Page 216 - Special Topic Session (STS) - Volume 1
P. 216
STS425 Nur I. et al.
The estimation process that produces a good model is also able to provide
identification of the first step analysis for each regime change in each daily
transaction. Giving values on latent variables during the EM iteration process
to converge can be obtained an empirical record of the membership of each
daily transaction data as a member of which regime. At the same time, the
distribution of the run length of each regime would be known.
Figure 2 Plot of the Dominated Regime 2 of AALI and Regime 1 of SSMS during the Trading.
(a) Regime 2 of AALI and (b) Regime 1 of SSMS.
There is almost the same number of hitting time and ARL of all regime in
AALI, those are 227 and 228 times with ARL 12.79 and 12.04, respectively.
Figure 3 demonstrates the run length distribution for the second regime of
AALI and the first regime of SSMS. The longest time to stay in each regime as
a maximum run length (MRL) is 342 run length daily transaction for Regime 2
(see Figure 3. (a)) and 223 for Regime 1. SSMS trading time which is only one-
third AALI, on the other hand, has been divided into three structural change
regimes. Its first regime is the most frequently visited, which is 110 times with
ARL 6.236 and MRL 27 days of transactions (see Figure 3. (b)). Next is followed
by Regime 2 with 61 times visited having ARL 4.279 and MRL 31 days of
transactions. Finally, the third regime with 52 times visited with ARL 3.577 and
MRL 17 days of transactions.
Figure 3 Run Length distribution Regime 1 of AALI and SSMS.
(a) Regime 2 of AALI, and (b) Regime 1 of SSMS
205 | I S I W S C 2 0 1 9