On most fractals you will be bridging. If a person is modeling, there is also a requirement to use information from the markets. This, especially, applied to things related to making money. you are making up two things which are the responsibility of the market to give to you. Momentum (p) is not related to price. The slope (m) of price is the trend no matter how you chose to determine it. You can take dm/dt if you want but remember it is not momentum; this is called price acceleration. m is called price velocity. All of this is a function of using standard time periods. the rate of change of angles as you use them in conjunction with a linear variation in position is trite. Especially trite. Check a recent Rahula thread where he made the same mistake in his OP re: p.
Assume my price curve is my simple moving average -- this is a position curve. Assume my trend (velocity) is the first derivative of this curve. Given that mass is nonsensical when discussing 'price', we simply assume unit mass (m=1). Momentum therefore is: p = mv = v To find the magnitude of momentum, we have to find the magnitude of velocity. Velocity, in our case, is a 1-dimensional vector: st (where s is our slope). Therefore, the magnitude of this vector would be sqrt((st)^2) -- or just |s|t. And what is s? Oh, right ... our 'price acceleration' -- and the second derivative of our price curve. And what? It is directly related to our momentum magnitude? Odd. While in multi-dimensional physics, talking about acceleration and momentum doesn't make sense, in this trivial one dimensional case, it actually does. Considering we already know the direction, my discussion of momentum was strictly about its magnitude. Given that I could either refer to it by an arbitrary value or by angle, I chose to use the angle. As for your comment on my system being 'trite' and damnit's comment that my system is too simple for the league we are in -- perhaps you boys could provide some constructive criticism instead of wasting your time simply being negative. As for calling me 'boy,' damnit -- I prefer not to be patronized. If you have concrete evidence of the futility of my method, then by all means, post away. However, if you are going to be disrespectful, I will have to politely ask you to refrain from posting in my thread. Thanks.
May I ask what you are basing the moving averages on? What kind of price groupings (time, tick, volume bars basically)? That is going to make a huge difference in this method. Just as an aside, if I were putting this together, I would use constant volume bars because they tie volume to price, which is important, and they don't print very long bars. Look at a chart of yesterday (or today for that matter) for why I say this. Think about this: sometimes, things don't go parabolic on time bars. Sometimes, they jump or otherwise go batshit. Something to think about. EDIT: just caught it. You are using 1 minute bars. The price curve will sometimes not be a curve, unfortunately. You might want to take this into account.
I'm glad to respond. Most people use price displacement, velocity and acceleration. When you added in a lookup table to relate contracts as c=15 time x contracts plus b, where b is 15. I felt you were using a two dimensional system invloving mass. Mass was represented by its usual trading variable, namely volume. My mistakes. I felt suggesting you were bridging was a constructive comment. I felt it was constructive to suggest to you that the elements you introduced to cause the bridging would have been obtained from the markets in order to preclude bridging. As a constructive comment I suggest, further, that you use math in support of making money instead of using what you know in math and seeing if it allows you to take money out of the markets. Price change occurs, in my opinion, in response to market forces which "show". Gathering the raw data which exhibits and reports to you the forces, either kinetic or potential, will better support your effort to make money from the markets. You may find that a two dimensional system exceeds a one dimensional system for describing he systems. I am wedded to an input of five degrees of freedom and they grow to about 70, then by proper use, 5 outputs are produced. You have an entry and exit system based upon one so far. And it based on an arbitrary single equation that relates measured angles to contract number. the market is providing you with the angles as a consequence of fixed, arbitrary values you chose for a fractal you chose and this precipitates bridging of price movement because of your contributions. I didn't cause the bridging but I recognize it through one thing or another. you may think I am being derogoratory since I read, that in what you say to me. What actually may have happened, is that your cause may have been advanced in a couple of ways. I could post a list of other market considerations that you must, by all means, take into account. I also feel that self discovery is a really terrific experience for one who is working hard and purposefully. So I will only suggest one. You need to formula some rendition of volatility which happens to be a nice connector between the two nodes: price and volume. Another modeler other than you, this week, noted a real long list of pattern type nodes. He was embarking on coding them up with an as yet unchosen software programming language. A very tough situation for him. the crux of it was that he would be correcting coding errors that would create other coding errors as yet unseen. In a very different way that is where you are headed. By mentioning bridging and, now, volatility to you, I hope to head you off as early as possible. The momentum thing was just there to connect you to a second variable and you came acrosswith mass in a new york minute to show me your work at the level it was (I hope) was another intentional oversimplification that prevents it from working. I am compromising at this point; you win on the "leaving your thread alone" from this point on. Damit sends me pm saying I am a POS so you are in good company. LOL.
xiaodre -- Definitely food for thought, though volume makes the problem two dimensional. In my extremely humble and inexperienced opinion, I have found that generally, volume information can generally be extrapolated by examining the rate of change of price. cd23 -- I cannot thank you enough for your post. Rereading mine, I certainly came off a bit abrasive, so your post was definitely unnecessary on your part. Having said that, I really do appreciate your insight. As I said, my history is with mechanical investing systems, not trading systems. My limited paper-trading experience has lead me to use this type of system -- and perhaps through overly judicious use of Occham's razor, I over simplified what I perceived the patterns I was recognizing to be. While I may seeing things on my charts, perhaps they are spurious correlations to the underlying factors. As I just said, I felt that volume specifically was manifested in the rate of change of the price curve -- but using it as 'mass' seems like a really unique concept. I know you said you wouldn't respond to this thread again, but can I ask what you mean by 'bridging'? I am unfamiliar with this term. Hook -- Yes, this system was intended to be a day trading system only. Thanks for the responses!
I think to wait for a minute for the bar to form, it may be too late in a panicked market. That's why I advocate volume bars, if you aren't going to study volume alone. A panicked YM market will move anywhere from 50 to 130 points in a minute. Look at today's action. The swings after 13471 was broken were intense. The price becomes not curvelike. Minute bars don't print fast enough. Now, if you use a low tick chart, you might get a smoother curve, but the trade decisions would come so damn fast that you would need to be a long time player with tons of experience to parse this info. I've watched tick charts and they make beautiful fluid curves (except the ER2 which frigging gaps down in the middle of RTH), but they are too fast for me. I personally think you are better off looking for a system that's alot simpler, but good luck. EDIT: oh, and one last point: you cannot extrapolate volume from price movement. Case in point: large price movements during low volume periods. Like, this morning or any other AH trading moves. You seem to be assuming many things that may really hurt you. Please try this out on the simulator before you go live with it.
If you were to run your original system in simulation, with realistic fills, I can pretty much guarantee you the slippage would wipe your account out, I believe this is what damnit was referring to. It may be better to have only 2 or 3 "scale-ins", and base them on price action rather than the slope of the displaced moving averaged vs. its original. However, this is an interesting idea to be used as: 1. A solo entry signal with multiple scale-ins according to price continued moving in the direction of the primary trend, and 2. A multi-scale out exit signal as price begins to reach its climax. Play around with the slope geometries and see what works best for your needs. *** The technique of using the slope value to fade the trend requires an extraordinary amount of knowledge and experience to execute successfully, consistently. As fading the current trend will yield the most profit if done correctly, and it will wipe your account out if done incorrectly. Tread carefully on that one. It would probably be better to use the highest increment slope to begin to scale-out of a position which you have entered successfully, in the direction of the primary trend ... and just patiently look for another setup in another tradeable indice, once again, in the direction of the primary trend. *** As mentioned previously, constant volume bars would most likely give you far superior results to minute bars, as time isn't important when determining price. *** Good trading, JJ P.S. If you do some work and search around for what you can find regarding Turtle Traders I am sure you will find much food for thought. P.P.S. Anyone is free to disagree with anything I've just wrote.
I really can't thank you guys enough. I think you are certainly right that a constant-time bar will not suffice for a mechanical trading system. However, I am currently trying to figure out if it makes more sense to use constant-tick, constant-volume, or constant-range. Constant-tick and constant-volume bars seem to make the most sense as far as information delivered. My next question would simply be what range, tick-scale, or volume amount to choose For the moment, I am trying to find 10 years or so of historic level 2 quotes. Being a programmer, I prefer to hand code my system tests -- I find that it allows me to have much more flexibility in my system. I found a pretty cheap data-set for s&p eminis for the last 10 years (only $30!) -- but it is minute bars, which means it would be impossible to create volume-bars, range-bars, or tick-bars. Nobody seems to offer historic level 2 quotes, unfortunately. Path to self-discovery, here I come!
Bridging. You said in the OP: "The system is based on the theory of trend (first derivative of the price curve) and momentum (second derivative of the price curve). For my system, I use a 1 minute chart and a 50 period moving average to represent my price curve. At any given moment, I look at the direction of the 50 period moving average to get the current trend. To find momentum, I take the current value of the MA and compare it to the value of the MA 13 periods ago (~50/4)." People who trand trade to take money out of the markets daily see 50 minutes as a long time in comparison the the profit segments they take. For example, use a 5 minute chart or your oneminute chart and sketch in these profit segments for a day. Notice that in pairs they make cyclical waves across the day. Take a 50 minute segment and only use that to connect points at each end of the 50 minute segment. No matter how you do it, this segment is not representative of the profit making segments drawn before. Many trades are occurring under a line that is 50 minutes long. If the 50 was a boat and it was floating on the waves of the market, then it would be bridging the money making pairs of each of the waves the boat is floating upon. To make money, it is necessary to use each price movement to the fullest extent. To state this simply and elegantly, in the limit of extracting money from the market, an exit is also simultaneously the next entry for making the next full segment of money. Your relatively long line (50 minutes long) floats (bridges many of these moments of simultaneous exit/entry and, therefore misses many trading cycles. The daily limit of extraction, in its limit approximates a gross of seven times the H-L range and six times the net. I shoot for 50% of the 6 times daily h-l as the practical performance level. Testing systems happens on a fairly fast data set level relative to human sensory levels and rates of perception. A one minute bar is not in the ball park at all. use the smallest tick range available for achieving sensible pattern recongnition otherwise you are bridging the data that others are using to make money. I didn't recognize that my vocabulary was inappropriate. If you can understand what I mean be bridging, then most of my prior comments will begin to have meaning.