A while ago in a post on the dangers of forecasting, I modelled oil prices with a simple sine curve model. It fitted the recent data (from 1960) pretty well, and at the time captured the price peaks in the early 1980’s and around 2010. It was a bit of a joke – the idea was to press home the point that most (oil) price forecasts are pretty useless, and that even a simple sine curve could do better.
One thing the sine curve predicted was a decline in price from the peaks of 2010-2012. Up until 6 months ago, this looked pretty foolish. And then…
Recent events have started to make the sine curve model really pretty amazing – lo and behold, prices have dropped rapidly and significantly, and even faster than the model predicted. Almost like magic!
Here I’ve extended the model, with a linear base line to make it look even better, like this:
Of course, the point here is still not that Optimeering has discovered a stunningly effective way of forecasting the oil price (dammit), but rather to show that good story telling (in this case, the story being the magic sine curve) as a prognosis may seem effective (or even brilliant) but actually delivers very little value. I can concoct a story to explain the historical oil price; I can concoct a story to “predict” tomorrow’s oil price; I can even make that story look pretty clever via the use of a fancy function and some numbers. But it would not be much more than made up numbers. As I wrote in my last blog:
A reasonable sounding justification could easily be constructed for this forecast, built around a fundamental business or economic cycle argumentation, driven by supply side cycles (high prices driving innovation, low prices promoting existing resource use only until these are used up) with a lagging demand side response. However, it would still be rubbish.
Price forecasting can be a useful input into decision making in government and business. However, as with many things, buyer beware – and an over-reliance on a single number or prediction can only lead one way (and it’s not up).
And just to prove myself wrong… according to our model (hah!) the oil price will recover over the next months to somewhere around 70 USD/bbl, before continuing a decline over the coming 7 years or so to around 25 USD/bbl (2013 USD, that is). Then, just wait for the rebound! We’ll return to this in 12 months time to see how things have gone (and to see if I can make up a good story to explain why the model didn’t quite get it right but why next time it will…)