Why do we pay any attention to predictions?

Reuters reported today that Oil rises after IEA reports sharp decline in investment. This caught my attention:

quote:

---

"The IEA [International Energy Agency] said oil was unlikely to return to $80 a barrel before the end of the decade, despite cuts in investment, as annual demand growth struggles to top 1 million barrels per day."

---

The IEA predicts that oil won't likely return to $80/barrel before 2020, yet look at this graph combining forward looking projections for 2007-2010, with nary a hint of the sub-$50/barrel prices we're experiencing today: That witch doctors and prognosticators exist has far more to do with human gullibility than with any supposed talent for foretelling the future. The tools have become exceedingly sophisticated but the results have remained the same.--Percy

Couldn't agree more. I see this behavior as being most pervasive in the world of Wall Street. For decades (if not centuries), anyone with a shred of honesty in the financial realm will flat out tell you that no one can make effective predictions on how investments will perform.Despite that, we still have numerous stock analysts, financial gurus, prognosticators, etc that continue to tout the economic acumen as a means to lure investors to buy into their managed portfolios. Yet ironically, when the market turns south, they all throw their arms in the air and just say 'well, sometimes the market goes up, sometimes it goes down'.No one exemplifies this concept more than that caffeine factory Jim Cramer on CNBS. (No that's not a typo)
He has been on TV for years, hyping up his stock picks. Yet when one scrutinizes how well his portfolio has performed, he pretty much ends up doing what everyone else does: his trades barely matched the performance of the overall market and often times actually performed worse.Seeking Alpha did a nice breakdown of how Jim's 2007 stock picks performed: To summarize:

quote:

---

"In January of 2007, Jim Cramer made a total of 194 bullish calls and 123 bearish calls. Out of the 194 bullish calls, 60 are right calls. Out of the 123 bearish calls, 53 are right calls. The accuracy of Jim Cramer’s bullish calls is 30.93% and that of his bearish calls is 43.1%. The combined accuracy is 35.6%."

---

Yet despite the above, he still has a devout audience and continues to be showcased as a financial guru on a flagship financial network.Edited by Diomedes, : No reason given.

Despite the inaccuracy, I submit that the 2020 prediction still provides useful information. Obviously the prediction cannot take into account unknown factors. But the prediction does provide some appreciation for the changes in factors that are discussed in the article, namely the effect of decreasing investment due to the current poor economics and decreases in production.

I've been in the oil industry for 36 years now, admittedly as a chemist, not a financial guy. I get asked pretty often, "when is the price going to get back up to where we can make some money??"
I answer, "nobody knows and those that claim to know are lying or self-deluded." I've seen five crashes now, none of them predicted in the absence of other predictions predicting growth. I would like to see a little recovery soon to put some of my neighbors back into jobs, but I refuse to even guess when/if that will happen.Like you say, there are very sophisticated tools to do this, but GIGO still reigns over all.

Why? Well, I guess a lot of things are predictable. And it's so counterintuitive to think that our data are worthless. Coragyps speaks dismissively of GIGO, but that's not the situation, is it? We know --- well, economists know --- all sorts of things, with reasonable accuracy, which are both true and relevant. So, not garbage. How is it that having a good picture of the present state of the economy is worthless for predicting its future state? Dammit, that should be useful. Iff there was any justice.

The best predictor we have of the future is what's happened in the past and forecasting works(ish) so long as nothing structural changes in the system you're studying.Forecasting has been described as driving a car with no front window, being directed by passengers looking out of the rear window. The trouble is that you can't see the cow in the road from the back window.

Chaos. It's not sufficient to have a good picture - even assuming your models are 100% perfect simulations of the way the world works, you're still going to come up with incorrect predctions if you don't know the variables to plug in precisely. I saw a nice demonstration once with some of the best weather-predicting software in existence, in which simulations were played out simultaneously with only tiny, fractional differences in some of the variables. It was only necessary to play the simulations more than about 5 days into the future before the predictions diverged completely.I imagine the same thing is true of economics, except that our models aren't as good as those for meteorology.

Economic systems are much more stable than weather systems - I don't think you would describe them as chaotic. Certainly economic models just aren't as good as meteorogical ones which is a bit of a puzzle considering the amount of dosh to be made. Edited by Tangle, : No reason given.

Well, they are more "stable" in a sort of informal sense, which is why I'd have better luck predicting the price of eggs in two weeks' time than whether it'll be raining. But this is just to say that most of the time it changes more slowly.

I think you've just demonstrated the case. There is nowhere near the chaotic processes involved in egg production and markets as there are in weather events. Plus we know the majority of variables involved in egg markets and production. The unexpected events of things like salmonella and bird flu throw straight line forecasts off kilter periodically but they're not of the same complexity as weather variables. So, barring totally unforeseeable events, we can forecast egg prices with some confidence for long periods.

So our predictions are good so long as nothing unpredictable happens.Great. Also, my house is fireproof except in the event of a fire.

Correct. Unless, of course, the forecaster is omniscient.The argument is that egg production is a less chaotic system than weather and can therefore be more reliably forecast. Not that it can be totally known.

I can think of 2 reasons:1. Many people like to be the one who knows. It builds their ego or makes them feel important or worthy or something. These people would love predictions. I know what’s going to happen, and remember I’m the one who told you! I am the bringer of knowledge!2. A feeling of security. Knowing what’s going to happen can grant some people a reprieve from feeling anxious about an unknown future.I'm sure there's more. Anyone else?

Do you all think global warming forecasts 100 years out, climate change predictions for the rest of this century, are likewise without any value?

Sure, there's thousands. We need forecasts when we set out to build anything requiring infrastructure - new roads, airports, water supply, hospitals, the Internet etc etc. You have to plan ahead and build stuff for when it's going to be needed - planning ahead requires a demand forecast.I get asked every year to forecast the growth in the business in order to build the network and hire the right people so that we can meet demand when It occurs. I can do with an accuarcy of +/- 2% - barring a data centre or two burning down.