We turn data into business decisions

The focus in real estate, construction and project finance lies on sound portfolio and project management. The aim is to reduce model risk, optimize project costs and subsequently, improve return on investment. 

Our services include:

• Cashflow estimation and modeling
• Property evaluation
• Identification and tracking of economic- and market trends
• Financial modeling: feasibility, risk and rentability studies
• Risk management and development of hedging strategies

The media and retail markets are highly competitive and marketing focused. In an online environment, customer attention is volatile. High conversion rates can only be achieved if the customer journey is without any hurdles and concluded in a timely manner.

Our services include:

• Inventory optimization
• Sales predictions
• Client optimized pricing evaluation
• Strategic sale campaign management
• Social Media analytics
• Affiliate and Non-Affiliate Marketing Performance measurement

The financial industry has a well-established data infrastructure. Hence, there is great potential for increasing efficiency in front office and reporting applications. The key to success is to understand data in relation to the respective markets and investment instruments. Any conclusion drawn has to be tested against proven financial models and viewed in the light of correlation breakdowns.

Our services include:

• Portfolio metrics and optimization with respect to the mandate
• Performance attribution
• Risk modeling: IRR, FX, credit and liquidity. 
• Hedge performance analytics
• Trade momentum analysis

The design of the dataset is the first and the most consequential step in the data analysis process. It sets the boundaries for what can and cannot be achieved and therefore must be tailored to the relevant business hypothesis, yet has to be flexible enough for future use cases.

This task includes:

• Consolidation of structured and unstructured data from various sources 
• Efficient database design, ranging from simple PostgreSQL to Spark SQL Hadoop Hives
• Descriptive database analysis to eliminate statistical pitfalls
• Data classification and clustering reports for future project documentation

Big Data and Artificial Intelligence are the dominant buzzwords in the industry. However, even the most basic Neural Networks require an incredible amount of data to return reliable results. In order to guarantee efficiency, we choose the right tool for the task, since addressing complex questions with even more complex solutions is rarely good advice.

Our toolset includes:

• Simple multivariate OLS regressions
• PCA dimensionality reduction techniques
• Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM)
• Auto Encoders and Convolutional Neural Networks (CNN)
• Boltzmann Machines and Generative Adversarial Networks (GAN)

The way variables in big datasets interact with each other forms a complex system. Without dimensionality reduction techniques we would, so to speak, miss the forest for the trees. The complexity that is still inherent to the final model itself is called model risk. It is a result of the temporal instability of relationships and the black-box nature of deep learning neural networks.

Our approach to manage model risk includes:

• Performance analysis of model convergence
• Robustness tests of the model
• Continuous reevaluation of input data on changes in volatility regimes and correlation breakdowns
• Heuristic interpretation of the model design features