# How to use physics models in Time Series Analysis and Forecasting.

I have been studying methods of Technical Analysis for several years and I am disappointed. I actually do not consider it useful. I have not met anyone who can constantly win in the market using these technics.
Right now I am more lining towards methods derived from physics and here comes my questions. Do you know some methods which can be considered as indicators for measuring market fluctuations and “the heat”? Is this approach worth investigating?

## 1 Answer

I am not a physicist, but I thought about some approaches based on physics several months ago. Some of them are easy to implement and some are really hard. The list below is made from the easiest method to the hardest:

-          You can start from the basic physics of the movement and measure the velocity of the time series ( based on v = road/ time). You can calculate road using different approaches:

a)      Using a sum of absolute values of returns or price changes.

b)      Using a sum of returns or  price changes.

c)       Compare price changes between point A and B in time series.

It will give you some kind of volatility indicator for the time series.

-          Then you can analyze the road - s (from the equation s = 0.5*a *t^2) using a model with acceleration

-          Moreover, you can measure the “kinetic energy” of the time series using the equation E_kin =  0.5 * m* v^2 where mass can be considered as the volume of the security.

-          You can use the Fourier transformation https://en.wikipedia.org/wiki/Fourier_transform and try to decompose the time series into a series of wavelets. This might be very hard computationally . In this case you need to analyze how to predict how long this wavelet models will last in time. An advantage of this approach is the fact that you can make a forecast when you have the model calculated.

-          Lastly, there is a lot of methods for derivative pricing based on physics. For example the most popular model for options pricing – the Black Scholes model is based on the equation which is identical to the heat diffusion equation.

• not a "physician" or not a "physicist" ;-) – vonjd Sep 7 '15 at 11:13
• Hehe such a terrible mistake. Thanks for noticing it :) – Robert Szóstakowski Sep 7 '15 at 11:19
• I have done some Fourier Analysis of Stocks which you can see here: Fourier Analysis of AAPL. For most stocks, I don't think it is very useful, but for some, you can find some interesting cyclical trends. – Greg Thatcher Sep 22 '15 at 18:46