Big Data Systems

We have designed algorithmic trading systems that operate in derivatives markets in an unattended way, according to the predictions generated by the AI Big Data Models. The systems are available through Tading Motion platform:

Intraday Big Data Algorithmic Trading Systems:

During the night, InvestMood's intraday systems analyze Big Data information and qualify it as good, bad or neutral using natural language processing algorithms. With this information, artificial intelligence models are created. Theese models generates everyday a prediction about the next session Ibex trend. Based on the probability offered by the model, these systems are left out of the market if the probability is low or they enter in to the market long or short, keeping their position open until the market close or until the position closes with a dynamic stop loss and take profit.

The big data systems of intraday algorithmic trading developed by InvestMood to date are:

Investors' Mood Daily Trend (Long Learning): This big data system of algorithmic trading daily tracks any communication, both specialized and non-specialized, mentioning Ibex on twitter, radio, TV and press. The obtained information is automatically analyzed and, subsequently, qualified as “good”, “bad” or “neutral” according to a natural language processing algorithm. Then, harnessing the qualifications of communications’ feeling about Ibex, a predictive model is created to obtain Ibex’s trend in the next trading session (upward or downward trends). The system automatically adjusts to short or long at Future Ibex opening according to the obtained prediction and close position at 17:30h. The predictive model is continually evolving by the use of machine learning that enables to the model to learn from the made mistakes.
The algorithmic trading big data system is available on the platform Trading Motion in two different versions:
  1. Investors' Mood Daily Trend: This system operates everyday according to the obtained prediction even when the last has not a high probability to succeed.
  2. Investors' Mood Daily Trend Low Risk: this system opens position only if there exist clear signs and the probability to succeed is higher than 80%.

Investors' Mood Daily Trend (Entrenamiento corto): This system was born as an evolution of the previous systems. It also operates intraday and is based on the same investor sentiment information as the previous systems, but proposes two differentiators. The first is that the artificial intelligence model built to make the prediction has used a 6 months learning period, neglecting the previous historical information. With this, it aims to be more agile (it enters the market if the probability of rise or fall is greater than 80%) and adapt more quickly to the sensations of the market. On the other hand it incorporates a trailing stop loss and take profit that optimize the profitability of the system in the sessions in which there are changes of trend throughout the day.

  1. IA 6M Learning Ibex Mood:

Big Data MiniIbex Trend (combinación de sistemas): This big data trading system operates with the future of the MiniIbex so it only requires a daily guarantee of only € 200. The algorithm that sends the orders to the market has been developed combining the previous systems, it operates whenever both models, long and short learng predict the same trend within a probability higher than 70%. You can see the evolution of this model in:

  1. Big Data MiniIbex Trend:

Continuous Big Data Algorithmic Trading Systems:

Following the paper "Investment signs based on a risk aversion index" of Professor Dr. Raúl Gómez Martínez, published in 2013 by IEDEE, the volume of searches carried out in Google on economic-financial terms has explanatory and predictive capacity on the evolution of the markets. Since the year 2004 in which Google began to publish these statistics, it is observed that in bearish markets the searches of terms such as crash, recession or short selling are higher, while in bull markets the searches on financial products are more abundant. Bearing this in mind, InvestMood has created big data algorithmic trading systems that take long or short positions based on an advanced analytical model in which the explanatory variables are those volumes of searches while the target variables is evolution (up / down) of main stock index. The models release their predictions on a weekly or monthly basis:

Google Week Trend Systems: These systems train their artificial intelligence models every Sunday, using a 5-year history of weekly Google search statistics of dozens of economic-financial issues, and issue a prediction for next week's trend. The system maintains a long or short open position until there is a new prediction in the opposite direction or until the position is closed with an optimized management of trailing stop loss and take profit, the system remains out of the market until the new prediction re-entering the opening of Monday.

Sistemas Google Monthly Trend: On the first day of each month the models are trained with all the history available since 2004 and a prediction is released for that month. The system remains open continuously, long or short, according to the prediction obtained.