JakeBrodskyPE said:
Like any data-driven forecast, it starts with a chaotic system and it goes downhill from there. Please don't tell me how good you think weather forecasts are. I've tried to use them for piloting a small plane. If I had been able to rely on those forecasts better, I'd have been able to pay for a lot more flight time. I can't tell you how many times I've read forecasts for winds and temperatures aloft, worked out my wind vector and temperatures at altitude and discovered that I shouldn't have even bothered looking at the forecast.
Financial investments are only accurate over a very long term and only within an order of magnitude. The shorter the time span, the less accurate they tend to be. I've read and invested and I'll say this: If you know what a random walk is, then you'll know that most financial forecasts are barely worth the paper they're printed on. And we're supposed to believe what these people think the job market will look like in ten years? I wish they could be that accurate, but if you ask them, any with a shred of honesty will tell you that this is all an educated guess at best.
Believe what you want, but don't be surprised if things don't end up anything like what was predicted.
Jake, how good weather forecasts are depends on the time point and the specificity of the forecast. Weather systems are based on non-linear dynamical systems that are chaotic (i.e. sensitive to initial conditions), so any forecast will depend crucially on whether the initial conditions on which the forecast has made has changed. Forecasts within 1 to 3 days usually tend to be fairly accurate, but any forecast beyond that time frame will degrade in accuracy.
It's also worth keeping in mind what we mean by "accuracy", since any good forecast should be inherently probabilistic, and should include some measure of the probability or uncertainty of the forecast (e.g. confidence interval, standard error, etc.). For example, if the weather forecast is for a 60% chance of rain the next day, and it does not rain, then can you really state that the forecast is inaccurate? After all, there was a 40% chance that it will not rain, and that's a high percentage.
Of course, your response is that all forecasts are garbage, but I think that is a disservice. Forecasts can be very useful for decision-making, so long as the following are considered:
1. You're aware of the assumptions that goes into the model from which the forecast is made.
2. You're aware of the measure of uncertainty of the forecast.
3. You're modest and careful in how to interpret the forecast.
Of course, there are other considerations depending on the type of forecast, which I can't immediately think of for the moment.