Quote from drjekyllus:
So we use the months where the earth is getting hotter as proof of global warming and we disregard the months where it got cooler.
No, because that would be equally unfair.
A minimum sample size would be at least 30 to fill out a normal distribution, so 30 months would be the absolute minimum to use if you were trying to gauge monthly increases or decreases.
I see how this game works. Can we apply this trading? We could keep the winners and completely write off the losers.
BTW, the Earth is 4.5 Billion years old, what is a good sample size?
It depends on what you're trying to predict, naturally. Just like the markets.
The NYSE is over a 100 years old, but you don't need to use all 100 years to predict, just as you don't need to use 4.5 billion years to predict what will happen to the Earth.