A moving average (MA) is a trend-following or
a lagging indicator which is typically used in identifying the direction
of the trends and in determining the support and resistance levels because it
is based on past price movements. The two basic and commonly used MAs are
Simple Moving Average; It is a simple
average of a security over a defined number of time periods,
Exponential Moving Average; It gives more
weight age to more recent prices.
This technical indicator is composed of a center line and
two price channels normally called bands located above and below the line.
Central line is the Exponential Moving Average and the bands are standard
deviations of the selected script. The price channels/bands expands and
contracts depending of the price movement of that particular stock.
the selected stock prices continually touch the upper Bollinger Band, the
script is thought to be overbought indicating the selling indication,
conversely, when they continually touch the lower band, prices are thought to
be oversold, triggering a strong selling signal.
a strong uptrend, prices usually fluctuate between the upper band and the
20-day moving average.
Modern methods of time series forecasting are developed
on the idea that past prices movements can be explain about its future behavior
The approach of
building a time series forecasting is to first specify a model. The model used
in forecasting generally composed of statistical formulation based on the
dynamic relationships between variables that we infer from (also known as
information set) and variables closely related to which we observe. The most
common approach of time series forecasting is derived from regression analysis.
A regression model includes a linear parametric relation between an explanatory
variables set and dependent variables.
Wiener (1949) and Kolmogorov (1941) are considered to be
pioneers in linear prediction field while most of the techniques used today for
linear prediction differ from there models but it is evident that their
solutions to same geometrical problems are still equivalent and provides
Weiner’s work was
mostly relevant to today’s times series forecasting in his work he rigorously
formulate the problem of “Signal extraction”; given observations on time series
degraded by additive noise.
Given the historical
perspective of mass systems of equation models popular for macro-economic
forecasters in mid-19th century it became apparent that the
forecasting models derived from signal extraction forecasted similarly as
compared to those based on complicated systems of economic relationship
equations formed as individual, interconnected, classical regressions.