HtAG (Higher than Average Growth) is a Real Estate Analytics portal that assists its customers in making accurate and timely property-related decisions. HtAG leverages the benefits of machine learning algorithms to rank the growth potential of different Local Government Areas and suburbs, Australia wide. Underpinned by advantages of Data Science and Big Data, HtAG processes large volumes of property statistics to classify the long-term capital value of a particular geographical location.
Here are the main differentiating factors:
– We have near real-time view of the market thanks to our Web Crawler solution (see DATA section below for details). Data lag in reports produced by other companies is at least 1 month.
– We provide detailed property market data. For example, we do not only report general median price values but further segregate the data based on dwelling type and number of bedrooms. This assists our clients when determining the appropriate ROI of an individual listing.
– We provide forecasted median rent values which assists our customers with making more informed decision as to the cash flow potential of a particular locality. Other companies provide only historical data.
– We provide forecasted median price values which assist our customers in better ascertaining the capital growth potential of a particular area. With our ‘ranking’ option, customers are also well placed to compare the investment potential of different areas based on their year-on-year value change as well as on their forecasted growth. Other companies provide only historical data.
– We provide growth rate cycle analysis data which combined with the ‘forecasted median price’ option gives you a more accurate understanding of the locality’s position in the growth cycle. Consequently, the ability to accurately forecast property values permit our customers to have a more informed understanding of whether a particular area has potential for further capital growth or is heading for a downturn. This can assist our clients with the timing of their purchase to take advantage of the cyclical movements of the market to maximise their returns. For example, buying in a declining market would mean that clients have to wait additional time until they realise the full capital gain potential, which limits their capacity to leverage into additional purchases in the future.
– Our service provides more value for money which is reflected in the pricing of our different subscriptions.
This question was answered in detail on our forum. Click here for an expanded answer.
You can purchase a subscription for as low as 16.99 AUD a month. See this link for more info.
We collect Australian property market data via a Web Crawler which systematically browses major real estate portals and agent sites in the country and indexes past sales, current listing and rental data. The collected data is combined with additional offline datasets, cleansed and analysed for key market trends to assist real estate professionals with market research. A portion of our dataset is in the public domain. You are welcome to explore it via this link.
HtAG crawler collects property sales, listing and rental data continuously and as a result, we are able to report the key real estate market metrics in near real-time. Our reports are updated as soon as data for a particular locality reaches statistical significance for the current calendar quarter and after new sales data becomes available thereon. On average the data is updated every fortnight.
HtAG reports are produced for the house (detached) and unit (apartment) segments. Some areas only have reports in the house segment due to no data in the unit segment and vice versa. There are ~430 area reports for houses and ~260 for units. ~150 areas are duplicated across the unit and house segments.
Our overall median price for any given suburb or council area can be calculated in 2 different ways depending on the composition of the underlying data:
– If there is significant number of sales recorded per number of bedrooms (i.e. 2,3,4 and 5 bedroom houses) within the area, we first calculate median values for the relevant bedroom time-series. The overall value is then produced by taking a mean (average) of all of the bedroom medians.
– If there is no significant number of sales to produce the number of bedroom time-series, the overall price is presented as a true median value.
Our customers find that the reported median values are of most relevance when they can be attributed to the number of bedrooms of properties within the geographical boundaries of their research scope. Therefore, we put emphasis on producing median values per number of bedrooms first.
Moreover, the overall value of a suburb is represented more equitably by averaging the bedroom medians, because this approach guarantees that each bedroom series is given an equal weighting. This eliminates statistical bias in markets dominated by a particular property type i.e. a 3 bedroom house. Other data providers commonly report median prices as a median value unbalanced through bedroom series averaging.
Where there is not enough historical sales data to produce median values, the last known median value is used to impute the data in the preceding quarters. For example if the sales activity within an area accumulates a significant number of sales in Q1 2018 with no prior sales data recorded, the median values between Q1 2008 – Q4 2017 are imputed using the Q1 2018 data.
Imputed historical values are common in new/developing markets with new land releases or recent surges in developer activity. Some rural locations with low sales volumes can also exhibit this data pattern.
Our forecast model works off a machine learning algorithm that predicts median prices with an accuracy of 95%, 2 years into the future. This level of accuracy is observed in areas with high volume of sales data.
Our accuracy is represented as an error rate. This question is best answered with an example. A suburb’s current median price is $500K with a 2 year forecast at $520K. 2 years into the future the actual median price is $530K. Use this formula to calculate the error: (actual-forecast)/actual *100 or (530K-520K)/530K * 100 equal to 1.9% rounded error.
To be transparent about the accuracy of the data in our reports, we have devised the confidence metric. High confidence strongly correlates with the number of sales recorded in any given locality. To simplify, the more there are sales – the higher is the accuracy. However, there are factors other than sales that influence this metric. Past model performance is one of them.
High and medium confidence indicate that the forecast error will be below 5% and 10% respectively. Low confidence data has an error rate below 15%.
As a rule of thumb at least 30 sales average per quarter at LGA (Council) level or 10 sales average per quarter at suburb level are needed for the model to produce high and median confidence forecasts. We simplify all of the above by allocating the confidence metric (High, Medium, Low) to areas and suburbs.
At HtAG, we divorce the analysis of causality between fundamental market variables such as population and unemployment and property price changes. By recording real-time movements in the property market, the plethora of causes that impact on fluctuations in property prices have already been accounted for within the collected Big Data that serves as an input into HtAG’s algorithms. At HtAG, any market changes become an instantaneous input into the HtAG prediction model ensuring the continuous agility of our predictions and alignment with existing market conditions.
Our philosophy is simple: apply a methodology that produces the best outcome. We conducted numerous tests and found that external variables (used as regressors) have no material impact on the accuracy of our model. Here are some of the variables that were tested: Population, Unemployment, Building Approvals, Stock on Market, Migration etc.
Instead of relying on external variables, our algorithm learns off a combination of historical and recent — days and weeks — sales data to produce the median values and forecasts. Recent data is given higher weightage than the historical data for forecasting. In addition, sales data from neighbouring localities and/or parent LGAs is inputted as regressors. This approach was tested rigorously and was proven to produce the best results. Click here for an expanded answer.
It is true that there is no “crystal ball” technology to predict the future, but because of the cyclical nature of property markets, it is possible to model the future with a high degree of accuracy. We are able to model and forecast property market trends with a high degree of accuracy thanks to our large dataset, which is continuously updated by our crawler solution. This question is answered in detail with examples on our forum. Click here to view it.
Our growth rate cycle (GRC) graphs plot the YoY rate of change and not the actual median price. The decreasing values on the graph that are above the 0 X-Axis (thick red line) indicate that the rate of growth will remain positive but significantly lower than the growth in the preceding years. In contrast to that, an increasing trend that is below the 0 line, indicates that the prices are declining but at a slower rate. First scenario highlights that the market has reached its’ peak and the growth rate is beginning to slow down. Second scenario is representative of a declining market that has the potential to reach its’ bottom in the near future.
We did our best to recognise typos and phonetic searches. In a rare case when you can’t see your locality in search results, try searching for a shortened version of the name i.e. “Druitt” instead of “Mount Druitt”. Note that our reports are produced at LGA level with suburb level data embedded within, so the search result returned for “Bondi” will be titled “Median Price Forecast for Houses in Waverley Council, NSW”.
There is property market data for ~6,200 suburbs in Australia. If you can’t find your suburb, it means that there was no statistically significant number of sales recorded for that locality. At a minimum we need at least 2-3 sales per quarter to be able to calculate the median price for a suburb. If a suburb does not have the minimum required number of sales, it is absent from our ranking tables. Nonetheless, your search will take you to the Council Area page, allowing you to lookup similar data for neighbouring localities. We are working on a feature that will allow you to examine data for suburbs with low sales activity in the future.
Local Government Area is an administrative division of Australia that a local government is responsible for. LGAs commonly have the same boundaries as council areas. On average there are 10-15 suburbs within an LGA. However some can have as little as 1 or as many as 70 suburbs.
Rental yield is the value you generate from an investment property represented as a percentage. High yield equates to a greater cash flow. Our customers use the rental yield metric to evaluate the expected income from investing in a particular location and to compare locations in terms of cash flow.
Our reports show the gross rental yield, which is calculated by dividing the relevant median price by yearly rental income, multiplied by 100. The higher the percentage, the better. As a rule of thumb, most investors seek rental yields of 3% and above. High rental yields (7% and above) indicate that the property market is under-valued, whereas low rental yields highlight potential for an over-valued market.
Not at this stage. However, if you are assessing an actual property, feel free to examine the metrics reported for the relevant suburb and number of bedrooms.
We are working on a feature that will enable you to generate PDFs of all pages on our site. Contact us if you need a council area or suburb report generated in the meantime. Cost is 175AUD, discount applies if you have an existing Professional subscription with us. You are welcome to co-brand the report with your logo. Here is a sample report.
Yes, it is. Contact us for details.
Yes. Contact us for details.