Skip to content
Subscribe
Notify of
35 Comments
Oldest
Newest
Inline Feedbacks
View all comments
VoZangre
16 years ago

Data, data, and more
Data, data, and more data…

more or less as expected…

thanks, Rich.

a new loyal reader.

ps- can you make a graph with faux wood paneling AND faux crown molding, cos, ya know… I think that crown molding will vastly increase the worth of your data…

๐Ÿ˜‰

sdrealtor
16 years ago
Reply to  VoZangre

Great new additions to your
Great new additions to your reporting. I cast my vote for the Doug Henning Background.

Artifact
16 years ago
Reply to  sdrealtor

I don’t remember which
I don’t remember which thread now – it is referenced but not linked in thread referenced here (with the plot by waiting to exhale) – but I plotted out the inventory data going back to 2000 or maybe even 1999 using that same annual plot of inventory with each year plotted separately – I may still have the code for that if I can find it – I will look tomorrow. It really showed an interesting trend as I recall.

I am trying to think of a creative way to combine the two sets of plots so you get the yearly comparison and the housing type breakout. The tough part is keeping the figure readable – If I come up with something I will try and post a figure to get some input/opinion/etc.

bsrsharma
16 years ago
Reply to  Rich Toscano

Rich: In the wake of the new
Rich: In the wake of the new credit crunch reality, it may be interesting if you split the data set into two: Those selling for less than $521K and those that sell for more than $521K (A proxy for the ones needing Jumbos). I am wondering if the ones needing Jumbo are gliding down towards the Conforming line. After the meltdown in mortgage markets is complete, I am wondering if Jumbo loans will be few and far between.

Artifact
16 years ago
Reply to  Rich Toscano

Sorry – I was incorrect in
Sorry – I was incorrect in my post – I used data from the BMIT site to plot sales from 2005 to 2007 – I plotted NOT’s from 2000 to 2007 plus 1996 for reference – I cannot remember the data source for that one, but I found it either through this site or BMIT – Here are those plots – for inventory, 2006 and 2007 are similar (until now) and 2005 is different. It is interesting that based on the data I had, the peek in 2005 was later than 2006. The BMIT data did not go back earlier in that year though. For sales, it just keeps getting worse, not much else to say there. The NOT’s were the plot I recalled as the interesting trend. This years NOT’s started off similar to 1996, then got worse. NOT’s stayed basically the same from 2000 until midway through 2006.

Thinking more about it, I agree that to make plots like these and spearate the housing types needs to be on separate figures – otherise it is just too messy to read. I don’t have those data – if someone can point me at them I can work that up pretty easily I think. The more years the better!

[img_assist|nid=5059|title= Monthly Comparison of Inventory|desc=|link=node|align=left|width=466|height=441]
[img_assist|nid=5060|title= Monthly Comparison of Sales|desc=|link=node|align=left|width=466|height=441]
[img_assist|nid=5061|title= Monthly Comparison of NOT’s|desc=|link=node|align=left|width=466|height=441]

cr
cr
16 years ago
Reply to  Artifact

Great post Rich – nice data
Great post Rich – nice data too, but I’d really like to see the graphs in granite and stainless steel.

CricketOnTheHearth
16 years ago
Reply to  cr

I’m still chuckling.
Great

I’m still chuckling.
Great backgrounds. If you want, I have some nice grass, or a couple of kinds of uber-pretentious black granite and black marble you could use. Failing that, there’s the popcorn, paper clips, cherries, and so forth.

On a serious note, thanks also to Artifact for your graphs. Especially the NOTs one, I just have to say ‘wow’. This market really blew up in the last year and a half.

>chirp<

Artifact
16 years ago

I posted this figure in the
I posted this figure in the Sandicor thread, but it is relavent here – this is the monthly sales data (without separating housing type) back to 1999.

[img_assist|nid=5078|title= Monthly Sales Data from Sandicor|desc=|link=node|align=left|width=466|height=362]

AKguy
16 years ago

Love the puppy. As for the
Love the puppy. As for the shaded grey background, flat grey is OK by me. The data speak for themselves.

nccoastalseller
16 years ago

For the graph background,
For the graph background, can you put up picture of Gallagher (80’s physical prop comedian) smashing GC’s head with the sledge-o-matic.

hd
hd
16 years ago

Have you considered, or do
Have you considered, or do you have the capacity to do a scatter plot rather than a line plot? That is, plot house prices at a given square foot in month 1 vs. house prices at a given square foot in month N? This graph would tend to have a slope of 1 if the price was unchanged, and this slope could be compared from month N to N-1 etc. The slope would not be biased by the type of houses sold that month. It would even be possible to plot a few months on one graph (busy, but not impossible). I know it’s sexier to say the “median price is down 5%” rather than “the slope of median prices are down 5%” but I think it works out to the same thing, and it’s more accurate.

Artifact
16 years ago
Reply to  hd

HD –
I think maybe a way to

HD –

I think maybe a way to do that would be to plot each month as a year-over-year comparison for each size. You could only compare 2 years that way but I agree that it would be interesting. If there had been no change in pricing, the slope of the line should be 1 – otherwise the slope will be less or greater than 1 depending on the change and which year is plotted on which axis. I think to simplify it the best best would be to split the square footage into categories (e.g. 1000 to 1050 sqaure feet or 2700 to 2750 square feet).

This plot could be taken a different direction and you could plot zip codes year-over-year the same way. I think what you will see there is that the relationship is not linear – meaning at this point the lower end zip codes are changing faster than upper end – so the relationship would non-linear.

T

hd
hd
16 years ago
Reply to  Rich Toscano

The slope changes because a
The slope changes because a fixed percent increase will affect higher home values more than lower home values. If the increase is very non-homogenous (and the data very non-linear), the other suggestion of “binning” different home price ranges would have to be employed to make a quantitative (rather than visual) comparison, unless the data was a simple curve (like there being a steadily higher increase in the price of large homes vs. small homes).

If I’m wrong on this (math not really being my forte, and not having the data to check) then I think the x or y-intercept would change, rather than the slope.

hd
hd
16 years ago
Reply to  Rich Toscano

What I think would work (of
What I think would work (of course I have scatter plots on the brain) is doing the scatter plot and then looking for some kind of pattern… maybe you would see that houses in a particular price range are fairly linear and you could create your bin around that group. That way you would give your readers a nice, familiar “mean percent change in price” rather than slope change or scatter graph. If the linar part of the curve changes (translates up in price) from one year to the next, that’s OK, just change the bin. As long as you state that the bin has changed in the footnote, I think you’re still OK. To figure out the linear part of the data, you’d have to set yourself up with an r-squared threashold. [img_assist|nid=5149|title=example of scatter plots with linear curve fits, various degrees of certainty (Graphpad Prism)|desc=for r squared = 0 there is no linear relationship, for r-squared = 1, knowing X allows you with perfect certainty to predict Y.|link=node|align=left|width=466|height=297]

brandyn
16 years ago

Back in Time…
I’d be

Back in Time…

I’d be curious occasionally to see back-in-time comparisons on price (both inflation adjusted and nominal) just to see how quickly and how far we are traveling back in time in terms of housing valuations. Useful as a quick reference to anyone who entered or exitted the market in the last so-many years to know when they’ve reached parity with today.

On the backgrounds, the center gradient looks distractingly like a problem with my screen. ๐Ÿ™‚

Try some soft drop-shadows behind each chart…