RiskButler.com™



Financial Forecasting


Risk Butler says

"Welcome to Risk Butler! I forecast stock prices and more. To get started say, for example: TELL ME ABOUT NETFLIX"


You ask (or answer)

Virtual assistant loading!


This is a Beta version.

Examples of questions

Virtual assistant loading!

All data is fetched, natural language text/speech is interpreted, and risk calculations are made each time you pose a question. Note the impressive speed - though after the first question.

Stock exchanges included:

USA stock exchanges, European stock exchanges, foreign exchange rates, some crypto currency prices and a few commodities.

Assistant Notes

10.000 Monte Carlo simulations are used for the calculations. "Chance" is measured as the average of simulations between the 99 percent quantile and maximum, and "Risk" is measured as the average of simulations between minimum and the 1 percent quantile. The risk measure is also known as Expected Shortfall.

Risk calculations are solely based on recent past movements of the stock in question, meaning that if future movements turn out to be much different then the forecast will not reflect this new behavior. We have better models but they are too slow to respond in the time required for a quick correspondance.

Warning!

Your profit or loss can be higher than reported.

Data, Machine Learning & AI

RiskButler.com is predictive Fintech software, services and consulting using Data, Machine Learning & Artificial Intelligence (AI).

We have a scalable cloud infrastructure where individual modules can be used locally too.

A core competence and focus area is forecasting of financial market prices and your portfolio values. This covers e.g. market price risk, counterparty credit risk (OTC derivatives) and credit risk.

We engage with your business to find the best solutions: Perfect solutions for asset management and financial institutions' client services and much more!

Four interfaces

There are four ways of using RiskButler.com analytics and forecasting capabilities:

  • via mobile/browser web app
  • via an API (Application Programming Interface)
  • by fetching pre-simulated forecasting data (think weather data) and other data
  • via virtual assistants & platforms (like Google Assistant or Home, Alexa, Cortana, Facebook Messenger)

An Example

Predictions and Backtesting

This is an example of on-going seven (7) day forecasts of the EUR/USD foreign exchange rate, using a particular so-called stochastic model.

For exchange rate trading purposes we want to predict EUR/USD rates (blue line) as well as possible, and these forecasts are pictured in the first diagram as the mean values of the simulations (green line).

In another context, a risk management context, we can use the same model algorithm, but instead of looking at the means (averages) of the simulations we instead turn our focus to the high and low forecast values, so-called percentiles.

In the second diagram, the two lower and two upper curves correspond to high and low quantiles: these are also forecasted. The objective is that the inner curve should not "break" one of the surrounding paths more often than expected by the statistical model and algorithm.

A backtest is a method to check the forecasting accuracy of a particular prediction model. RiskButler.com includes a wide variety of models that are specified in purely mathematical terms.

Backtesting is a model control method where you go back in time and pretend you do not know the past - and from here estimates are made forward in time, and thereafter compared with actual realized market prices. This is the way these two diagrams have been created.

Prices

Backtest example: simulated mean vs. actual

Limits

Backtest example: simulated outliers vs. actual