CRGreathouse wrote:You need to take the geometric averages, not the arithmetic averages. So if you have a 50% loss followed by a 100% gain, followed by a 3% gain, you should average them as cuberoot((1 - 0.50)(1 + 1.00)(1 + 1.03)) = 1.009901... or about a 0.99% gain per day. (Arithmetic averages would make this seems like a 17.67% increase per day...)
But this will make the average even lower.
From 12/31/1999 to 01/06/2012 the value dropped about 13%, or 1.16% per year. Of course your choice of starting and ending dates can change things a lot, and I don't know if anything funny happened with the index in that time.
DynamiCAM wrote:I have attached my template. GEOMEAN in the f(x) menu hasn't given me a numerical value it says error.. not sure why this is either. I need an excel/math whiz on this!!
Denis wrote:Well, I see geometric mean as worthless...when working with numerous periods like 3036 in this case.
CRGreathouse wrote:> You'll get poor results, then. The geometric mean is the only sensible way to deal with increases of this type.
Agree, CR; but the OP "seems" to be inquiring about the effectiveness of averaging daily changes.
> Consider a savings account (bond, etc.) yielding a 3% return each year over a span of 30 years.
> The arithmetic mean gives an average return of 4.75%. How is that sensible?
Agree, sure not indicative of much; but the annual income remains $30 per $1000, of course.
> Even better, maybe you swich banks after 15 years to another offering the same rate.
> Withthe arithmetic mean you get an average return of 3.72%.
> Why did that change when you're making the exact same amount of money?
Well, ok, but no need to switch banks for that; you can just switch account numbers in the same bank
CRGreathouse wrote:From 12/31/1999 to 01/06/2012 the value dropped about 13%, or 1.16% per year. Of course your choice of starting and ending dates can change things a lot, and I don't know if anything funny happened with the index in that time.
DynamiCAM wrote:I was thinking Geometric as well. However, as you said, this would cause my "return" to be even lower. I know for a fact that I've heard in the last decade the S&P has returned around 0.4%-1%, and this number is w/o dividend reinvestment.
Erimess wrote:DynamiCAM wrote:I was thinking Geometric as well. However, as you said, this would cause my "return" to be even lower. I know for a fact that I've heard in the last decade the S&P has returned around 0.4%-1%, and this number is w/o dividend reinvestment.
But your time period isn't "the last decade." Change time periods even slightly, depending on what's been going on, can have drastic changes in the returns. The market was pretty hot the last of the 90's and peaked around beginning of 2000. It was up to 1400 something, and then went down under 900 in the post-911 crash. Comparing the last decade is starting in the middle of that big crash when the thing was lower, and yeah, we're back down to about that point with little return. But if you compare today to 1999, you're hitting when it was near the peak and there's going to be a loss.
And yes, it would be without dividend reinvestment, because an index doesn't have dividends. But there are absolutely different ways you can define a return. I get stuff off Morningstar and while it might be interesting to know exactly how they calculate stuff, I've always been more concerned with things on a comparative basis, and they already have risk numbers on there.
You're in finance and interning at Merrill Lynch and they've never taught you how to do this?
DynamiCAM wrote:I've been taught to think and do my own work which provides me a true understanding.
DynamiCAM wrote:I also calculated Jan 00' to Dec 31 2009' FYI. And no, they don't teach us pre-made tools created by companies for "investors" to aimlessly point and click and blindly assume correctiveness. I've been taught to think and do my own work which provides me a true understanding. That being said thanks for the condescending help you offered.
Users browsing this forum: No registered users and 2 guests