Predicting randomness
- Thread starter oddiduro
- Start date
That does not invalidate that approach. Topology has the same "descriptiveness" feel to it in relation to other mathematics that are more constructive. But that description often gets us very valuable information. For example, there is a theorem in topology that says that no matter how you comb a hairy ball there will always be a colic [not all the hairs will lay flat] (that is stated in english; obviously there are rigorous mathematical definitions to all those pictures, i.e., there does not exist an everywhere nonzero tangent vector field on the 2-sphere S^2.) This has great applications. For example, it tells us that some where on the surface of the earth there must be a spot where the wind [horizontal wind velocity] is not blowing.
This is an old philosophical debate that goes back at least one century, where some mathematicians declared that unless you could construct an object, it did not exist and doing math in this [non-constructivist] way was mental masturbation (formally they wanted to forbid the use of Axiom of Choice in proofs.) No modern mathematician worries too much about this today.
nitro
very nice, odd. can you PM me or post how to access the original paper??
thanks,
I am sympathetic to the fact that we all start from a position of ignorance.
The idea that trend does not exist is incorrect. One can see trend by simply taking a chart, posting it on a wall, and asking a nine year old child to tell you whether the line is moving up, down or straight across.
The idea that trend "happens" in hindsight is both incorrect, and not useful to a professional. I know that trending behavior will occur at some point. I know from experience that it is possible to predict approximately when trending will occur (sorry if you are having trouble with that one). After that my interest is limited.
I see the some here (on ET) believe differently. That is probably why they have difficulty making money. As it turns out, the difference between a professional who makes consistently good money, and the retail trader who consistently losses it, is the ability to do things that most people believe can't be done.
It is (in my opinion) very similar to when folks believed that no one would ever run a mile in less than 4 minutes. Today the world record is 3 min 43 seconds, and many high school athletes have broken 4 minutes.
I wish you well, and hope you learn to overcome your preconceived limitations. In the meantime, I make my living anticipating and betting on occurrences of trending behavior.
Steve
The idea that trend does not exist is incorrect. One can see trend by simply taking a chart, posting it on a wall, and asking a nine year old child to tell you whether the line is moving up, down or straight across.
The idea that trend "happens" in hindsight is both incorrect, and not useful to a professional. I know that trending behavior will occur at some point. I know from experience that it is possible to predict approximately when trending will occur (sorry if you are having trouble with that one). After that my interest is limited.
I see the some here (on ET) believe differently. That is probably why they have difficulty making money. As it turns out, the difference between a professional who makes consistently good money, and the retail trader who consistently losses it, is the ability to do things that most people believe can't be done.
It is (in my opinion) very similar to when folks believed that no one would ever run a mile in less than 4 minutes. Today the world record is 3 min 43 seconds, and many high school athletes have broken 4 minutes.
I wish you well, and hope you learn to overcome your preconceived limitations. In the meantime, I make my living anticipating and betting on occurrences of trending behavior.
Steve
position of ignorance? please !
here is an objective test that is readily available, understandable to non math minded, and repeatable---from larry connors " how markets really work" chapter 3
multiple higher highs ( trend ) and multiple lower lows ( trend ) are generally accepted as strong or conversely weak markets.
testing was conducted from 1989 to 2003 over a 15 year period on the SP and NDX.
1. the market lost money within one week after 3 or more consecutive days of higher highs.
2. mulitple days of lower lows outperformed the average daily gain.
3. multiple day lows far outperformed multiple day highs
4. multiple day lows in the nasdaq outperformed multiple day highs.
5. returns increased following consecutive days of market declines and decreased folowing consecutive days of market gains.
this is basic testing of strategies--- trend followers are starting out behind the eight ball per basic statistical analysis--- not a place any trader wants to be.
if you can provide ANY tests or studies that indicate otherwise, i am open to new ideas.
thank you,
surfer
perhaps another Et'er can explain the point with a little more clarity.
best,
surfer
best,
surfer
Larry Connors is a smart business man. He knows that naive people like you will purchase just about anything given the right pitch. He started his own publishing company "Connors Research Group" and hired a good programmer named Connor Sen. They forwarded a copy to my office shortly after the first print. What you have done is to take a piece of the text and post it here, (without permission) and without any context except the one that serves you best.
Wish I could be gentler with you, but your "ignorance" is of a very special kind, that borders on "stupid". My guess is that you are "stuck on stupid" and maybe a gentle tap on the head is all you need.
Take moment to read the rest of the book. Skip the pictures and try to understand the commentary. I will be here if you need any help with the text.
Steve
Wish I could be gentler with you, but your "ignorance" is of a very special kind, that borders on "stupid". My guess is that you are "stuck on stupid" and maybe a gentle tap on the head is all you need.
Take moment to read the rest of the book. Skip the pictures and try to understand the commentary. I will be here if you need any help with the text.
Steve
you are funny!!
are these tests not accurate? when i repeat them, i get the same results.
I doubt that you find me funny. I don't contest the work, only the significance you give it.
Each market has its character. I specialize in the S&P. If you look at the charts on one time period, say 5 minute candles, you see a intraday market that cycles back and forth between price points. Retail traders (here on ET) complain that this is the reason why it is hard to make money trading indexes. When I trade this market, I "see it" much differently. Prior to the market, I can identify price points where buying and selling ARE LIKELY to take place. I look to get in (and out) based on these price points. The move from one price point to another requires the market to trend. I (and others I assure you) make a living identifying these price points. For my purposes, all that matters it that the trend I am trying to capture is of a magnitude sufficient to overcome risk.
If for instance I am in a market and am operating with 5 min candles, I may have to risk 1.5 points. If that is true, I need to be able to capture a trend that is at least 3 x 1.5 =4.5 pts. That being the case, I look to see if the cycling activity I see will support that or not. In order to do so, I need to have the tools to find an edge that offers me at least 4.5 points net of expenses, on a regular basis. There are many ways to do this. One approach is to look at "systems" (indicators). Another is to look for temporal windows where the market trends (for at least 4.5 points). Another would be to trade news, events, etc. and finally one could incorporate multiple systems, trading not only a specific setup, but also the failure of that setup. Either way, what we are looking for is trending activity on a micro or macro basis.
The problem with your "interpretation" is that you went right by the most valuable part of the work. It is "as if" it were "hiding in plain sight".
Hope this helps.
Steve
Edit:
I hope that others who read my comments realize that I am in part responding to the first post by Oddi. In his comment he mentioned that it seems impossible to predict the occurence of trending activity. In part, what I am saying Oddi, is that all you have to do is to look at the size of jumps (from one price point to another) during an average intraday session. Import a couple of charts into Excel and change the data from prices to price jumps (one pt, half pt, two pts, etc) and look at the distribution. You will see (eventually) that even markets that look like they are consolidating, actually exhibit "trend". All you have to do is study the distribution of those jumps. If you look at the data long enough you will start to see the patterns "crystalize" in front of you. Its sort of like those dot pictures that you can buy. When you look at them you see nothing, but if you focus just beyond the dots, in just a few seconds you start to see the picture behind the dots.
Each market has its character. I specialize in the S&P. If you look at the charts on one time period, say 5 minute candles, you see a intraday market that cycles back and forth between price points. Retail traders (here on ET) complain that this is the reason why it is hard to make money trading indexes. When I trade this market, I "see it" much differently. Prior to the market, I can identify price points where buying and selling ARE LIKELY to take place. I look to get in (and out) based on these price points. The move from one price point to another requires the market to trend. I (and others I assure you) make a living identifying these price points. For my purposes, all that matters it that the trend I am trying to capture is of a magnitude sufficient to overcome risk.
If for instance I am in a market and am operating with 5 min candles, I may have to risk 1.5 points. If that is true, I need to be able to capture a trend that is at least 3 x 1.5 =4.5 pts. That being the case, I look to see if the cycling activity I see will support that or not. In order to do so, I need to have the tools to find an edge that offers me at least 4.5 points net of expenses, on a regular basis. There are many ways to do this. One approach is to look at "systems" (indicators). Another is to look for temporal windows where the market trends (for at least 4.5 points). Another would be to trade news, events, etc. and finally one could incorporate multiple systems, trading not only a specific setup, but also the failure of that setup. Either way, what we are looking for is trending activity on a micro or macro basis.
The problem with your "interpretation" is that you went right by the most valuable part of the work. It is "as if" it were "hiding in plain sight".
Hope this helps.
Steve
Edit:
I hope that others who read my comments realize that I am in part responding to the first post by Oddi. In his comment he mentioned that it seems impossible to predict the occurence of trending activity. In part, what I am saying Oddi, is that all you have to do is to look at the size of jumps (from one price point to another) during an average intraday session. Import a couple of charts into Excel and change the data from prices to price jumps (one pt, half pt, two pts, etc) and look at the distribution. You will see (eventually) that even markets that look like they are consolidating, actually exhibit "trend". All you have to do is study the distribution of those jumps. If you look at the data long enough you will start to see the patterns "crystalize" in front of you. Its sort of like those dot pictures that you can buy. When you look at them you see nothing, but if you focus just beyond the dots, in just a few seconds you start to see the picture behind the dots.