Financial Market Context with Alternative Data & Machine Intelligence

Dimitri Huwyler, Head of Quantitative Strategy and Aleksandar Pramov, Quantitative Researcher, Next Gate Capital | May 21, 2018

View an extract of this session held at the London Big Data and Machine Learning Revolution event in April 2018.You can also access the full video and slides.


Successful market timing is a tantalizing holy grail for investors. On both side, investors and researchers have discovered that the market timing is harder than it might seems. At Next Gate Capital they think that this is a perfect research playground for new machine intelligence techniques and new alternative dataset. They use classic variables to build economic climate and global sentiment indicators, enhanced with news sentiments, particularly on politics and monetary policy (two fields very difficult to handle with classic dataset) and economy. They cover a practical example of enhancement of a trend following strategy.

Full Video access slides

Big Data Machine Learning Finance

It doesn't come as a surprise if I tell you that this revolution has pushed our Quants finance into a new era. We see two main reasons behind this. The first is something we have talked a lot about today, which is the access to a new type of data, which by construction and essence, captures very complementary and additional information compared to traditional datasets, such as fundamental data, analysis estimate or pricing to name a few. The second is the remarkable advancement in machine learning due to different reasons, algorithmic breakthrough, but also the ability for small firms today to access extremely computing power at very cheap costs.

Now our business will continue to be highly impacted by this revolution, but we don't see it as a negative disruptor or as a threat but an opportunity to revisit existing problems with new tools.

Mapping Financial Markets of New World of Investment

We talk about quantifying the investment process which is the ability to systemize more traditional, qualitative investment approach. We define that in five steps:

  1. Context Analysis
  2. Forecast or Signal Generation
  3. Signal Combination and Conditioning
  4. Portfolio Construction and Risk Control
  5. Trading or Cost Control

financial market context

Defining The Context: The building blocks

Context Analysis

The context is a summary of the health and propitious aspect of the environment in which evolve the mapping financial markets.

Define market regimes by:

  • Risk Management
  • Strategic Allocation
  • Flow Positioning

Context is broken down into two categories:

1. Global Economic Climate

  • Geopolitics / Politics
  • State of the Economy
  • Monetary Policy

This specific indicator was built using RavenPack data. It shows how news can build an indicator, that can quantify and react to geopolitical events and intention.

2. Global Sentiment

  • Positioning and Flow
  • Volatility
  • Prices and Volume
  • Correlation and Dispersion

Trend Following (TF) and Context Analysis

Starting point is Time Series Momentum:

  • Time Series Momentum, attributed to various behavioral patterns of investors
  • Systematic CTA have a relevant market share
  • We illustrate an enhancement of an existing strategy with the context analysis

An illustration of a long-short TF strategy, our benchmark:

  • Investment Universe: Cross Asset
  • Signal Generation: Time Series Momentum 12 month lag
  • Portfolio Construction and Risk Control: Risk parity optimization with a 10% target volatility
  • Re-balancing: Monthly
  • Currency: USD

Investment Decision

Motivation for enhancement:

Include Context Analysis to enhance the strategy, make it more reactive to changes in the context.

Context Analysis in the TF: inputs and Outputs (1)

  • We explore a large set of news sentiment indicators related to the context, stratified by region and/or asset class.
  • Constructed at topic level, multiple group level. Approach in selection based on experience and past findings in the literature.
  • At a first passage used model-free diagnostics to explore individual relationships between instruments forward returns and the indicators.
  • We filtered by entity relevance >75, events similarity days >1 and we created a daily average of the ESS. This measured the average news sentiment daily for the respective entities.

Context Analysis in the TF: inputs and Outputs (2)

Here we have to establish the relationship goals within our universe. So the first options was to build individual models for each asset. The other options that we will more, is a sort of overlay which comes as a top down approach to our systematic investment strategy. This is done by clusters of instruments. A simple way to cluster them is to look at their historical correlations and also interpreting the end results. We end up with subcategories of asset classes.

financial market context

Mapping financial markets: Putting it All Together

  • Combining the output of the predictive model and the TSMOM
  • Can apply the re-weighting on the signal level or weight level directly
  • Keep point: Separate the TSMOM signal and the Context Analysis
  • Keep monthly signal and rebalance more often eg weekly, based on Context Analysis

Summary and Оutlook for Mapping Financial Markets

  • Enhancing strategies with well established investment signals by using alternative data, such as news sentiment data that measures the context, is an area worth having further investigation.
  • Alternative Data have opened many more opportunities for explorations in the different steps of the investment process.
  • Now that we have the tools to quantify and the data to construct context indicators, we should include a systematic context analysis in the investment process.
  • It is important to remain cautious when analyzing such new data sources because their economic interpretability is not always straightforward and requires human expertise.

Full Video access slides

Generation AI

RavenPack’s events have become global, with attendance exceeding 250 buy-side professionals at the London Big Data and Machine Learning Revolution in April 2018. RavenPack Research Symposium returns to New York on September 12th.

Register Now

By providing your personal information and submitting your details, you acknowledge that you have read, understood, and agreed to our Privacy Statement and you accept our Terms and Conditions. We will handle your personal information in compliance with our Privacy Statement. You can exercise your rights of access, rectification, erasure, restriction of processing, data portability, and objection by emailing us at in accordance with the GDPRs. You also are agreeing to receive occasional updates and communications from RavenPack about resources, events, products, or services that may be of interest to you.

Data Insights

Read More