Will data science disrupt financial modelling?

Is the financial modelling profession dead, and if not will it be disrupted by data science like so many other industries?

The answer is a little more complicated than a simple yes or no.

Firstly, financial modelling is a profession that in my view should be disrupted by data science. It is very manual, takes a lot of time, is prone to subjectivity and errors and is performed in an environment that isn’t always ideal. But whether it should be and whether it can be (in the short term) are two different statements.

Microsoft Excel is dead, long live Microsoft Excel?

Will data science build models for us?

Reasons why Financial Modelling will be Disrupted by data science

  • Depending on the type of financial modelling being performed, it can be largely automated. I’ve automated detailed free cash flow forecasts and valuations with key inputs and VBA.  So this is very achievable and should be considered by any financial modelling centre, bank and financial institution.
  • Standardisation does exist in with respect to modelling practices and financial reporting standards. This helps to build financial model templates that can be adapted to different inputs and requirements. With standardisation comes the ability for AI to learn from models and build them.
  • Large-scale data exists for certain types of financial modelling for AI learning purposes – so that algorithms can not only build models, but keep them up to date and improve them over time.
  • Errors are costly, and a considerable amount is invested in reducing errors from financial modelling and Excel training to financial model audits.
  • Returns need to be optimised and it is well worth using AI to optimize returns for investors.
  • Multiple scenarios and sensitivities need to be run, and this is sometimes does using Monte Carlo analysis.

But that’s just the good stuff.

Machine learning is already better at forecasting than we are!

Reasons why Financial Modelling won’t be Disrupted by data science

  • Certain types of financial modelling, for example, infrastructure and project finance modelling, is highly bespoke with very few public models for comparison and learning purposes.
  • Certain financial structures require unique modelling and structuring, especially jurisdiction specific modelling.
  • Models are sometimes exceptionally different and unique, and employ different methodologies, so they are difficult for algorithms to learn from and generate automatically.
  • There is an unwillingness from investment bankers to outsource the model building to a computer, as understanding the financial model is key to understanding the deal, especially what the key drivers of a deal are and what is negotiable.
  • Financial model templates do exist which can form a great starting point in the financial model building process. Therefore automation beyond using these templates is difficult as this is where the model customisation comes in.

So the answer to whether financial modelling will be disrupted is very dependent on two factors:

  1. What is the type of financial model is being built – a corporate valuation model and an infrastructure finance model for a wind farm are very different beasts.
  2. What time-frame are we talking about? Eventually, we expect all disciplines to be disrupted to some extent, however, in the short term large-scale disruption of financial model is probably not the lowest hanging fruit for data scientists to demonstrate value in corporations.

In the meantime, financial modelling, and more specifically financial modelling in Excel is here to stay!

Good luck and happy financial modelling!

Matthew


One Reply to “Will data science disrupt financial modelling?”

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.