Archive for the 'Forecasting' Category

Quote of the day

Published
by
Sam Stewart
on February 19, 2009
in Forecasting and Mathematics
. Comments

Essentially, all models are wrong, but some are useful. George E. P. Box

As somebody who spends a significant amount of time at work building and maintaining models I think it is important I remember this myself and remind stakeholders of the limitations and context of the model outputs.

Six Rules for Effective Forecasting

Published
by
Sam Stewart
on October 23, 2008
in Business, Forecasting, Harvard Business Review and Podcast
. Comments

The Harvard Business Review (HBR) IdeaCast has an excellent interview with Silicon Valley forecaster Paul Saffo as a follow up to HBR article Six Rules for Effective Forecasting (premium content).

Paul Saffo’s Six Rules for Effective Forecasting

  1. Define a cone of uncertainty - Visualise the uncertainty and try to encompass all reasonable possibilities
  2. Look for the S curve - Everything interesting behaves like an S-Curve e.g. Moore’s law
  3. Embrace the things that don’t fit - These anomalies/discrepancies maybe subtle indicators of future changes.
  4. Hold strong opinions weakly - Draw quick conclusions and  then systematically dismantle them. Forecasting proceeds as a sequence of failed forecasts, be your own worst critic.
  5. Look back twice as far as you look forward - Identify patterns, history does not repeat but it may rhyme.
  6. Know when not to make a forecast - There are situations where the level of uncertainty precludes the ability to make any sort of meaningful forecast.

Saffo illustrates these rules with examples from technology including forecasting the emergence of the iPhone and global events such as the fall of the Berlin Wall.

Additional important points Saffo raises include:

  • Differentiating between effective and accurate forecasting
  • Embracing uncertainty
  • Use the past as illumination not support
  • Don’t cherry pick history to support your conclusions
  • “Never mistake a clear view for a short distance” new technology will take time to become an overnight success

Two rules that I think are important for forecasters are:

Check your ego at the door

As a forecaster you will be wrong more often than you are correct. It is important to acknowledge and learn from your mistakes rather than denying them or trying to explain mistakes away.

Transparency

Utilise a transparent framework and engage stakeholders. Transparency will allow you to learn from your forecasts and allow stakeholders to understand the context and limitations of the forecast.

The podcast is definitely worth listening to if you are interested in forecasting.