Auquan Blog

Could we have seen this coming? A Tullow Oil Case Study

June 30, 2020 by Chandini

I recently gave a keynote at CogX on Adoption of AI in investment management. Post the talk, we spoke with quite a few discretionary managers looking to adopt more data driven processes in their investment decision making, while still keeping their subjective human touch. One common theme that emerged was how reliant every asset manager is on daily news and information flow to stay on top of their portfolio activity. And how flawed the current process is.

Let’s take the example of Tullow oil, which is a multinational oil and gas exploration company headquartered in London. Tullow's main production comes from Jubilee and TEN oil fields in Ghana. Typical news coverage about Tullow includes any direct updates from the company as well as any local news regarding Tullow’s operations.

On November 13, 2019 Tullow Oil (TLW) shares fell 27% after warning its production will miss targets following continued operational problems at its Ghana fields. Shares further plunged 60% on Dec 9 2019 to a 16-year low after the company surprised investors by slashing its production forecast for the fourth time in the year, due largely to lower production from the TEN and Jubilee fields in Ghana, scrapping its dividend and announcing that its chief executive and exploration director had left. The company’s market cap fell by more than 1 billion$ to 2.2 billion$.

“Could we have seen this coming using data science?”, one of our clients with a significant position in Tullow Oil asked us in a recent meeting. “Were there any early warning signals?”

We turned to our Knowledge Graph and Portfolio Monitoring tool to see what news the tool thought was relevant for Tullow over the last year.

Among others, our model ranked this story from Aug 2019 as highly relevant to Tullow. Lessons to be learnt from Ghana's excess electricity shambles The model was monitoring not only Tullow Oil, but also Ghana National Petroleum Corporation (GNPC), Ghana Gas in the news. These stories from late October to mid November were also ranked high.

Why is this the case and why is this important?

Using a knowledge graph, we are able to monitor news that directly mentions the company also it’s key people, products, subsidiaries, suppliers, clients, competitors etc. This, combined with ML algorithm to rank news, allows us to filter through thousands of news sources, across millions of news items, for news that are significantly relevant to your portfolio names, even if it doesn’t mention the company directly.

Our knowledge graph flagged Ghana National Petroleum Corporation and Ghana Gas as clients of Tullow and deemed it important enough to monitor. It turns out, Ghana’s agreement with Tullow entitles it to free deliveries of gas that the company extracts with crude from its fields, which is under the purview of GNPC and Ghana Gas. This is win-win for both, because it allows Tullow to dispose of large unwanted amounts of gas associated with crude oil extraction.

However, the first news states that Ghana has an energy surplus (in excess of 50%) as well as an overhang for gas, a large part of which has been contracted on a take-or-pay basis. This means that Ghana is contractually obliged to pay, whether the gas is utilised or not (in 2020, they are looking at annual excess gas capacity charges of US$550 - US$850 million yearly).

This implies that the country's contractual obligations would render some of Tullow’s (free) deliveries surplus to requirement. Low demand from GNPC meant Tullow would have to inject the unwanted gas back into the reservoirs, a process which would weigh on production.

Sure enough, when Tullow cut production forecasts in Nov and Dec 2019, it cited low demand from GNPC as one of the reasons, adding that discussions on increasing gas offtake were ongoing, because increased gas offtake will reduce the amount of gas being reinjected into the fields, improving oil production over time.

This is just one example of how extensive monitoring and comprehensive coverage of news can help you avoid significant losses to your portfolio.

Another client talked about a company that filed for bankruptcy in 2019 - three weeks before, there was a news of a court judgement against a lesser known subsidiary of the company for a few hundred million dollars. The client’s fund missed this news, because “it’s impossible to track online information for every subsidiary of every company in your portfolio, especially across sources no one’s looking at”. If noted, could have prompted the client to reduce their position and the subsequent losses.

Manually, yes, it is impossible to track information for every subsidiary of every company, at least in a economically sensible way. But data science can let you track not just the company but also complex, exponentially increasing web of other connections that can affect a given company (both directly and indirectly) - it’s key people, products, subsidiaries, suppliers, clients, competitors, operational locations, trademarks, lawsuits etc. Using data driven tools allows you to look at more data sources than you currently look at, and search for more keywords than you currently know to search for, ensuring you never miss critical information about your portfolio.

Share this Article: