Saturday Links: Sunday edition! Yeah, that's right. I didn't get around to writing this yesterday.
How the Flash Crash Trader’s $50 Million Fortune Vanished -- Bloomberg Markets -- Investigative piece on Navinder Singh Sarao's fortune, which is tied up in a complicated web of offshore investments. Sarao, dubbed the "Flash Crash Trader" and "The Hound of Hounslow", was convicted for fraudulent trading in the financial markets. Interestingly, aged almost 40, he lived in and traded from his bedroom in his parents' house. He went on to make roughly $50 million from futures trading. US regulators claim he helped cause the Flash Crash of May 2010. After having been extradited by his home country England and subsequently sent to Chicago, Sarao was ordered to pay $38.4 million to the CFTC and US Justice Department. I'm working on a post about the accusations myself; watch this space for more. I'm not going to defend Sarao's trading strategies, which may have involved an HFT practice called "spoofing", but I think it's ridiculous this matter had to be settled in the US and I must say it's nothing short of embarrassing if one individual trader can break the international financial markets from his bedroom. I mean, if that's really true, how stable and trustworthy can today's automated markets be?!
A Litany of Problems With p-values -- Statistical Thinking -- Informative post about the shortcomings of null hypothesis testing and p-values. Practitioners will find this extremely useful. From personal experience, I can say that null hypothesis testing only looks easy from the outside. It's extremely easy to calculate p-values, after all. Press a button in the statistical software of your choice. But to make educated inferences you should look beyond the numbers and ask yourself whether you would fully trust your statistical tests: If you're testing a potential trading idea, would you put your own money into it?