Research and Modifications


JAB SPECULATE LLC attributes its exceptional performance to superior research methods. All the models tested are robust with very few degrees of freedom/parameters. Many proprietary statistical techniques are examined including analysing multiple time period subsets, market-by-market as well as sector analysis and correlation, risk/reward analyses, parameter degradation studies, slippage analysis, and drawdown analysis. An independent research team develops and tests systems and then trade projects and cross-check all results using different research platforms. Experience helps in developing a successful trading system. The most common and most dangerous error made in system development is curve fitting. One thing we have learned over the last years of trading is that curve fitting cannot be understood by theory alone. There are many statistical traps that can only be learned by trading systems real-time. Statistics require many assumptions.

It is extremely difficult to know which of these assumptions are valid in the real world until they are actually put into practice in the real world. This takes time. JAB has learned many lessons by comparing more than 25 years of real-time trading to historical simulations.
Many methods commonly accepted by economists, statisticians, and even the CTA community-at-large have major flaws when executed in the real world. This is because the assumptions are valid only in theory. JAB’s success comes from original thinking and a refusal to blindly accept statistical models, economic theories, and even “common wisdom” prevailing in the investment industry. JAB uses no assumptions in research that have not been proven through extensive testing and real-time experience.

JAB began its trading career after much research and testing and in early August 2006 implementing its original strategy, a single system trading 21 markets with an average holding period of 2-4 weeks. Over the past 9 years, this original system has evolved as a result of new research discoveries. Since 2015 our research has focused on incorporating non-correlated systems in an effort to reduce overall program volatility while maintaining our target annual rate of return of 15-20%. Prior to January 2015 our annualized daily standard deviation was as high as 33.8%. However, since our modifications in 2015, our annualized daily standard deviation has dropped to 12.1%.