Distribuições hiperbólicas generalizadas: aplicações ao mercado português

O objectivo desta tese é discutir o uso das distribuições hiperbólicas generalizadas como modelo para os retornos logarítmicos de 4 activos do mercado de capitais Português. Os activos em análise são o índice Português PSI 20 e as 3 maiores empresas pertencentes ao PSI 20: PT, EDP e BCP. Os dados são constituidos pelos valores de fecho diário durante mais de 8 anos. Utilizando o software R procederemos à estimação dos parâmetros das distribuições para ajustamento aos dados empíricos. Para medir o grau de ajustamento das distribuições aos dados empíricos usamos os gráficos QQ-plots e 4 distâncias: Kolmogorov-Smirnov, Kuiper, Anderson-Darling e Fajardo-Farias-Ornelas. Os resultados obtidos permitem concluir que o melhor ajustamento é feito pela hiperbólica generalizada e em seguida a distribuição normal inversa gaussiana. Todas as distribuições desta família ajustam-se muito melhor que a distribuição normal. Por último temos uma aplicação ao cálculo do preço de derivados financeiros, nomeadamente a fórmula de uma opção de compra Europeia no modelo discutido.

The possibility of primordial black hole direct detection

This thesis explores the possibility of directly detecting blackbody emission from Primordial Black Holes (PBHs). A PBH might form when a cosmological density uctuation with wavenumber k, that was once stretched to scales much larger than the Hubble radius during ination, reenters inside the Hubble radius at some later epoch. By modeling these uctuations with a running{tilt power{law spectrum (n(k) = n0 + a1(k)n1 + a2(k)n2 + a3(k)n3; n0 = 0:951; n1 = ????0:055; n2 and n3 unknown) each pair (n2,n3) gives a di erent n(k) curve with a maximum value (n+) located at some instant (t+). The (n+,t+) parameter space [(1:20,10????23 s) to (2:00,109 s)] has t+ = 10????23 s{109 s and n+ = 1:20{2:00 in order to encompass the formation of PBHs in the mass range 1015 g{1010M (from the ones exploding at present to the most massive known). It was evenly sampled: n+ every 0.02; t+ every order of magnitude. We thus have 41 33 = 1353 di erent cases. However, 820 of these ( 61%) are excluded (because they would provide a PBH population large enough to close the Universe) and we are left with 533 cases for further study. Although only sub{stellar PBHs ( 1M ) are hot enough to be detected at large distances we studied PBHs with 1015 g{1010M and determined how many might have formed and still exist in the Universe. Thus, for each of the 533 (n+,t+) pairs we determined the fraction of the Universe going into PBHs at each epoch ( ), the PBH density parameter (PBH), the PBH number density (nPBH), the total number of PBHs in the Universe (N), and the distance to the nearest one (d). As a rst result, 14% of these (72 cases) give, at least, one PBH within the observable Universe, one{third being sub{stellar and the remaining evenly spliting into stellar, intermediate mass and supermassive. Secondly, we found that the nearest stellar mass PBH might be at 32 pc, while the nearest intermediate mass and supermassive PBHs might be 100 and 1000 times farther, respectively. Finally, for 6% of the cases (four in 72) we might have substellar mass PBHs within 1 pc. One of these cases implies a population of 105 PBHs, with a mass of 1018 g(similar to Halley's comet), within the Oort cloud, which means that the nearest PBH might be as close as 103 AU. Such a PBH could be directly detected with a probability of 10????21 (cf. 10????32 for low{energy neutrinos). We speculate in this possibility.