Strategies.
Partner with Leibniz Group and experience the power of systematic investment strategies that have stood the test of time. Our expertise, cutting-edge technology, and commitment to excellence position us among the esteemed ranks of industry pioneers.
Invest with confidence, knowing that our strategies have a proven track record and rival the success of renowned firms. Experience transparency, trust, and tailored solutions that align with your objectives and risk appetite.
Strategies Overview
DISLOCATION
Inception: 2007
Type: Machine Learning Stat-Arb
The Leibniz DISLOCATION program is a set of machine learning and AI-based equity statistical arbitrage strategies that aim to profit from "dislocation", or disagreement among market participants about security values. It generates consistent, low-volatility returns that are not correlated to the broader market. The program uses diversified, liquid S&P1500 names, with an average of 1000 live positions and up to 2000 orders per day. The portfolio is constructed using a combination of optimization techniques, including closed-form, stochastically robust, and meta-heuristic methods.
EMOTION
Inception: 2011
Type: Multi-Strategy Stat-Arb
The Leibniz EMOTION program is a systematic investment strategy that has been utilized since 2011. The program trades globally diversified cash equities and ETFs, focusing on medium-term investor underreaction and short-term investor overreaction, operating with low frequency and capacity constraints. EMOTION aims to manage drawdowns and has been used to anticipate earnings surprises and identify ETF pairs with potential for price divergence.
SENTINEL
Inception: 2017
Type: AI/ML Multi-Strategy
SENTINEL is a deep learning-based, systematic trading program that utilizes daily sentiment data to predict price trends in a variety of global markets. The first sub- strategies went live in 2017. SENTINEL analyzes unstructured public data content from thousands of sources related to the economy and politics and uses proprietary tools to filter and map the data into price trend predictors. By doing this SENTINEL offers a unique perspective on potential market movements. SENTINEL uses supervised learning of classifiers across a broad spectrum of machine learning methodologies to improve prediction quality and to automatically create a diverse set of trading models and combines them into a multi-model, multi-scale strategy for high returns, maximized alpha/Sharpe and minimized drawdowns.
SWARM
Inception: 2015
Type: Machine Learning FX
The trading strategy uses a combination of neural networks and advanced statistical methods to trade a diverse range of financial products, primarily spot foreign exchange but also using futures market data. The sub-strategies are chosen using a Monte Carlo simulation and Markov Chain Monte Carlo process, and are combined into a single network with varying weightings. The strategy holds positions for varying lengths of time, typically 2 days, and has a high turnover. Since its inception in 2015, the strategy has evolved to become a system of non-optimized sub-strategies with unequal weightings, allowing for a more flexible and dynamic approach to the market.
REGIMENT
Inception: 2015
Type: Mid-Frequency FX
The REGIMENT investment program is a fully automated strategy that seeks to profit from volatility in the foreign exchange market using both directional and mean-reverting approaches. The program employs quantitative analysis, order routing algorithms, low latency technology and liquidity management to identify and exploit market inefficiencies. It is designed to adapt to different market conditions and risk regimes, using a diverse range of trading and hedging strategies. The program also includes a master risk layer that runs thousands of concurrent computations per second to manage risk.
P-REACTION
Inception: 2016
Type: NLP Stat-Arb
Leibniz P-REACTION rigorously applies a data-driven framework using machine learning methods and natural language processing to predict how investor behavior impacts asset prices and leads to short-term mispricing surrounding earnings related news and events in a medium-frequency approach. P-REACTION digests and pre-acts as as well as reacts to many thousand events via its news-driven strategies in a fully systematic manner. Many years of development work and attention to detail now allow the strategy a reliable approach and fast reaction time to market moving news by exploiting behavioral biases and market inefficiencies.
