If we were to model random behavior of stock, whether the stock generates a positive daily return of 10%, which random variable would best depict this phenomenon? I was thinking a uniform random variable (continuous)?
Is this a mathematical exercise or are you trying to model the 'real world'. If a mathematical exercise then first, you would need to answer a couple of questions about the return. For example, are you saying that the mean (average) return of the stock is 10%? Is there a given standard deviation for the value or is this just a one parameter random variable? In other words, describe the problem better.My thought is that it may be a continuous random variable? As the return on stock may yield a percentage which is not countable and thus not discrete. However, could it be a Bernoulli random variable? Because there are two outcomes, one is that the stock gives a return in excess of 10% in a given single day (which is a success) or that it doesn't produce a return in excess of 10% which is a failure. Also, these events are mutually exclusive as if one happens, the other doesn't. Am I on the right track?
Yes, I think the Bernoulli distribution would be most appropriate. Yes/No type questions are typically binomial distributions and the Bernoulli distribution is a binomial distribution for a single trial, i.e. in a given day.The question just merely asks, if you were to model random behaviour of stock, whether or not it gives a return in excess of 10% in a given single day, what random variable would be the most appropriate for this phenomenon.
The question also asks, if you were to use the binomial distribution to explain this phenomenon for trade activity over 1 year (only including weekdays), then what assumptions would you have to make and are these reasonable?
What im thinking is that in order for this to be a binomial distribution, the assumptions would be that n = 260 (260 weekdays) and that a p success occurs when the return of a stock exceeds 10% and a failure occurs when this does not happen and rather it is less than 10%. I would say that the binomial distribution is not reasonable to use due to the large value of n and when computing specific values, our actual equation would contain huge numbers. Am I on the right track with this?