Our investment principles
A short history of the S&P500 (1975-2017)
Lessons learned? Financial history is there for all to read, yet market participants have short memories. Additionally, each generation of investors seems to believe it is different this time. This inevitably leads to each investor generation repeating the mistakes of former generations. The collective behavior of investors and human nature does not change over time as it goes through cycles of fear, greed and exuberance, which predictably leads to asset price booms and busts. Financial crises, as shown in the video, are interrelated and interdependent phenomena; they share the same root cause(s) and thus constitute an integral part of the system and its dynamics.
Our proprietary software
Our proprietary analytics engine is based on sophisticated LPPLS based algorithms analyzing daily market prices of thousands of stocks globally. These algorithms look for specific patterns over multiple time horizons (74 per stock) that are caused by hierarchical feedback of market participants. Next generation behavioral finance insights are at the core of our unique approach.
Our engine detects price patterns (LPPLS signatures) and exploits forming bubbles. It has the ability to anticipate endogenous directional price movements days or weeks in advance. Contrary to many other traditional quantitative and machine learning approaches which try to predict all the time, our approach inherently distinguishes random periods from periods where limited visibility can be achieved. Detecting such pockets of predictability is key.
The technology underlying our analytics engine is based on decades of academic research and years of hands-on experience developing a practical LPPLS approach which is both proprietary and – while still requiring enormous computational power – is orders of magnitude faster than other approaches and is also implementable in portfolios. Our analytics engine has been extensively backtested in different economic cycles over the past 40 years and has shown consistent outperformance. The core software is enhanced by deep machine learning algorithms.
Technology we use
We are standing on the shoulders of giants as we use decades of academic research and open-source software tools. All of our production systems run Linux and PostgreSQL. For data analysis, machine learning and back tests, we are heavy users of Python and its full scientific stack. For our core LPPLS engine we use Cython/C/C++ and Intel MKL to push the performance boundaries of Python. We implement the data collecting and preprocessing systems that need the highest data throughput in Java / Scala.
At SIMAG® we have benefited hugely from open-source technologies. We contribute back to the community whenever we can.