How to identify and eliminate statistical outliers?

Discussion in 'Property Information Resources & Tools' started by Lettie_S, 18th May, 2021.

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

    Lettie_S Active Member

    Joined:
    29th Jan, 2021
    Posts:
    33
    Location:
    Aus
    Say a property is sold 1.2 million in Chelmer in Nov 2020. onthehouse.com.au shows the median price has grown 16 percent since Nov 2020, give or take. So, it might be worth around 1.39 million now?

    https://www.onthehouse.com.au/suburb/qld/chelmer-4068

    But I suspect it's the beefed up increase in median price. All suburbs have statistical outliers, where a couple of 4~5 million sales will beef up the average/median price. (more prevalent in expensive suburbs)

    How to identify/eliminate these statistical outliers?
     
    Last edited: 18th May, 2021
  2. Erica

    Erica Well-Known Member

    Joined:
    19th Jun, 2015
    Posts:
    308
    Location:
    Adelaide

    I personally ignore any median prices published for a suburb if there has been less that 25 sales per quarter.
     
  3. Lettie_S

    Lettie_S Active Member

    Joined:
    29th Jan, 2021
    Posts:
    33
    Location:
    Aus
    Fair point, but the question still stands.
    If the suburb has had more than 25 sales per quarter, how would you eliminate those statistical outliers?
     
  4. thatbum

    thatbum Well-Known Member

    Joined:
    18th Jun, 2015
    Posts:
    5,792
    Location:
    Perth, WA
    You don't. Because median house stats are pretty much useless for any real DD for property investing. No point trying to polish that turd.
     
  5. Mark F

    Mark F Well-Known Member

    Joined:
    29th Jan, 2020
    Posts:
    1,029
    Location:
    Canberra
    A standard way in statistics to remove outliers is to remove those properties that lie more than 2 standard deviations from the mean/median.
     
    Lettie_S and devank like this.
  6. devank

    devank Well-Known Member

    Joined:
    18th Jun, 2015
    Posts:
    1,669
    Location:
    Inner West - Sydney
    You are confusing mean with median.
    Outliers will have higher impact on the mean (average). That is why property prices aren't expressed in mean.

    Median is the middle point after sorting the prices in assenting/decenting order.

    Say you have 100 properties sold.
    First you sort them by their prices.
    Median is the 50th property's price.
    An extreme value at the higher end wouldn't make much difference.
     
    Dave Radelaide, craigc and Firefly99 like this.
  7. The Y-man

    The Y-man Moderator Staff Member

    Joined:
    18th Jun, 2015
    Posts:
    13,443
    Location:
    Melbourne
    I know this is going to sound stupid - but I do property investing through "gut feel" (I have a big one for this purpose) and intuition. The reality is that experience gives you that intuition, and it certainly isn't 100% correct BUT I have found it more useful than trawling through stats...

    The Y-man
     
    Dave Radelaide likes this.
  8. Trainee

    Trainee Well-Known Member

    Joined:
    24th May, 2017
    Posts:
    10,260
    Location:
    Australia
    If you are looking at buying a particular property type in a small number of suburbs, you should know the market well enough not to need high level statistics.

    You should have a spreadsheet showing exactly how each house is different and have a good idea what is due to the market and whats due to the house.

    the other thing is dont calculate to the dollar how much you think a property should sell for. Choose to buy in a buyers market instead.
     
    Sackie, MTR and The Y-man like this.
  9. devank

    devank Well-Known Member

    Joined:
    18th Jun, 2015
    Posts:
    1,669
    Location:
    Inner West - Sydney
    In a normal data set, one standard deviation covers about 68% of the middle curve.
    Two SD, covers about 95%
    Three SD covers about 99.7%

    If I'm blindly removing end points, then I generally use 3xSD.
     
    Firefly99 likes this.
  10. boganfromlogan

    boganfromlogan Well-Known Member

    Joined:
    10th Jan, 2017
    Posts:
    3,332
    Location:
    Brisbane
    Wow! are there people that use stats for property investing?

    Property is about people. Income is rent from ..... drum roll ......... a person. Value is perceived ......... drum roll by people.

    We currently have spruikers using spreadsheets and stats and tax law to seduce mum and dad investors into buying into a set of shiny fittings, cobbled together with a shiny agent, and shiny spreadsheets. What is missing is land, bricks and mortar and ............. people.

    So i would suggest you remove all the standard deviants, the non-standard ones too, find an appealing property to a tenant (or yourself), beef up the land component, try some solid bones, and do the shiny reno yourself.

    Then outlying deviants, and other spreadsheet induced anti-social elements can be eliminated right at the outset.

    If none of that works ask @The Y-man . His method works well
     
    Dave Radelaide and The Y-man like this.
  11. Lettie_S

    Lettie_S Active Member

    Joined:
    29th Jan, 2021
    Posts:
    33
    Location:
    Aus
    I should have mentioned it would be a Primary Place of Residence, but I find your points well and truly valid.

    In saying that, as a future investor I should start from somewhere to develop my skills. So, tell me good Sir/Madam, if the following property was on the market now, what would you think a fair valuation would be now? It has been sold @ $1,775,000 in September 2020

    https://www.realestate.com.au/sold/property-house-qld-hamilton-133950774

    What are the very first steps? I will then try to expand my knowledge and develop some skills over time!

    Thank you :)
     
    Last edited: 18th May, 2021
  12. boganfromlogan

    boganfromlogan Well-Known Member

    Joined:
    10th Jan, 2017
    Posts:
    3,332
    Location:
    Brisbane
    I would have to offer my kidney, i suspect. I might have to throw in another body part, but my left leg needs work, so it might have to be something else.

    My speciality is in more modest abodes, with less snooty neighbours.

    Others may be able to help (and may have better organs to trade)
     
  13. Firefly99

    Firefly99 Well-Known Member

    Joined:
    24th Jul, 2020
    Posts:
    1,727
    Location:
    Qld
    Using a median generally removers outliers (well it doesn’t remove them but they don’t affect the result) - that’s why people use median over mean for things like property values.
     
  14. Lettie_S

    Lettie_S Active Member

    Joined:
    29th Jan, 2021
    Posts:
    33
    Location:
    Aus
    Point taken.
    How about this one?
    https://www.realestate.com.au/sold/property-house-qld-bilinga-133848550
    I just want to learn the first steps in valuing a property.
     
  15. boganfromlogan

    boganfromlogan Well-Known Member

    Joined:
    10th Jan, 2017
    Posts:
    3,332
    Location:
    Brisbane
    Seriously though, the quick way to finding 'true value' in a suburb IMHO is to attend a Public Trustee auction in the suburb. The sales price often represents a fair value of the land and the bones of a property (cos the finish is often lacking). Then add some for good fixtures and fittings. I attended a great one in 4007 early in the pandemic. Old Qlder ( I think someone is raising it now ) and it sold in 800s in Oriel St from memory.

    Does anyone else remember that?
     
  16. Firefly99

    Firefly99 Well-Known Member

    Joined:
    24th Jul, 2020
    Posts:
    1,727
    Location:
    Qld
    A couple of 4-5 million sales will not beef up the median price unless the dataset is very small.
     
  17. boganfromlogan

    boganfromlogan Well-Known Member

    Joined:
    10th Jan, 2017
    Posts:
    3,332
    Location:
    Brisbane
    that is a tricky one ..... although 'access to transport' is clearly a plus.

    Qld Globe has the land valuations, that should be a key input.

    Replacement value of the building is also key. In a bricks and mortar sense.

    Proximity to transport (good and bad) is a factor.

    Potential views is also worth looking at, but valuers tend to look at the 'most valuable use' of the existing property. ie. don't necessarily think about development potential unless that potential blows the present use out of the water.

    If your house was in the place where the runway was being extended (for example) just like the 'castle' then it might be useful to consider that. Good and Bad.

    Kinda loaded question really.
     
  18. The Y-man

    The Y-man Moderator Staff Member

    Joined:
    18th Jun, 2015
    Posts:
    13,443
    Location:
    Melbourne
    We pick an area out - typically somewhere we wouldn't mind living ourselves (and we do use public transport, commute, want to go to the supermarket etc) and then watch the sales of properties in our price range over a few months - in our neck of the woods that's typically auctions.

    Most importantly we look for properties that get passed in - that's when I know the market in that area is starting to saturate (more stock than buyers) and a good "base level" of pricing I can make offers on is set.

    The Y-man
     
  19. Trainee

    Trainee Well-Known Member

    Joined:
    24th May, 2017
    Posts:
    10,260
    Location:
    Australia
    What you should be thinking is. Here is a property. Compared to my list of 20 recent sales of similar properties in the same suburb, this is how its different, or the same. You should also know if the market is rising fast. But if you choose a buyers market this isnt a problem.

    you would know about how much it is. Plus minus a couple of %.
     
    Lettie_S, thatbum and Firefly99 like this.

Build Passive Income WITHOUT Dropping $15K On Buyers Agents Each Time! Helping People Achieve PASSIVE INCOME Using Our Unique Data-Driven System, So You Can Confidently Buy Top 5% Growth & Cashflow Property, Anywhere In Australia