What do you consider noise?

Discussion in 'Technical Analysis' started by mcteague, May 4, 2012.

  1. mcteague


    Being new to trading, I am fascinated by the observation that watching a chart in different time frames often shows a completely different story. It almost seems that you could trade daily, hourly, or by 5 minute charts in totally different ways and still be correct.

    Although It seems valuable to look at a chart in different time frames, some questions come up. If you trade on closing prices for example looking at faster time frames might give you additional insight about the mood of the market. But you might see lots of fluctuation that means little to you. What is sometimes called noise. And if you are a faster trader, say based on 15 minute charts, you might want to look at daily and even weekly ones to gauge the overall trend. But paying to much attention to the overall trend might make you miss opportunities that are occurring during the time frame you actually trade in.

    So my basic questions are: How many different time frames do you look at a position from? And, How do you distinguish noise (or irrelevant information) from things that are or could be valuable. Thanks
  2. Brass


    Anything that annoys me I consider to be noise. Of course, one man's noise can be another man's music.
  3. +1. In other words, there is no noise, there is only data.
  4. I agree with it.

    The more you try to figure out the answer you seek, the crazier you will become, and you will be chasing your tail. remeber einstien says: "everything is relative." It's one of the few absolute truths.

    What it comes down to is coming up with a plan, having rules for the way you draw your trendlines, or having the same set of time frames you look at, every time.

    For me, noise is when price swings are stagnant. Since I need to capture a price trend, if price swings overlap, then I could trade the range (buy bottom, sell top), but I much prefer waiting for price trends that aren't overlapping recent previous ones.

    Someone else might define 5 min swings as noise, but the 5 min trader might consider the 1 min swings as noise. as I said, all einstien's theory. For me, I don't count 1 bar as part of a price swing, it is noise. But remember, that 1 bar can be subdivided and therefore it's really not just 1 bar. (i'm going crazy mysefl)

    Just do the same thing all the time, define your edge 100% so there is no question whether you enter/exit or not, and have positive expectancy. Net a few points/day, then scale. E-Z.

    By the way, I think 20 trade is good as Mark Douglas says in his book. THe fool think that his trade model over 20 years is better then last 20 trade. I consider 20 year data noise, while other think 20 year trades is music.
  5. To me, this is kind of like asking "What is outside the universe?" Nothing.
  6. All of the abstract answers aside, from previous posters, in statistical terms the markets exhibit much more noise in lower time frames, ie 5 and 15 minute bars, compared to hourly and daily bars. I consider noise where the properties of the price chart (frequency, amplitude, trend sustainability, etc) exhibit values close to those of a random walk - in terms of distribution and value ranges.

    If you think about it, it's unlikely that during 30 1-minute bars you will see anything other than noise generated by people hedging, entering/exiting long held positions, a few speculators looking at completely different time frames etc. So it's intuitive that lower time-frames will be filled with such noise and the chart would exhibit descriptive statistics that are close to a random walk. This has been discussed previously in many threads, and it often leads to heated debate which fizzles out due to nobody being prepared to shovel out tomes of statistical testing to back up arguments. Also, "noise" is different on different markets, so some properties may keep while others exhibit normal distributions in the short term.

    Basically, I consider everything noise except for the market events that generate the "fat tails" of the almost-normal-distribution which make it non-random only on occasions. Those fat tails are generated by news and other inflow of information, as well as fundamental speculating for longer term trend establishments, and a bunch of other things. This is surely not noise, but in order to anticipate it, you either need insider info or a time machine. So it's best to consider everything noise.
  7. Brass


  8. Well at least your response is congruent with what you would expect to come from a person living in a world in which everything is noise.
  9. ==========
    Best examole of noise is a 5 minute chart;
    its too long for the few profitable day traders.

    My favorite swing position trade/invest time frames include monthly weekly, daily;
    i dont mind entering on a 5 minute chart, exit on that one,
    but bid/ask is more important than 5 min ''NOISE''
  10. Living in a noisy world has it's benefits, like extracting profits from a quasi-random walk chart. Nothing wrong with noise, imho.
    #10     May 4, 2012