On Thu, 03 Jul 2008 11:29:00 +0200, scott wrote:

>> I think both figures are useful - year over year (or quarter over
>> quarter) is something that "the street" looks at and affects stock
>> price.
> 
> Exploiting "noise" in the profit signal to try and make money is fine,
> but...
> 
>> It's the beginning of building a trend -
> 
> ...how can you use just one or two points of data to decide that a
> *trend* is starting?

It's two points of data in an overall view of what the company does.  
Only those two points of data isn't enough, you have to look at the 
policies the company has implemented as well.

>> and if the analysts use
>> it as one indicator of a company's direction (from a profit
>> standpoint), it makes sense to be aware of it and know where you're
>> going.
> 
> I don't think any real analysts look at solely the headline numbers to
> decide what to do.  They will look at the detailed result history to
> determine if the trend is changing, they can't do that with just one or
> two numbers.

Like I said, it's an indicator.  There is more analysis that's done, of 
course, when you talk about Wall Street analysts.  But an improvement in 
year-over-year (or quarter-over-quarter) is an indicator that something 
good happened in the quarter or year.  Then look at the trend over 5 
years - 5 points or 20 points, take your pick - and that *is* a trend.

Another indicator is the difference between looking at GAAP vs. non-GAAP 
numbers.  non-GAAP is generally a better indicator (as I understand it) 
because it reflects ordinary business and not extraordinary expenses for 
reorganizations or other (hopefully) one-time expenses.  So while a 
company may show a loss from one quarter to the next, if you look at the 
non-GAAP financials, you might see a trend of increased sales.  That's a 
good thing, regardless of the GAAP numbers which might include costs for 
litigation, acquisitions, and the like.

> For example, suppose your monthly profit fell from 104 last month to 90
> this month.  Some people then go off on one based *solely* on these
> figures, they send out emails to staff saying that profits have dropped
> by 13% and it's been the biggest drop for over a year blah blah blah",
> they try to "correct" whatever they think has gone wrong this month etc.
> 
> Then next month the profit is 107, they feel all good about themselves
> and maybe get a pat on the back from their boss if he is equally stupid.

Monthly is too frequently, I think you'd find most analysts would agree 
with that.  Quarterly or annually are the numbers that I typically see 
used.

> But, take a look at the monthly profit figures running up to the
> 104,90,107 incident:
> 
> 38,64,45,67,50,50,76,66,66,77,70,76,81,73,91,82,80,73,76,93
> 
> The clever person will notice that on average profits seem to rising at
> 5 per month, and try to find the reasons for why this is and how to go
> about increasing it.  The stupid person wastes effort on trying to
> explain and correct the change every month.
> 
> It's not just profits, but plenty of other things (like production
> volumes) that suffer from people trying to analyse the noise.

It largely depends on what you're trying to get out of the numbers.  But 
again, a monthly period is too short because of those variations.  I 
wouldn't decide a company was doing good because they had more sales on 
July 3 than July 2.  But I might decide a company was a good investment 
if I saw their revenues increased in FY08 over FY07.  Especially if the 
FY07 revenues (or more appropriately, the non-GAAP profits) were over the 
FY06 profits.

Jim

Post a reply to this message