Oil price forecast from models: not a substitute for an investor's instinct


Change Weekly Data:

Data Source: EIA


"You don't need to be a rocket scientist. Investing is not a game where the guy with the 160 IQ beats the guy with 130 IQ." - Warrent Buffett



The crude oil price recorded its lowest ever value on April 20 this year, when it went negative for the first time in its history.

What it meant in practice, despite being relatively brief, the oil producers had to pay their buyers to store oil!

A few countries cashed in on the opportunity, including China, the US and India; for some, the lack of storage facilities and corresponding transport challenges got in the way, when the new-found fortune amplified the enthusiasm for going on a buying spree.

On one hand, as far as buyers were concerned, it was like discovering a fountain of fresh water in an arid land.

On the other hand, for investors and oil producers, it was their worst nightmare.

The investors and producers have been relying on model-based forecasts for decades, as that was the only way they can get an insight into the future of the market, apart from the analyses of the experts.

Neither the mathematical models nor predictions of the expert saw the oil crash coming. On the contrary, the predictions were a steady growth, punctuated by random geopolitical events in the Middle East.

The above chart, based on data from the EIA, the US Energy Information Administration, shows how crude oil price fluctuated over a period of decades; in recent years, you can see the sharp dip on April 20th, when the price crashed to an all-time-law.

No model predicted that dip; in future, modellers may even refer to it – in retrospect, of course – as a one-off blip; these models, however, cannot sweeps the flaws in modelling under the carpet, as the latter cannot be ignored easily.

Based on what you see in the above chart, I may be able to come up with mathematical formula to mimic its past behaviour of the curve – except the price crash on April 20th.

If I say my formula is reliable enough to forecast any future oil price on weekly basis, in the same breath, I claim to know every single parameter that determines the oil price at any given time in future – an impossible feat even if I know enough statistics and computer coding for the process.

This is the danger of relying completely on mathematical modelling; there are many latent factors at play: some modellers simply do not know they exist; there are others who cannot quantify them even if they vaguely identify them.

You ignore these undeniable facts and make predictions from a mathematical model – only to find your investments go up in smoke at some point, when you can least afford to happen on your watch.

It is not just models for oil price that received the wrath of the investors recently. Those which predicted Coronavirus deaths in the West are not very far behind them.

Of course, investors need a way to guess the oil price in future for their businesses and which makes perfect in economic sense. The results on many fronts, however, show that they need to take their instinct on board too before making disastrous mistake.

Predictions from the models and individual instincts are complementary, but neither is a substitute for the other.









Popular posts from this blog

Oil price closer to $70 a barrel: it's more realistic, sustainable and better than pursuing unrealistic goals

Oil Price and Lingering Suez Factor

What caused the downward trend in crude oil price?

Crude Oil: latest news