Unraveling the Impact of Data Quality on Institutional Investments and Market Dynamics

Explore the Nexus of Data Quality, Fake News, and Institutional Investment Algorithms

In a chaotic and fast-moving world, executives often rely on their gut for decision making. Read our post on how to empower executives to use data to make decisions.

Chaos Unleashed: Nasdaq’s Data Error

A data error on the Nasdeq exchange caused chaos last week Monday, as the introduction of test data into the live systems vastly impacted the shares prices of a number of tech companies. Two notable examples: Amazon prices crashed by 87%, while Zynga bounced up by an amazing 3,292%

The Human Error Connection

In this case, the error can be linked to human error/data quality.

Arguably, most data quality errors can be tracked as human errors.

However, what is of more interest to me is the impact that an error, or manipulation, of this nature could potentially have on the machine learning algorithms that, increasingly, drive institutional investment.

Machine Learning’s Predicament

At its most simplistic, machine learning relies on pattern recognition and on copying or applying learned human behaviour to similar/related patterns.

For example, every time the stock price goes down by 10% implement a stop loss and automatically sell.

How many stop-loss calls may have been triggered by Monday’s error?

More importantly, what have the algorithms learned from the experience? Should a similar event occur in the future, will the machines recognise this as an error (maybe it won’t be) and do nothing, expecting the markets to recover? Can we predict their result?

Market Manipulation Unleashed

If a small subset of businesses trigger real actions based on an error, how will other machines react to their decision? The stock market tends to follow the pack and the actions of a few influential funds could again trigger chaos.

Sentiment Manipulation: A Market Threat

Which brings me to my last question.

Could we use social sentiment to trigger chaos?

South Africans are well aware of the risk posed by fake news, following the well-publicized efforts of a British “PR” firm to influence local sentiment and support the theft of state assets by the South African president and his cronies.

For most of us, it is difficult to tell the difference between reality and fake news stories promoted through social media. A key strategy of these PR campaigns is the creation of so-called news outlets that present the desired ‘reality” as fact, and propagate this reality through social media.

This kind of manipulation was used to influence the outcomes of last year’s #Brexit and the US Presidential campaigns, as discussed in  How is Big Data affecting the world we live in?

It seems to me that it is just a question of time before an unscrupulous consortium applies this kind of tactic to the markets.

How would the market react to the news that Warren Buffet, or a similar well-respected titan of the industry was recommending selling his stock in XXX large company, or sector?

Conclusion

Let’s assume, for now, that the algorithms do not try to predict sentiment, but act only on market activity. Initially, there would be no impact.

But as enough small investors start to react, the market may start to move. How would the algorithms react? if one or two large funds sell too, then the market would probably crash.

Are machines better at telling fact from fiction than people?

Discover effective strategies for encouraging executives to utilize analytics with insights from How to Get Executives to Use Analytics, emphasizing the importance of building trust in data-driven insights.

Understand the challenges of managing big data with insights from Why Is Big Data Hard to Manage?, exploring the complexities associated with handling large and diverse datasets