Save content
Have you found this content useful? Use the button above to save it to your profile.
Lots of little lightbulbs

Collective forecasting taps the wisdom of crowds

4th Jan 2017
Save content
Have you found this content useful? Use the button above to save it to your profile.

Tom Herbert takes a look at a new innovation in the world of forecasting – harnessing independent judgements from a diverse crowd of people to improve forecasting accuracy and avoid groupthink.

In his landmark 2005 book ‘The Wisdom of Crowds’ James Surowiecki explored the idea that decisions taken by a large group are always better than those made by small numbers of 'experts', even if the individuals within the large group aren't particularly well-informed or intelligent.

This built on a theory from Charles Darwin’s cousin Francis Galton in 1907, who observed that the average of all the entries in a ‘guess the weight of the ox’ competition at a country fair was remarkably accurate – even beating the individual guesses of alleged cattle experts.

The essence of the ‘wisdom of crowds’ theory centres on how the average judgement converges on the right solution.

However, Surowiecki himself admitted that the crowd is far from fail-safe, with many factors influencing the accuracy of crowd judgement. These include how independent people’s decisions are from one another (the more independent the better the accuracy), the diversity of the crowd (again, the more diverse the better) and the amount of transparent information available.

While last year’s EU referendum and US election results generated conflicting opinions about the general wisdom of crowds, what was even more startling was the failure of traditional methods of forecasting results: polling data was wildly inaccurate, gloomy economic forecasts were disproved and even the bookies, that well-known bastion of predictive stability, were proved wrong.

Collective forecasting

This uncertainty had led to a surge in interest in forecasting tools and techniques, and one interesting innovation that has emerged over the past year is around harnessing the wisdom of crowds in what has been dubbed ‘collective forecasting’.

The idea behind the concept is to open up forecasts to a diverse crowd of people in order to improve forecasting accuracy.

One of the main drivers behind the collective forecasting movement is Australian firm percypt, a forecasting platform that aims to deliver better forecasts through collective intelligence.

The company started life as Almanis, a global prediction platform where users could stake points predicting the outcome of geopolitical or economic events that last year generated around 40,000 forecasts. percepyt is the private version of this, and is currently available free for up to 50 users (although there are plans to introduce a paid upgrade later this year).

The platform runs alongside more traditional forecasting tools as a basis for enquiry, with the decision maker in charge setting a series of questions and putting them to a range of employees and stakeholders within the business.

From an accounting and finance point of view this could mean any relevant estimate, so a company could potentially open their budget to the crowd to forecast what the actuals will be, or a finance director could set a question such as “what will the sick pay cost be in the next quarter?”

It could also be used for compliance – when a company audit is due go to the platform and ask your users what issues auditors will find and what severity will be attached to them, turning regulatory compliance from an annual event into a live, real-time monitoring exercise.

Participants are also asked to provide a narrative by commenting on why they have made the forecasts they have.

These forecasts can then be weighted to an amount of points (e.g. 1,000), with users competing to get on leaderboards and be singled out as the most accurate to potentially win prizes or bonuses. The platform is anonymous by default, but can be set up so names are visible.

Framing and posting questions ‘is an art’

Karl Mattingly, CEO at percypt told AccountingWEB that one of the keys to good collective forecasting is around the quality of questions posed.

“Framing and posting the questions is an art”, said Mattingly, “the better the questions the more engaged the users are and the better the forecasting accuracy”.

While Mattingly admits that is it is “early days” for their collective forecasting tool, they are starting to see their network developing as partners strive for improvements in forecasting accuracy.

So far Mattingly has found that two types of person enjoy using the platform: young, more junior staff who are tech savvy and are well-versed with social media, and CEOs and board members. Leadership in particular are keen on the tool as it is as a way of cutting through the layers of management to get a heat map of their organisation – as Mattingly puts it, to “virtually walk the shop floor”.

With percypt’s concept only a matter of months old it is too early to tell if their collective forecasting model will prove a success, but in this uncertain ‘post-expert’ world it certainly isn’t beyond the realms of possibility that businesses looking to get ahead of the competition may turn to their employees to provide that competitive advantage.

Replies (1)

Please login or register to join the discussion.

By thegreatgrumbleduke
05th Jan 2017 11:47

Its an interesting idea, I like it, but it seems like a lot of work to set up and run along side existing packages without guaranteed results. Will keep an eye on this though.

Thanks (0)