Location score by Empower Wealth

Discussion in 'Property Information Resources & Tools' started by SydneyInvestor, 28th Mar, 2018.

Join Australia's most dynamic and respected property investment community
Tags:
  1. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    Members can see a 3 year historical chart for all metrics including the DSR.

    However, I wouldn't put too much value on the change in DSR. The DSR is a measure of current supply & demand balance. It is already a lead indicator.

    When a market's supply and demand are out of balance, there is pressure on the market to re-balance. For every point the DSR moves above balance, there is more pressure on its direction to reverse and come back down to balance.

    I would rather buy in a market with a DSR of 70 even if it is trending down, than buy in a market with a DSR of 60 that is trending up. You don't know if 60 was the top DSR and from now it will reverse.

    Every month a trend extends is one month closer to that trend's end.
     
    Toby likes this.
  2. Toby

    Toby Well-Known Member

    Joined:
    19th Jun, 2017
    Posts:
    144
    Location:
    Melbourne
    Thanks Jeremy.

    Do you think there is anything to be said about comparing the dsr trend is they are at the same dsr score? And also if the suburbs dsr is the same, potentially the suburb that more recently got to that score could have potential to remain high for longer than the second suburb that has had a high dsr for a while?
     
  3. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    Yes, Toby, that sounds like a useful approach.

    Note however, that the DSR+ contains a metric that kind of already considers that case. It's called the MCT - Market Cycle Timing. It's an attempt to measure how likely a property market is starting its next growth surge. It's based on changes in prices recently and changes over the last 10 years.

    If growth has been below long-term averages for a long while, but recently there's been an uncharacteristic uptick in values, then the MCT score will be higher. The assumption being that the market is about to enter its next growth spurt.

    If however the property market has had double-digit growth in the last 4 years, it will have a very low MCT score. The assumption being that it has just finished its last boom.

    There's certainly no harm in checking how long a market has had a high DSR+ as you suggest. Looking at markets from many angles is the key to avoid misreading an anomalous indicator.
     
  4. Jeremy86

    Jeremy86 Active Member

    Joined:
    18th Jan, 2017
    Posts:
    25
    Location:
    Windsor, Brisbane
    Hi Jeremy

    Just picking up on this thread - interested to hear a little more of how DSR or Location Score can be used as a leading indicator.

    I've seen the 5min vid on the Location score website, which shows the top 250 suburbs outperforming the averages.However this is comparing current suburb scores versus previous years capital growth. Was there any back testing or demonstration of previous Location Score versus previous capital growth?
    For example can you see the 2013 top 250 suburbs versus the forward capital growth from 2013?

    Furthermore would also be interested specifically in what Scores were being shown at 2018 before the downturn in Syd and Melb.

    Cheers,

    Jeremy
     
  5. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    Hi Jeremy86,

    The video your referred to actually does show the performance of past scores. Those weren't current LocationScores, they were historical ones. However, I'll admit that wasn't made clear in the video. Perhaps this video makes it clearer...



    This video is a performance review of the DSR which is an improved version of LocationScore.

    Both the LocationScore and DSR have been scoring property markets every month for nearly ten years now. So, there's an enormous amount of history that can be assessed. The video above goes through that and summarises the overall performance up to the date of recording, which was early 2019.

    Note that data scientists derive algorithms from historical data. They reserve a portion of the data for testing the algorithm. So, without even using the algorithm in the real world, they can claim a "likely" performance if the algorithm were to be used in the future. However, the video above is not like that. It is a performance review of actual suburbs that were picked by an algorithm that already existed and was published back in 2010. These are real historical results not theoretical future ones. So, the LocationScore and its big brother the DSR are indeed lead indicators of capital growth.

    The average LocationScore for Sydney peaked at 73 (considered "Good") in March 2015. The first sign of poor future growth for Sydney came in March 2017 with an average LocationScore of 61 which is considered "Ok".

    The average LocationScore for Melbourne was more volatile. It peaked at 68 (considered "Good") three times: Aug 2016, Jan 2017 and Dec 2017. It didn't come down into the "Ok" range until January 2019.

    So, the LocScore suggested the end of growth for Sydney around early 2017 and for Melbourne around early 2019.
     
    Jeremy86 and Codie like this.
  6. Jeremy86

    Jeremy86 Active Member

    Joined:
    18th Jan, 2017
    Posts:
    25
    Location:
    Windsor, Brisbane
    Thanks for the further info and the vid.

    So the scores and success rates for those top 100 are taken from the beginning of each of those 3 year periods.

    I also assume that there would be plenty of other suburbs outside of the DSR top 100 that would have beaten the DSR average?
    But I guess your message is the top 100 provides you the best chance of getting this right, for the forward three years anyway.
     
  7. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    The scores are as they were at the start of a 3-year period. The growth is calculated as the change in values from the start to the end of that same 3-year period. We can only measure the performance after 3 years have passed since the scores were published.

    And yes, there would be suburbs that outperformed the DSR top 100 average growth rate. In fact, some of those suburbs are within the list of DSR top 100 suburbs.

    But yes, there would also be suburbs outside the top 100 that outperformed the DSR top 100 average and even some that outperformed the best suburb in the DSR top 100.

    Whether there are plenty of such suburbs or not comes down to your definition of "plenty". But it's not a bad way to look at performance. In other words, "how many great suburbs did the DSR top 100 miss?"

    A word of warning though, the reason why I average over 100 suburbs rather than look at individual cases is to avoid anomalies in median growth calculations. You may have noticed there are reports floating around listing the top growth suburbs for the last year or last 5 years or whatever. The vast majority of the top performers are quite often statistical anomalies with medians. Whenever you look at the top N of a data set without careful filtering, you're top and bottom 1% are more than likely "outliers", not true figures. You could easily obtain one of these reports and assume there were "plenty" of markets that beat the best of the DSR top 100. But the growth calcs are more than likely anomalies for at least the top 20 cases unless some very careful filtering is performed. Core Logic usually do a good job in that respect. Not many others do however.

    Here's a blog I wrote on this exact phenomenon...

    Misleading medians
     
    Jeremy86 likes this.
  8. fols

    fols Well-Known Member

    Joined:
    19th Jun, 2015
    Posts:
    737
    Location:
    Sydney
    Hey @datageek. I like the data driven approach & tools you provide. Just on the SUA trackers, are you able to please provide some thoughts on why all areas dropped so heavily in the latest NOV 19 snapshot? Major cap cities dropping some 5 to 6Pts which didn't seem to match what was happening on the ground. Appreciate any insights.
     
  9. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    I'd like to point out that a dip or jump of 5 points is not uncommon even for our largest significant urban area of Sydney. There could be a number of reasons. It's not uncommon to see dips around xmas/NYE. But I suspect November's data wasn't a sudden drop but a return to "normal" and that it was actually September's data that was "abnormally" high. Perhaps there was some excitement around spring. Or perhaps a bunch of mortgages on lower rates finally got approved.

    BTW, nobody ever truly knows the reason why a property metric changes. We could speculate that: the May election result had an impact on growth rates for the rest of 2019; or that APRAs restriction easing affected mortgage approval rates; or auctions were low in Easter because of holidays. All are plausible arguments. But we can never know for sure.
     
  10. Jeremy86

    Jeremy86 Active Member

    Joined:
    18th Jan, 2017
    Posts:
    25
    Location:
    Windsor, Brisbane
    Hi again Jeremy

    Signed up this morning so having a play around with the site.

    So with the variation you mention there from month to month, how much do the top 100 or 250 suburbs change each month and do many suburbs interchange or come in and out of the list?
    As this could obviously influence your shortlisting quite a lot depending on what month you were looking at....

    And if there was interchanging, the demonstrations you give of forward capital growth each month for the top 250 would actually be many many more than 250 suburbs beating the national average as there are many new suburbs getting added to the top 250 each month?

    Cheers

    Jeremy
     
  11. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    Jeremy86,

    Yes, the list does change a lot from month to month. However, that shouldn't matter for short-listing.

    According to historical calculations approximately 48% of the top 100 by DSR+ are not present the following month. This doesn't invalidate them a month later it just means that positions at the top are closely fought.

    There are more than 10,000 markets scored each month, the top 100 represents the top 1%. If a market's DSR+ drops from one month to the next by only a point or two, that's often enough to move it out of the top 100. They could be back in again the following month.

    If you were to take the top 250 instead of the top 100, the markets towards the top of the list are less likely to drop out from month to month. However, their rank in the top 250 might change significantly. A drop of a point might put them down 20 suburbs lower in the list.

    The point is, there are a lot of markets in Australia.

    I check the DSR+ historical chart for each top market looking for a recent one-off spike. If I see that the last few months have consistently been around the same score as the current month, then I'm more confident that market's current score wasn't a slightly higher anomaly.
     
    Jeremy86 likes this.
  12. What does the fox say

    What does the fox say Active Member

    Joined:
    4th Nov, 2019
    Posts:
    38
    Location:
    Sydney
    Hi Ben (& Jeremy), I've got a DSR lite subscription at the moment, and am trying to pick the relationship between the DSR graph and median price value - essentially I want to give myself the best chance of picking a suburb which is just about to hit it's CG straps; obviously there is no exact science to this, but I feel that using these tools is as scientific approach as it'll get.

    I shortlisted two suburbs in SA over the weekend; Old Reynella and Birkenhead. Old Reynella's median house price is very much the same as it was 3 years ago and is developing a little tail after a dip in 2019, whilst Birkenhead is a train wreck - almost 25% cheaper than what it was 3 yrs ago!!.

    So my question is, at what point does the DSR start to push up the purchase price? Because both suburbs DSR's have been impressive for the last 3 yrs. The way I shortlist is if it's above 75 DSR and it's been in that space multiple times throughout 36 months or not trending down quickly, then it makes the initial shortlist. Also, does the tool provide any mechanism to understand how much developable land is available or what developments have been approved in the last 18 months? Obviously, these are also factors that will affect CG, but not necessarily DSR scores. I've looked over a few suburbs in Brisbane where the DSR has rated well over the 36 months, but there's no CG to speak of yet, presumably because of the land supply component.

    DSR to Price Layer.PNG
     
    Spiralkut likes this.
  13. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    Hi "What Does the Fox Say",

    Both those markets have had impressive DSRs recently (demand to supply ratios). I would've expected more growth to date. But don't right them off just yet.


    To answer your 1st question:
    There's no exact point when we can say that prices should start going up in response to a certain DSR. All we can say is that there's more "probability" of prices rising, the higher the DSR is. In some cases prices don't respond at all, even with a really good DSR. There's always going to be some risk with investing and the DSR isn't perfect.

    Keep in mind that the basic DSR as seen in the Light membership plan only considers 8 variables. There may be dozens of variables affecting price growth for a particular market. The supe'd up DSR+ has 17 variables. One of those, checks for a flat period followed by recent price growth.

    In fact, the basic DSR has a 20% failure rate. By that I mean: out of the top 100 markets by DSR for any given month in the last 10 years, on average, 20 of the top 100 will failed to grow by more than the national average growth rate over the 3 years that followed.

    However, Old Reynella didn't pop into the top 100 until November 2018. So we still have till Nov 2021 before we can right it off as a fail. Birkenhead first appeared in the top 100 in May 2018.


    As for your other question:
    We're currently working on a project with RMIT developing an "infill risk" metric. It looks for vacant land which represents over-supply risk. Here's how I approach it manually at the mo':

    1) Switch between street and satellite views in Google maps to gauge vacant land for your suburb of interest. Ideally, you don't want any in the suburb or its immediate neighbours.

    2) Check Email alerts of planning applications near you | PlanningAlerts or forecast.id or the local council's website. Ideally, you want to find zero development and low population forecasts. Anything less than a pop. growth of about 2% pa is fine.
     
    Niche likes this.
  14. What does the fox say

    What does the fox say Active Member

    Joined:
    4th Nov, 2019
    Posts:
    38
    Location:
    Sydney
    Thanks Jeremy, really appreciate your response and thank you for your transparency on the tool. Based on those two SA suburbs, does the DSR+ provide any further insights into why there may not have been any CG to date?
     
  15. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    Hi WDTFS,

    Yes, there is a significant difference. The basic DSR is more than two standard deviations above average.

    Ctx Ruler - DSR - OLD REYNELLA 5161 SA Houses 2020-04.png
    That means it is exceptional by basic DSR standards. However, not by DSR+ standards...

    Ctx Ruler - DSR_PLUS - OLD REYNELLA 5161 SA Houses 2020-04.png
    Although the DSR+ still puts Old Reynella houses in the above average group, it's not as exceptional - only one std deviation above avg.

    It's a similar story for Birkenhead, but not such a dramatic difference.
    Both are still good markets using either metric.
     
    NickWCBA, Niche, craigc and 3 others like this.
  16. 2020 Property Investor

    2020 Property Investor Active Member

    Joined:
    13th Jan, 2020
    Posts:
    37
    Location:
    Sydney
    Hi @datageek

    New to this forum and found this thread interesting. Regarding the above quote, interested to know if the convergence of capital growth you studied occurred at a city or national level? Also when you say long term, are we talking 10, 20 30 years or more?

    If the convergence you mention is at a city level, then I presume this would indicate Sydney or Melbourne should really be the target for investors given their higher rates of growth over the long term compared to the other major capital cities?

    Thanks in advance
     
  17. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    Hi 2020PI,

    Comparing major significant urban areas (i.e. cities) across the country over a 30 year period, the data suggests that none of them outperform over the long-term. If you compare start and end values, it looks like there are clear winners. But if you examine the historical growth paths, you'll notice that the lead changes whenever one of them has a boom.

    This is the nature of compound growth - whatever price changes happened in the past, no matter how large they seemed to be at that time, a time comes in the future when those "massive" gains are dwarfed by relatively modest gains recently.

    It's a little harder to see this pattern at the suburb level, given smaller samples and anomalies in medians, but there's enough there to confirm it's the same story.
     
  18. Toby

    Toby Well-Known Member

    Joined:
    19th Jun, 2017
    Posts:
    144
    Location:
    Melbourne
    @datageek do you think you will soon have enough data to begin the SUA tracker again?
     
  19. 2020 Property Investor

    2020 Property Investor Active Member

    Joined:
    13th Jan, 2020
    Posts:
    37
    Location:
    Sydney

    Interesting thanks for the response.
     
  20. datageek

    datageek Well-Known Member

    Joined:
    18th Apr, 2017
    Posts:
    229
    Location:
    Australia
    Sorry for the tardy response Toby. Those SUA Tracker reports are available again on LocationScore. The latest one is there for June.