|
Long-term forecasting of BFI using chaos cycle
theory and maritime technical analysis
Alexandros M. Goulielmos
Ex Professor Marine Economics Department of Maritime
Studies, University of Piraeus, 80 Karaoli and
Dimitriou St., Piraeus 18534, Greece.
E-mail:
ag@unipi.gr
and
am.goulielmos@hotmail.com
Accepted 04 December, 2012
|
|
We tackle
the problem to produce reliable long-run freight
rates forecasts in maritime economy. Available tools
developed in academia aimed at predicting stock
prices with short term correlation, using models
like GARCH. Moreover, Chaos theory models developed
since 1963 by Mandelbrot, using Rescaled Range
Analysis, provide short-run forecasts, the range of
which depends on Lyapunov’s exponent. Moreover,
“Maritime” Technical Analysis, due to Hampton,
supports the existence of short-run (3-4 years) and
long-run (16-24 years) shipping cycles. So, the
paper applies rescaled range analysis and maritime
technical analysis to produce long-term forecasts
through cycles. Since 1981, and in 1973, and at the
end of 2008, shipping has experienced dramatic drops
in freight markets, which Mandelbrot, in another
context, described as the “joker” in the pack and
“Noah” effect. Applying BDS and other tests on BPI
time series of daily and weekly rates (1999-2011),
we found: non-normality, long term correlation and
chaos. The Hurst exponent found 0.93<1.00,
indicating a very strong ‘black’ noise. The
‘Lyapunov’ exponent allowed forecasting up to 6
days/weeks. In such a case, to obtain long term
forecasting we calculated cycles on the principle
that those persistent cycles will be repeated.
Cycles identified with Chaos theory were: 28 months
to 35 and 4 to 9 years -using Vn statistic. In
addition, Maritime technical analysis showed the
short term cycle, which ended in May 2011, and the
long term cycle, to end in 2017.
Keywords: forecasting freight rates index,
chaos and Vn statistic/H exponent, maritime
technical analysis, BPI 1999-2011, tests for
normality (JB) and iid (BDS
|
