Development process
Data Collection
In the first phase of our development process we collect data from a variety of traditional and alternative sources. This allows us to test our hypotheses in a more comprehensive and extensive way in the following phases of our strategy development framework.
Data Cleaning
We use automated algorithms to clean large amounts of data. The automated cleaning procedures are evaluated and verified by our data scientists to ensure that the data sets are suitable for use by our quantitative researchers.
Data Analysis
Using sophisticated and advanced statistical methods, our quantitative researchers analyze the processed data sets to uncover alpha signals. This step is most important, since it is hard to find market tendencies. As a consequence it is important to have an excellent team of quants that can uncover reliable market opportunities through vast amounts of noise.
Backtesting
After the data has been thoroughly analyzed, our quants apply the scientific method to the financial data by testing their hypotheses through backtesting experiments on long-time series data spanning multiple decades and different markets. This allows us to find stable statistical models that give us predictable results over the long term.
Pilot Test
Once our quantitative research team has found investment strategies that have performed well in backtesting, we test them on our own capital for a suitable amount of time before implementing them in a pool of strategies working with clients capital. This step is very important since it increases the probability that the strategy will work as expected in real market conditions and allows us to have the time to fix potential technological issues before live production.
Implementation
Having confirmed that pilot trading returns are correlated with the backtest and what we expected. Our programmers added a new robot to the main algorithm pool.