Our ‘Smart Factory’ concept is based on a combination of internet technologies, big data analyses, and artificial intelligence algorithms, allowing us to automate, configure and systematise the entire production (i.e. investment) process.

Our highly diversified investment universes are shaped by information theory methods. Machine learning algorithms provide accurate forecasts for many thousands of securities and markets. Combined with state-of-the-art risk assessments, this yields above-average optimisation results.



Our investment solutions are risk-adjusted; in other words, we consider the ratio of return and risk estimators, in order to achieve the best allocation by optimising both factors. Our approach also supports investment strategies with individual constraints regarding risk exposure (risk budget); such are implemented particularly in the case of absolute return strategies. Thus it is essential that we are able to take the potential market losses into account. For this purpose, we use the latest insights of financial mathematics, and evaluate market risk by applying the ‘Expected Shortfall’ risk metric. An Expected Shortfall of q per cent corresponds to the return in the worst q per cent of cases. In contrast to volatility or value at risk, Expected Shortfall incorporates the risk of extreme events; thus, it is a realistic measure for losses potentially incurred in crisis scenarios.

Based on machine learning methods, we have developed estimators which forecast the expected performance of a market over different time horizons (e.g. week, month, year). These forecasts are established on the basis of the prevailing market regime – which is automatically recognised for the time horizon to be analysed. Contrary to moving averages, regime-oriented return estimators can adjust the length of the regime in qualitatively different market scenarios. In phases of market stability, estimators calculate the return expectations on the basis of long-term time horizons, shifting to short-term horizons in times of crisis. The individual results are used as input data for classifiers of the machine learning procedures, determining whether a market is investable or not, and which return is to be expected for the next investment period.

Investors prefer a highly diversified universe, in order to already minimise the risk of market fluctuations within the scope of portfolio construction. Systematically establishing an investment universe requires criteria defining the forecasting quality of an asset on the one hand, and the degree of diversification of a universe on the other hand.

The forecasting quality of an asset is determined by the signal-to-noise ratio, an indicator which was originally used in the field of physics. In the financial sector, this measure indicates ‘purity’ of the positive and negative regimes of a security. The higher the signal-to-noise ratio, the easier it is to make predictions regarding a security.

The degree of diversification of a universe is based on measuring the information shared by assets in that universe. Assets with a high correlation, e.g. DAX 30 and CAC 40 constituents, share a lot of information – and therefore contribute little to diversification. Including assets with a low correlation, however – such as equities in the DAX 30 and MSCI Pakistan indices – increases diversification due to the low level of overlapping information.
The final investment universe combines markets with the highest degree of predictability and the highest level of diversification, via an algorithm.

For us, constructing a portfolio is the process of understanding your individual investment restrictions, and taking them into account when optimising allocations. These restrictions can be very varied: fund price floors, maximum annual drawdowns, individual risk budgets, market and asset class requirements, etc.

Our ‘Smart Factory’ allows the implementation of any sensible restriction and is thus capable of developing customised portfolio solutions.

Once investment restrictions have been defined and an optimal universe has been found, the algorithms calculate the optimum risk/return structure and thus the best allocation for your portfolio in the prevailing market situation.

GET Capital’s entire investment process is fully automated, thanks to the Quant 4.0 infrastructure: securities and portfolio data are compiled and collected in databases using computer programmes. Subsequently, key forecasts and indicators, such as performance, risk and return, are calculated. These indicators are included in the daily portfolio optimisation for each managed portfolio. In the last step of this process chain, the software ‘produces’ the trading orders to implement the optimum portfolio allocation, forwarding it to the trading partner following a review.

The process can be designed and realigned via the respective interfaces, e.g. by changing or re-defining your constraints, or adjusting the investment universe via software modules that intervene directly in the portfolio optimisation.

Automation, and the flexibility of our Quant 4.0 ‘Smart Factory’, allow for implementation of your individual catalogue of requirements.