Applied probability in oil and gas exploration

A two-day course that provides a solid mathematical basis for the quantitative concepts and methods used in assessing uncertainty in oil and gas exploration.

This course is unique in its explicit focus on the quantitative aspects of the explorationist’s toolbox. It is designed so that the pace and level of mathematical detail can be adjusted to accommodate a simultaneous range of backgrounds, approaches and comfort levels without compromising its fundamental learning goals.

A first principles approach to the material aims to help participants develop intuition and understanding, which will support them in all their quantitative activities and not only those explicitly covered in the course.

Target Group: Assurance teams, portfolio managers and analysts, exploration managers, project managers, technical specialists.

Prerequisites: The course is designed for geoscientists and engineers and starts from fundamentals. Facility with high-school mathematics and recent experience working with probability and uncertainty in practice will be valuable.

Format: The course is taught through a combination of classroom teaching, abundant exercises and cases. Typically, courses are closed and restricted to participants from individual companies, which allows discussion of specific practices and relevant “live” examples, as well as tailoring of course content to suit specific needs and background knowledge. Open courses are also available.

Course content

Day 1, morning | RISK

  • Probability as logic and the logic of probability
    • Definitions, axioms, rules and representations
    • Dependence, independence and weakest link
  • Resolving uncertainty
    • Bayes’ theorem
    • Evidence – a powerful reformulation of Bayes’ theorem
    • Using Bayes to get the most from databases


Day 1, afternoon | UNCERTAINTY

  • Discrete distributions
    • Mean, mode, median and percentiles
    • Relationship between statistics, distribution properties and parameters
  • Continuous distributions
    • The central limit theorem: Why it works, when it works, when it fails
    • Log-normal distributions: The Jekyll and Hyde of uncertainty quantification


Day 2, morning | APPLICATIONS

  • Modelling
    • Best practice for choosing and parameterizing distributions
    • Using historical data and expert opinion
    • Definition of success and relationship between risk and uncertainty
  • Lookbacks
    • Conventional lookback methodologies.
      What works, what doesn’t work. What looks like it works, but doesn’t
    • Eliciting systematic bias and confidence parameters


Day 2, afternoon | ELECTED TOPICS
Participants on closed courses may choose one or two of the following topics. Alternatively, our instructor can also address individual issues or examples brought by participants.

  • Variability and uncertainty
    • Models for distributed properties (e.g. porosity)
    • Using Bayes to use sparse measurements to update uncertainty distributions
  • Complex traps
    • Managing the mathematics of complex traps
    • Modelling leakage, charge limitations and seal capacity
  • Portfolio analysis.
    • Portfolio outcome space
    • Risk metrics
    • Efficient frontiers: Maximizing reward for risk or minimizing risk for reward
  • Venture modelling
    • Modelling dependencies
    • Dry hole tolerance.
  • Advanced use of data from databases
    • Extending analysis of discriminatory power of scores to include statistical significance
    • Updating probability of working petroleum system using indices (including statistical significance)
  • Training probabilistic intuition
  • Decision theory.
    • Value and utility
    • Value of information

Feedback from earlier courses

“Instructor’s style makes a fairly heavy topic enjoyable to follow”

“Good balance of theory and practical examples and exercises”

“Complex messages brought across in a very comprehensible way”

“For an advanced course, it was very much appreciated that the instructor was able to take ALL participants along the way all the time”

“The instructor has an extremely rare combination of deep mathematical understanding and an ability to teach it in an enjoyable way.”

“He’s like a mathematical magician”

about the instructor

Graeme Keith has a PhD in applied mathematics from Cambridge University (2000) and coming up on 20 years’ experience applying quantitative modelling to business decision making.

Before joining Decision Risk Analytics, Graeme held a number of senior management positions at Maersk Oil, including Global Exploration Portfolio Manager, Head of Risk Management and Head of Strategy.

Graeme is a fellow of the Institute of Mathematics and its Applications and teaches and supervises at Copenhagen University and the Technical University of Denmark.

You can contact Graeme directly on