Can "Noise" Be Traded?
- Thread starter ktmexc20
- Start date
In an earlier post to this thread, I suggested that "noise" can be a fairly relative term. Perhaps I should have just left it at that. In the final analysis, I think that each of us defines "noise" as the kind of price action that our respective methods cannot trade. I am guessing that no one will agree on a single, absolute (as opposed to relative) definition of noise because none of us trades in exactly the same way.
There are two definitions of market noise - random movements around a mean, or random movements with no mean whatsoever.
If you define noise as totally random movements with no mean, then it is untradeable on a directional basis. However, you may still be able to trade it in other ways. For example, volatility of the market may have a mean. The theoretical fair value between the bid/offer spread may also be a mean with an upper and lower boundary (upper boundary being the offer price, lower boundary the bid price). Thus if you go long vol much below the mean, and short much above, then you have a theoretical edge. If you make markets, being best bid and offer, then you also have a theoretical edge (which is the difference between the bid/offer price and the fair value price somewhere in betwee). Same if you get rebates, commissions, payment for order flow etc.
Anyone familiar with the Central Limit Theorem? If a selection of random outcomes have a mean, then you can predict the average result. A good example is a 6 sided die - the mean is 3.5, so you can "trade" it by buying below 3.5 and selling short above, exiting at 3.5. Despite having no clue what the next 10 rolls are going to bring, you will make a consistent profit.
A trading range, assuming one can predict one in advance, can be traded the same way. If you know the maximum adverse movement beyond the trading range, then you can place your stop beyond that, then enter somewhat inside that price, and exit at the middle of the range. If your estimates for the range envelope are reasonable accurate, and you are sufficiently demanding of a good entry price, then you will make money whilst the trading range stays in existence. You will of course lose money if there is a breakout from the range - but that is what your stop is for.
Obviously the main question is whether you can predict the continuation of a trading range, and the maximum "noise" excursion above or below the range before it qualifies as a breakout. But then again, *all* trading methods suffer from this problem of not being sure if the market state in question will persist.
There is another interesting point. If trends can be predicted in any way, then so can trading ranges - because a range is defined as the absence of a trend, the fact that no trend is being signalled is by itself a signal that a trading range is occuring. Unfortunately we cannot tell the boundaries of the range until it has been in place for some time, enough to produce at least one meaningful high and low. We also do not know the "drift" of a trading range (many ranges have a slight directional bias over time i.e. highs get a bit higher, and lows get a bit higher each time they are made; or vice versa) until at least 2 highs or lows have been made.
If you define noise as totally random movements with no mean, then it is untradeable on a directional basis. However, you may still be able to trade it in other ways. For example, volatility of the market may have a mean. The theoretical fair value between the bid/offer spread may also be a mean with an upper and lower boundary (upper boundary being the offer price, lower boundary the bid price). Thus if you go long vol much below the mean, and short much above, then you have a theoretical edge. If you make markets, being best bid and offer, then you also have a theoretical edge (which is the difference between the bid/offer price and the fair value price somewhere in betwee). Same if you get rebates, commissions, payment for order flow etc.
Anyone familiar with the Central Limit Theorem? If a selection of random outcomes have a mean, then you can predict the average result. A good example is a 6 sided die - the mean is 3.5, so you can "trade" it by buying below 3.5 and selling short above, exiting at 3.5. Despite having no clue what the next 10 rolls are going to bring, you will make a consistent profit.
A trading range, assuming one can predict one in advance, can be traded the same way. If you know the maximum adverse movement beyond the trading range, then you can place your stop beyond that, then enter somewhat inside that price, and exit at the middle of the range. If your estimates for the range envelope are reasonable accurate, and you are sufficiently demanding of a good entry price, then you will make money whilst the trading range stays in existence. You will of course lose money if there is a breakout from the range - but that is what your stop is for.
Obviously the main question is whether you can predict the continuation of a trading range, and the maximum "noise" excursion above or below the range before it qualifies as a breakout. But then again, *all* trading methods suffer from this problem of not being sure if the market state in question will persist.
There is another interesting point. If trends can be predicted in any way, then so can trading ranges - because a range is defined as the absence of a trend, the fact that no trend is being signalled is by itself a signal that a trading range is occuring. Unfortunately we cannot tell the boundaries of the range until it has been in place for some time, enough to produce at least one meaningful high and low. We also do not know the "drift" of a trading range (many ranges have a slight directional bias over time i.e. highs get a bit higher, and lows get a bit higher each time they are made; or vice versa) until at least 2 highs or lows have been made.
I have read the thread since it's inception and I have reviewed it.
There are some really profound comments here and I feel the diversity is a quality thing. Different strokes for different folks.
I do trade noise.
My approaches to the markets key off making money primarily.
I feel noise is everpresent and it only comes into dominance when the indicators of the market go dormant and the pace of the market goes into DU and VDU.
This inactivity puts the market into a state I call drift. It is also termed "chop" by some.
The easiest example of indicators demonstating a dominant noise condition is when realtavistic indicators cneter on the neutral and cannot reach the extreme levels where signals are given. Effectively the absence of delta (change) in price causes them to behave randomly.
By examining all fractals and crossing them with the range of possible volumes you get to see the bar tick length and it's std dev plus and minus reach convergence at lowest values.
On this matrix you can divide the performance into all the paces of market performance. Noise is a fringe zone directly opposite and away from the most productive zones of making money (higher and highest paces).
This is the zone where "random walk and chaos theory best applies as a solution.
So I trade noise using volume as my indicator for price timing.
to orient to this look at a no noise portion of today. See attached. you will see that the ES (5min) is cycling and that the YM/INDU (2min) leads the ES. Also notice that the high/ low range of volume is small on the leading indicator (smart money of YM).
The day will get to a noise level when the least volume reaches 80 contracts per 2 mins on YM. The ratio of H/L volume will gain to a level of up to 5 times.
I do not trade noise using the same rules I use for all other paces (slow, medium, fast, and extraordinary.).
The channels of noise are "super" containers. The key operant of noise is the "neutral or centering" values. As a court testifying scientist who testifies on hydraulogy I can't get to the sand bar and riffle (rapids) illustration.
I draw an opposite to meandering association which would be the path of least resistance model.
In noise there is randomness in a context of a bounding container. The best picture I find for making money is this: A rubber band attached to price value and also attached to the neutral. To say it as abo did I would say the line of MLR is the neutral and the boundary is one deviation where the band on price strteches no further.
The action is NOT to cycle from one side to the other. What is it? THe action is to pull away from neutral and then the stored potential energy is given up to pull to neutral as the kinetic action. DO NOT count on any underdamping that simulates cycling. All of this is because of the TC (time constant) of the system. The TC is MUCH longer than the harmonic or fundamental that cannot be created.
What insurance is there for when noise ends? The market wake up is telegraphed to you. It comes as a critical volume is reached on the extremes of the excursions. When the volume enters the slow pace region, you must discontinue the rubberband model. It is at this time you have the first potential for BO's instead of the guaranteed FBO's of noise. An adjunct to post noise periods (after noise)is that the volume forms gausians of about 10 two min bars. In noise the gausian pattern is not formable and the volume lines are independnat in color and sequences. This random bar sequencing is a true measure of noise dominance.
Specifically, what I do is marinate. I enter on the boundary of the channel. I make half the run by getting to neutral. I may have to return to the entry value again before I exit. I hold at the low risk of the boundary condition. when the price finally goes to neutral and then begins to move to the opposite side I hold and exit (and reverse) on the boundary opposite and come back to neutral at the end of the stretch.
What I described is making money by going away from neutral and back again. Each pair of these gives me two profit turns, where I can come within a tick (market orders dictate that at best) of the bounds.
When and as pace comes into play, I am holding and on the right side of the emerging move.
The best reason to trade noise is to learn to do washes. The second best reason is to be monitoring at all times and mentally getting away from the concept of edge trading. The third best reason is to have the experience of always being on the right side of the market on BO's. The fourth best reason is to be better able to deal with stalls in slow paced markets where, within a channel, you can always use the "noise" condition of a stall to continue to profit until the trend resumes or the FTT is reached to begin point 1 of the next trend.
With respect to this post: I have been told I am delusional, moderators call this gooblygook, and other attack me personally. The guardians of ET will assure you that my post will damage you. That all is fine with me. For all these people, feel free to crap on this in your own way. My view is that I contribute here in the way I do and It is not going to hurt anyone because they can ignore me or if they read it, they can be assured that by thinking about what I say, there will be an effect that, later may be helpful.