Shit !!! Ponpon
I’m not sure what you’re referring to specifically, but I do know that “breadth” for the fundamental law needs to correspond to the same length of time you are using to measure active return and active risk (or tracking error). So if your active return is measured as monthly alpha and your risk is monthly SD, then breadth is the number of independent decisions made in a month. Usually active return and tracking error is annualized, so breadth is typically the number of decisions per year. No reason it has to be a year, however, unless of course you’re in GIPS-land. (oooh, another bad memory!)
Not really correct! They said “if breadth increases by 2 the IR increases by 2” - IT IS NOT correct, the IR will increase only Breadth^(1/2)
That was a two-parter, second part incorrect discussion of IR effect. I fear I missed breadth and therefore the question.
yes it was stated the number of decision per year, but i thought it was the number of stocks that you follow…maybe it is the same, mistake again… Ponpon
If your stocks move independently of each other (which could happen if you short out the market factor and possibly any sector effects), and you make one tradeable decision on them per year, breadth could be the number of stocks you are following. It’s not really one-to-one, but it is true that the more stocks you follow, the more breadth you have. The problem is that it is only the non-systematic part of their variability that counts for breadth. I don’t know how you would measure breadth directly. It’s really more of an academic concept that tells you how increasing the number of things you try to predict helps you in comparison to increasing the accuracy of your predictions. You can supposedly make a lot of money getting things right only 51% of the time, but it’s easier to do that if you’re following 10,000 stocks and getting things right 51% of the time than when you’re following six european market indices with the same level of accuracy.
it says number of active independent decisions per year
i thought it said it was the number of transactions the manager did which is obviously wrong so I wrote no no. anyone remember the question?
Ok, so I buy AAPL. It goes up, I sell AAPL. I make a bunch of money, and get a free iPhone with soft dollars. Is that one active decision or two? Are these independent decisions? If AAPL goes down first and I double down before it goes up and I sell again, is that 3 independent decisions, or 4 (two buy lots, and two sell lots). Are these independent? After all, it’s the same company. So, most of the time, I’ve seen the “breadth” number either “backed out” of an IR and R^2, or it’s just used “conceptually” to justify looking for alpha by increasing the number of assets under consideration.
yes whether the defination of breadth is right or no? BREADTH IS THE NUMBER OF ACTIVE DECISIONS MADE BY A MANAGER IN THE YEAR was the sentence
wait so that which manager had the highest info ration it was manager D right? (4/8)*SQRT(100)
I don’t remember any specifics, but if that was the sentence, it’s not quite right… the decisions need to be independent. If I had seen the sentence though, I probably would have said it was correct, and then CFAI would blast me because I left out the word “independent.” That would not have been the reverse of the null hypothesis that independence was not necessary. (joke)
why is IB only calculated for an annual period? what if i want to calculate the IR over a quarter?
Ummm, one other person in the thread had the same thought I did. I thought the statement was wrong because the person said if breadth doubled, the IR would double. There’s a square root in the formula, which makes the statement incorrect.
frisian Wrote: ------------------------------------------------------- > Ummm, one other person in the thread had the same > thought I did. I thought the statement was wrong > because the person said if breadth doubled, the IR > would double. There’s a square root in the > formula, which makes the statement incorrect. you are correct.
needhelp Wrote: ------------------------------------------------------- > why is IB only calculated for an annual period? > what if i want to calculate the IR over a quarter? I assume IB was supposed to mean IR. Calculated over a year… mostly convention, I think. Most rates are annualized, so this follows that convention. Also, IRs look better over longer periods (SD in the denominator increases with SQRT(time), versus the alpha in the numerator increasing in proportion to time). It’s a bit like gaming the sharpe ratio. So most people stick with annual.
so in short, it’s… 1) True 2) False …?
Sounds likely, but I don’t want to get that specific on a public forum.
wait so that which manager had the highest info ration it was manager D right? (4/8)*SQRT(100) ____________________ should we multiply by ^1/2(100)? I supposed that just simply AReturn/ARisk. It was A or B (0,7. smth)
ponponpq Wrote: ------------------------------------------------------- > yes it was stated the number of decision per year, > but i thought it was the number of stocks that you > follow…maybe it is the same, mistake again… > > Ponpon yeah i thought they were defining the IC not breadth?