Predicting the Future -- Not.
- Thread starter vaughan
- Start date
medians are unreliable because of one, simple factor.
example.
"House Prices are falling"
House prices in perth are falling, with the median slipping over 10% to it's lowest level since Sept 2009, at $454,000.
what does that say?
it says to me that larger, more expensive homes aren't selling, and that smaller, cheaper homes are the flavour of the year.
considering the lending environment, that's not surprising.
if you sell 10 homes under 350k, and 10 homes over 450k, your median is 400k.
but if you sell 15 homes under 350k and 5 homes over 450k, what happens to your median? same volume of sales, different median.
you need to look at a numebr of factors, not just medians.
we need to look at the volume of sales UNDER the OLD median versus the volume of sales OVER the OLD median - and compare it to the new figures - to determine if we have slipping prices or just a shift to the purchase of homes in a different price bracket.
there is no "one indicator" that will tell us all we need to know.
but, that doesn't make for newspaper selling headlines.
Hmmm... the median is the *middle* value, not the average. These two sets of numbers have the same median value:
1, 50, 200
49, 50, 51
Medians are better for property values than averages because one large sale (or small sale) makes no difference. What matters is the sample size though, and all stats are less reliable with small samples.
I'm looking at property values and rents so I can calculate capital growth and rental yield Both of these needs to be worked out using medians.
But in the end we can only work with the data we can gather, and sales data is the only data we can get. How is it possible to determine whether the median is skewed by a predominance below-median properties, or whether the values have actually declined? There has to be an assumption that the sample of houses over the price range remains the same, and obviously that assumption is invalid when sale numbers are low, or if something makes the sales skew. With regard to your Perth example, what makes you think the median value HASN'T really dropped? That to me is a better explanation than assuming that higher-value houses are not selling as much.
With regard to other indicators, yes I agree: but which? Population growth? Employment? Demographics? Distance from nearest Westfield? After we have chosen the sats we then need to put into a single number that represents the metric we consider? What formulae do we use for that? How do they need to be weighted?
When do all these number start becoming chicken entrails?
I don't know. I need to break bread with somebody who knows statistics and talk this stuff through.
Thats your problem right there. You should go for an older measurement unit such as chains or hands.
I realised that Vaughan (smiley, point), but it can be interesting nontheless..
How about 'Prediction Errors'?
Prediction Errors Neuroscience
How about 'Prediction Errors'?
Prediction Errors Neuroscience
obviously you are not going to sell all 15 homes at 350k. the variance will be significant. you will have some at 260, some at 290, some at 320, some at 340.
my point still stands. "middle value" does not represent an average, a weighted average, or other useful guides to gauging a large databases general trend movements. it just identifies the middle value......whoop de freakin doo.
my point still stands. "middle value" does not represent an average, a weighted average, or other useful guides to gauging a large databases general trend movements. it just identifies the middle value......whoop de freakin doo.
How often have you read here that someone here bought property X just a year ago and it has increased in value 8.5% since? How do they measure this?
Sydney is lifting itself out of it's slump with 6% increase last year. (Whatever) What metric did they use?
i assume its measured against what others are selling for in similar condition in the immediate surrounds.
6% could have been measured by dropping unit sales.....how TF would i know, Mr Fish?!?!?
6% could have been measured by dropping unit sales.....how TF would i know, Mr Fish?!?!?
Clearly it isn't median values you use if you accept such figures. Like every one else here I am curious to know what metric you use to gauge market sentiment?
At the risk of stating the bleeding obvious it SHOULD be like for like sale sans renos. Who publishes these figures?
I agree though that "median" is a rubbish figure. I just don't accept it in either a rising or falling market so you must understand that I would love to know a better metric.
This thread was prompted by the "success stories" post by redwing -- not redwing's post, but the stories themselves.
It's a bit pessimistic, so be warned.
I'm new to IP, having bought my first in September 2011. As part of that process I've got my hands on a lot of sales and rent data from 2001 to 2011 for Sydney and have started analysing it. (I'm a database developer and one of my skills is pattern matching and problem solving, so I'm just looking for patterns.) My aim is to predict the future.
I need to cut this story really short because I can go on and on... so:
Capital Growth
For a lot of Sydney metro from Umina to Casula to Penrith, the median house price looks like this:
/\/
Starting at 2001 (where my data begins) there is a peak around 2004 and a trough around 2008, plus or minus a year or two. Depending on the suburb, the last peak is either just below or just above the 2003 value. Some suburbs (eg, Granville) have already peaked and are dropping again.
For the period 2001 to 2004 properties almost doubled in value -- Granville went from 220k to 400k. In the slump between 2004 to 2008 values dropped about 15% (Granville $320k) which is not good but hardly a crash.
From what I can work out, the period 1998 to 2001 was basically steady growth, so anybody that bought before 2001 did very, very well. People that bought after 2004 have been waiting to get back to their buy price, poor sods, and those that bought poorly and paid above market are still out of pocket. People that bought at the bottom around 2009 have seen some capital growth (eg, Quakers Hill) but nothing like the good times just 5 years before, and certainly not enough to make bold negatively geared strategies look good because in many suburbs the growth has peaked and reversed a bit.
Rent Growth
In generat the charts for rental growth for houses (units I have not looked at yet) look like this:
_/
... with the growth starting around 2004. Those people that held on through the peak of 2003 have seen their values drop (and return) but rent yield has been constantly growing.
The Future for CG
The bad news: if you were to take the previous 10 years and assume something similar will occur for the next 10 years, then I would not recommend property as an investment, especially one that relies on capital growth.
The good news: It's been a good time to buy since 2009 and this may continue for a while despite the market in Sydney dropping in some areas.
I have been searching for pattens and started to wish I had 30 years data to look at, but then I realised that that would be no use because the world has changed and each of the booms and busts have been for different and relatively unexpected (for some) global events, not what might be called "normal" market cycles. By that I mean: the GFC (bad); China and the resources boom (good); the woes in Europe (bad).
As far as predicting the future: I have decided I cannot do it. All that I know with certainty is that forecasts that assume constant positive growth (of any magnitude) are laughable in the next 5 years.
More to come later. I may edit or delete this post entirely. Do not take this as financial advice. YMMV etc.
What are your thoughts on Australia's money supply? Is it being trending up, down or constant? I believe that in the long run it will dictate where assets will be at.
Hera ya go
Infaltion?
No more land! It's all gone...
Ohh, the famous economic clock BS
And a classic example of bulls on parade catching falling swords the 2 lone voices of dissent. It's easy though (except on this forum) to predict that any strategy relying on prices always going up will incur large losses eventually.
http://www.somersoft.com/forums/showthread.php?t=8435
Infaltion?
19-08-2006 My recollections are that inflation brings higher interest rates, which brings more inflation, which then brings higher interest rates which eventually leads to a recession and or market crash.
Sorry to bring such good news
No more land! It's all gone...
Ohh, the famous economic clock BS
Sorry folks, but what is a market "aberration"???
There is no "resetting" of the cycle, this is just changing history to fit in with a nice little orderly theory of how a market should be.
As for predictions, there is still demand for property, which will hold up prices for a little longer. People are still hearing all the "boom" stories and want to jump on board. "I'm gettin into real estate investing" is still the talk of the backyard BBQ, and of course seminars, newsletters, radio & tv shows will still try to take the last wannabe investor's dollar.
When those dollars run out, and investors are borrowed to there eyeballs, is when then market slows and interest rates start rising. Then it will be interesting to see what happens.
I remember a time when I could by 4 bed doulbe garage at Mt Annan for 220-249K and they were hard up finding takers. You could buy a house in the Sydney western suburbs for ~100K and get $160 rent.
Unless human nature has changed, and I have no reason to believe it has, those times will be back when RE's will struggle to find buyers, and owners won't be able to meet repayments, and interest rates will high enough for many not to be able to afford there IP's.
I've seen all before and I'm sure I'll see it again.
always imho
And a classic example of bulls on parade catching falling swords the 2 lone voices of dissent. It's easy though (except on this forum) to predict that any strategy relying on prices always going up will incur large losses eventually.
http://www.somersoft.com/forums/showthread.php?t=8435
A long while ago I talked about this issue on Somersoft. One thing I found was that autocorrelation showed a slight tendency for one up year after another, but that was lost if you looked two years ahead so it's pretty weak.
I have done some work analysing randomness in financial markets & I can say with confidence that for this sort of strategy to work you have to have control over many aspects, or at least information other than simply pricing. For example in 2011 almost all Murray/Darling towns went down in value after going up the year before. This was reported as being because of the governments plans to buy back water rights. Correlation or other pricing analysis cannot forecast this, & this is the main reason why property (and many other markets) cannot be forecast using pricing analysis alone.
But in the end my personal conclusion is that property offers many aspects that can be reliably controlled by the investor. Forcasting the housing market could never be as strong a strategy as buying property that can be profitably controlled.
Pricing analysis that works tends to use other information or exploit market structure. For example where a stock trades at one price in early futures trade & opens at another price on the stock market the difference can be exploited.