Abstract:
Forecasting is becoming increasingly relevant to producers and consumers in all markets
today. Both for producers and consumers, forecasts are necessary to develop bidding
strategies as well as negotiating skills at the market in order for both parties to maximize
benefits. Due to fluctuations of the Kenya shilling strength, weather conditions, politics
and many other variables in the markets, prices of goods are highly volatile. The
volatility of prices following a pattern makes it a time series phenomenon. Everyone
with a new prediction method wants to try it out on returns from a speculative asset,
such as stock market prices. Papers continue to appear attempting to forecast stock
returns usually with very little success. This project is aimed at estimating the amount
of predictability of a time series data using the Hurst exponent index. The Hurst
exponent (H) is a dimensionless estimator for the predictability of a time series. Initially
defined by Harold Edwin Hurst to develop a law for regularities of the Nile water level,
it now finds applications in financial data such as stock prices. The Hurst Exponent
can be interpreted as a measure of the trendiness: To be more specific, different values
of Hurst exponent imply fundamentally different price behaviors. It is a statistical
measure used to classify time series into either a random series or a trend reinforcing
series. The larger the index value is, the stronger the trend. In this study we have
investigated the features of time series associated with different estimates of Hurst
exponent. It is shown that series with large Hurst exponent can be predicted more
accurately than those series with Hurst exponent value close to 0:50. The main focus of
this study is therefore to determine: Why Hurst exponent index swings from persistence
to anti-persistence. Estimating of the Hurst exponent for time series data plays a very
important role in research of processes which show properties of auto-correlation.
