Fat Tails Quantified and Resolved: A New Distribution to Reveal and Characterize the Risk and Opportunity Inherent in Leptokurtic Data
The Thorne Distribution is the result of years of after-the-day-job effort. It began innocently enough. Reading about what a pox on the financial world leptokurtic (i.e., “fat- or heavy-tailed”) data sets had become, I was amused. “Silly business types!” I thought to myself. “They should just have come to the Real Men of weird distributions—the spectroscopists like me!” And so I decided to spend an idle Saturday afternoon flipping through my broad repertoire of fat-tailed distributions to find one that would help the investment folk out. A month later, I conceded defeat; nothing I had was adequate to fit both center and tails of every leptokurtic/skewed distribution. But by then my curiosity was engaged, and so I launched what became a very long quest.
The Thorne Distribution is an odd duck. But it’s also seemingly infallible. I have tested it in a wide, deep, challenging spectrum of applications, and its performance has been uniformly excellent. It is in finance, however, that I have experimented the most. There, the distribution behaves almost as though it represents the innate nature of markets.
As a further test, I also developed a stock market trade-timing indicator that incorporates unique filters and smoothing algorithms that spring from and re-incorporate the Thorne Distribution; it delivers jaw-dropping returns. The indicator does not deliver long-range “predictions;” it simply gives highly accurate buy/sell/hold signals on a day-to-day basis.
Leptokurtic and/or skewed data is a fact of life, not just in finance, but in a myriad of diverse areas ranging from the turbulence characteristics in aerodynamics and hydrodynamics to internet traffic statistics to the epidemiology of disease outbreaks. If the Thorne Distribution finds application--proves useful!--in any of these, my time will have been well spent.
The paper I have written to introduce and explain my new distribution is available, open access, at: http://arxiv.org/abs/1110.6553. (See the PDF file in the Download section there for the complete text of the article.)
The Thorne Distribution is the result of years of after-the-day-job effort. It began innocently enough. Reading about what a pox on the financial world leptokurtic (i.e., “fat- or heavy-tailed”) data sets had become, I was amused. “Silly business types!” I thought to myself. “They should just have come to the Real Men of weird distributions—the spectroscopists like me!” And so I decided to spend an idle Saturday afternoon flipping through my broad repertoire of fat-tailed distributions to find one that would help the investment folk out. A month later, I conceded defeat; nothing I had was adequate to fit both center and tails of every leptokurtic/skewed distribution. But by then my curiosity was engaged, and so I launched what became a very long quest.
The Thorne Distribution is an odd duck. But it’s also seemingly infallible. I have tested it in a wide, deep, challenging spectrum of applications, and its performance has been uniformly excellent. It is in finance, however, that I have experimented the most. There, the distribution behaves almost as though it represents the innate nature of markets.
As a further test, I also developed a stock market trade-timing indicator that incorporates unique filters and smoothing algorithms that spring from and re-incorporate the Thorne Distribution; it delivers jaw-dropping returns. The indicator does not deliver long-range “predictions;” it simply gives highly accurate buy/sell/hold signals on a day-to-day basis.
Leptokurtic and/or skewed data is a fact of life, not just in finance, but in a myriad of diverse areas ranging from the turbulence characteristics in aerodynamics and hydrodynamics to internet traffic statistics to the epidemiology of disease outbreaks. If the Thorne Distribution finds application--proves useful!--in any of these, my time will have been well spent.
The paper I have written to introduce and explain my new distribution is available, open access, at: http://arxiv.org/abs/1110.6553. (See the PDF file in the Download section there for the complete text of the article.)