Sunday, 12 December 2010

The secrets of modelling

Probably you have heard more than once that a model is nothing but a few numbers, additions and subtractions... and that's it. A good model is no more than that... but ordered and structured in a smart way.
Are there any secrets / good practices on this? Yes, sure there are.
To develope any kind of model you need THREE KEY ISSUES TO MEET:

This is something obvious. If you are trying to model something, whatever, you need to know the way it works. It doesn't matter whether you are forecasting sales, you are making a financial model with its financial statements, or you are estimating the deductions in a highway project due to non availability events. YOU MUST KNOW THE WAY THE THING YOU ARE MODELLING WORKS. The better you know it, the more suitable your model will be.
Example 1: If you are building a financial model, you must own a robust financial background, you must be fluent managing financial statements, and you have to keep in mind the way every single movement impacts in both sides of the balance sheet.
Example 2: If you are forecasting the costs of a 30-year-term highway project you have to be sure which are the main cost drivers, the way the traffic assumptions impact on them, which costs are fixed costs, which ones depends on traffic, which ones are related to the length of the road, which ones depend on the toll plazas...

As we have said before, a model is no more than a few additions and subtractions, but they must be ordered in a smart way. A model is always focus on solving a problem. Thus, the modeller needs to be able to detect the problems, isolate them, and solve them, and the solution of a problem will be the data for the next one, and so on, that is the way that "additions and subtractions complex" is linked.

The way a plumber uses its wrench or spanner is the way a modeller should use the modelling software. Although most of financial models are built in Excel, spreadsheets are not the only tool used for modelling. There are different sofwares used for different applications (RiskLink for Catastrophe Modelling,  VISUM for  Traffic Forecasting...) but Excel is probably the most popular software for non-technical models or for those ones that not require a GIS software.

If any of the previous items fail, you will probably not succeed in developing a good model. For instance, you can be a great consultant with a great financial background and a great command of Excel, but you will need to spend a lot of time understanding your client business in order to build a good business model for his company. The same goes for investment banks, for example those ones investing in Infrastructure Projects... they have skilled modellers with wide financial backgrounds, but they often have to trust external technical advisors to get reliable business figures (income, costs...)

On the other hand you can be a finance guru, or a professional with more than ten years of trackrecord and deep knowledge of your business, but if you do not command any tool to reproduce your historical results in a consistent way and to forecast you future figures, you will not be able to build any kind of model.

Why a Modelling Blog?

Throughout my career I have seen lot of models. Hundreds of them. Many different models: business cases, project finance, valuation, cost or sales forecasting models...
Some of them were incredible models, really good ones, with great outputs, smart and pertinent charts, easy to follow and greatly structured.
Some other were simply awful.
There are a lot of people in the financial sector with a great financial background... but with very little modelling skills.
Modelling is something that you learn when you need it. There are no modelling subjects in the universities, maybe a few, but there is no better way to learn modelling than building models.
As a result of this situation, the sector is plenty of people who make models that really don't know how to make them, and only the experience and the willingness to learn and to develope better and better models are the tools to improve your performance.
Once said that, it is obvious that internet is a great source for learning more things, looking for example cases that you can use in your everyday's work modelling. The aim of this blog is to be useful as a portfolio of different cases you may meet while modelling.
I don't consider myself an excel guru or a project finance master, so please feel free to contact me for any suggestions, corrections you may want to make.