some oscillator to the mix. Then you will find out that only divergence of certain magnitude will do, then you will need to use certain magnitude of a price swing only to eliminate noise. And this is only about 30% of what you have to consider for a timing system. If you can code all of this and test it, you are damn good. Walter
Regarding the timeframe questions, I just found a way to program my favorite software to have bar interval as an input, so basically a system can be optimized over bar interval. Don't quite have the code done yet but should be interesting. I have found a system that seems to work from intraday bars so short that there is barely liquidity to support analysis all the way out to month bars, fairly good moves too. As far as "every system should do that" I can't say, I would say "every system should make $", I wouldn't want any more rigid rules imposed than already are.
...thanks for that clear explanation. I understand what you mean now. I am currently working on generic line-generating and -projecting code for use in the kind of approach you describe, but it is slow going. It looks so simple on the chart but the logic is hard for an old man to code. My guess is that when it's done it may be too slow for the split second calls I think are needed for intraday. Best regards. - Mike
EOD data is not a very fruitful place for making money. It is very commonly used by people though. I believe this common usage is a function of tradition and convenience. And the fact that most people have jobs or something like that. They scarf up what is there and use it. To make money you must go to where the opportunities lie. What apparently has happened, is that the literature has really built up and multiplied based on EOD. The values, OHLC, also are emphasized for some reason. I think it is fairly contagious at this point. When you look at the maths applied to the data, you also get to see how people mostly take what they know and read about and use that stuff for convenience. On the other hand, a person can start from scratch and figure it out. Timing is extremely important for making money but few approaches start with that as a basic idea and then work in support of enhancing timing. The second element of making money is change over time. It is a necessary ingredient. Working on understanding change is very difficult with the EOD and OHLC constraints. Turning to looking at the characteristics of change really creates a great ball park in which to play. Most people do not look at the whole picture. Any system is usually named for the piece it focuses upon. Anyway, that is why I see that you have not had a successful experience with timing. I focused upon timing from the start (1957). Nailing down the right time to begin to make money allowed me to build my approach around that and the one other facet: when does a money making trend end? Getting these two things straight completed my efforts in making sure I had a way to make money. I have used this stuff ever since. I cull stocks that are repeatable and reliable. Once I knew the exact events that preceded and started trends, I just sorted out the stocks that do that. The trend endings come as an equally simple sequence of events. As they signal their arrival one after another, I exit as the best price is reached. You probably noticed there is no OHLC and EOD ness in this. So you do not see timing for making money in those values. Indexes are less complex to deal with because they trade in a zero sum game and they are bilateral. Timing is reduced to just one facet as a consequence. You are not doing cycles of profits as in equities. You are just continuing to extracting profits from the progress of the zero sum game. Timing action comes down to one element of consideration. All you do is continually participate by being a winner all the time. Timing on this matter only focuses on where you stand relative to the opponent. It is a game where it is always your turn and never his. I express it in market terms as "always being on the right side of the trade". I monitor and make that determination as required. The timing I focus on is related to the expression: "if nothing is wrong, do not fix it". All i have to do then is have a short list of stuff to look for. This list comes from a very easy source. It's basis is really complex, mathematical and theory based. No one here has explored any of that territory except for one person who is much smarter than me. The pragmatic substitute is easily carried out and I would guess about 30% of the people here do that. For them it evolves from their trading logs the hard way. People write in logs what they did. Too bad. It is what you didn't do that you learn from more often. Writing down what you didn't do will greatly accelerate getting rich. The shortcut to making all the money a trend has to offer is found in knowing what stopped it. A trend will continue until it is stopped. That sentence is the KISS of bilateral trading in a zero sum game. All the theory will never be understood anywhere most likely. I like having it down but it is not important. What stops trends is a very short list. There are no surprises either. No suddenness either. A sequential ending occurs where it adds up to not continuing. I go along and finish each trend and then do as the market dictates. I follow along always staying on the right side of the trade. There is no one, hardly that does not know when they are on the wrong side of a trade. Most people can "feel" being on the right side. Almost everyone leaves early and knows it also. The less time a person stays in a trade is a measure of "fear". we know all this stuff. If you changed your log entries from what you did to what you didn't do, you would quickly find a list of why trends end. It is also the same list of what happens before a new trend begins. No EOD, OHLC data showing up here. What you use to back test is data that is not used to make money. What you use to make money is the facts you need to know to "stay on the right side of the trade". These facts are simple things. They are Boolean statements. I never use more than one at a time. The most I use is a pair. the pair have wording that change a little depending on one of three market sentiments. Naturally all of this is programable and very very accurate for optimizing making money. In a zero sum game there is only one thing on the table. As you keep yourself on the right side of the market, you take profits each time you are required to take an action to adjust which side of the market you are on. QED.
There are no needs for split second calls. To make money, the potential is offered by the market and the trader extracts it according to his efficiency. All you have to to is match your eficiency with the markets offerings to you. The calls are easily made by software on both side of the situation. The thing that is really on the table is understanding that both the market and the trader never have the capability to do anything suddenly or by jumping around. When you see either of those items in your view, you are simply looking in the wrong place to be operating. Think of the planes of Excel spread sheets; if you are getting surprises or seeing things jumping around, it is simply that you are on the wrong plane for the moment. When you get to where the right place is, the Q has either a yes or no answer. And it is the same Q too.
You have to test all components separately and when they do 70% or better, you put them together and test by simulated trading . You do not need split second calls, if your timing system is good, you know, I mean, you really know when you will be trading, ahead of time . Walter