Relating signal processing to trading
Hi All,
I would be grateful if some of the wise people could help me understand a few things:
What do people mean when they say that financial markets are non Gaussian and are stochastic time series?
Does non stationary mean the average over time varies?
What does overshooting mean in context of digital signal processing?
And the last one; it is said that the noise in trading is multiplicative in nature. What does that mean?1. Because some think that time series is random (see some more info about stochastic processes Stochastic process - Wikipedia, the free encyclopedia) - they are forgetting that you can forecast next price changes/ranges within standard distribution in a % that those critics are not prepared to accept (a simple question : what are the odds that on Monday EURUSD exchange rate will be 1.5? So where is the "randomness" then?). What they are regularly talking as examples are black swans - and black swans are just confirming that the time series is just partially random in its nature - or else there would be no black swans
2. Non stationary == new data will be added/changed in that financial time series
3. Not too much if you consider that it is applied to financial time series - financial time series overshoota all the time - see how the price temporarily retraces when big changes happen - because the price overshot
4. because of point 1 - the assumption that financial time series is random also implies that trading just ads noise (not the opposite). Simple averaging is removing the noise so that point can be argued as not valid - the purpose of technical analysis is to remove the noise and the "noise in trading is multiplicative in nature" is usually used as an argument that no analysis can predict (which is a gross misunderstanding of TA - the purpose of TA is not to predict - nobody is trying to play a Gods role, but to remove noise from the data and to help in estimation (not prediction)).
1. Because some think that time series is random (see some more info about stochastic processes Stochastic process - Wikipedia, the free encyclopedia) - they are forgetting that you can forecast next price changes/ranges within standard distribution in a % that those critics are not prepared to accept (a simple question : what are the odds that on Monday EURUSD exchange rate will be 1.5? So where is the "randomness" then?). What they are regularly talking as examples are black swans - and black swans are just confirming that the time series is just partially random in its nature - or else there would be no black swans
2. Non stationary == new data will be added/changed in that financial time series
3. Not too much if you consider that it is applied to financial time series - financial time series overshoota all the time - see how the price temporarily retraces when big changes happen - because the price overshot
4. because of point 1 - the assumption that financial time series is random also implies that trading just ads noise (not the opposite). Simple averaging is removing the noise so that point can be argued as not valid - the purpose of technical analysis is to remove the noise and the "noise in trading is multiplicative in nature" is usually used as an argument that no analysis can predict (which is a gross misunderstanding of TA - the purpose of TA is not to predict - nobody is trying to play a Gods role, but to remove noise from the data and to help in estimation (not prediction)).Thanks for coming back so quickly. Can you elaborate more on the overshooting thing; from what i gather digital filters might overshoot price if we try to reduce the lag? Didn't quite get this.
I was under the impression that most noise would be additive in nature (because of market makers' actions) and a little might be multiplicative when there is some fundamental activity (news, central banks).
Thanks for coming back so quickly. Can you elaborate more on the overshooting thing; from what i gather digital filters might overshoot price if we try to reduce the lag? Didn't quite get this. I was under the impression that most noise would be additive in nature (because of market makers' actions) and a little might be multiplicative when there is some fundamental activity (news, central banks).
The overshooting is a simple result of traders jumping on a trade too late (which is what the market makers are after - they need those buffers in order to exit without letting the price retrace completely to the starting point).
As of noise : it would be a noise if the market makers were not trading in the same direction. Don't forget one thing : when it comes to men, randomness is not what is a result of their actions. It is a human nature : we are tribe, not individuals
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hi All,
I would be grateful if some of the wise people could help me understand a few things:
What do people mean when they say that financial markets are non Gaussian and are stochastic time series?
Does non stationary mean the average over time varies?
What does overshooting mean in context of digital signal processing?
And the last one; it is said that the noise in trading is multiplicative in nature. What does that mean?