Periodic Reporting for period 1 - ImmAzing (Data Intelligence for Accurate and Transparent Expert Real Estate Valuation)
Okres sprawozdawczy: 2018-05-01 do 2018-08-31
Mortgage is a loan in which the house acts as a security. There is currently 7 trillion euro of outstanding mortgage debt secured against residential dwellings. In Europe, mortgage transactions require a valuation by a certified appraiser to establish a property’s value. Accurate and transparent property appraisal assures the lender that the asset is of adequate value to cover the outstanding debt should the loan default. Global Financial Crisis of the early 21st century highlighted instances where real estate portfolios did not match the valuations that investors expected. The lenders did not possess any documentation to support their property value claims and the investors were misinformed. To prevent the repeat of such a widescale build-up of non-performing loans, increasing regulation has been implemented that demands transparency in the property valuation process. Banks will be required to notify private consumers of their current Loan-to-Value figure on an annual basis. Regulators are working to avoid the spectre of valuation inaccuracies in the property boom of 2002-06 that preceded the collapse in market values during 2007-08.
Unbiased and independent real estate appraisals play a vital role in assisting individuals, businesses and governments to make informed decisions during real property transactions.
IMMazing is a comprehensive real estate valuation solution, developed by Data Science Services GmbH, that utilizes Automated Valuation Model technology (AVM).
The project addresses known shortcomings of the traditional property valuation process: long turnaround times, high labour costs, lack of accounting for the factors that make up the estimated property price, and reliance on valuers’ subjective judgement. DSS have collaborated so far with Austrian banks to establish that IMMazing can reduce valuation time and cost by 50 to 70%, remove the need for compulsory physical appraisal and satisfy the EU regulatory requirements through transparency of its results. The project’s software technology has been integrated into an online real estate broker platform to offer potential buyers a reliable insight into property values; such accurate and instant house appraisals have been previously available only to professional valuers.
IMMazing is an AVM solution that incorporates advanced Machine Learning algorithms in conjunction with statistical methods. Having tested the prototype in Austria, DSS is now considering deployment in other European markets.
Unbiased and independent real estate appraisals play a vital role in assisting individuals, businesses and governments to make informed decisions during real property transactions.
IMMazing is a comprehensive real estate valuation solution, developed by Data Science Services GmbH, that utilizes Automated Valuation Model technology (AVM).
The project addresses known shortcomings of the traditional property valuation process: long turnaround times, high labour costs, lack of accounting for the factors that make up the estimated property price, and reliance on valuers’ subjective judgement. DSS have collaborated so far with Austrian banks to establish that IMMazing can reduce valuation time and cost by 50 to 70%, remove the need for compulsory physical appraisal and satisfy the EU regulatory requirements through transparency of its results. The project’s software technology has been integrated into an online real estate broker platform to offer potential buyers a reliable insight into property values; such accurate and instant house appraisals have been previously available only to professional valuers.
IMMazing is an AVM solution that incorporates advanced Machine Learning algorithms in conjunction with statistical methods. Having tested the prototype in Austria, DSS is now considering deployment in other European markets.
The main technical activities have been related to understanding and benchmarking the Machine learning algorithms and we came to the conclusion that 4 components need to be taken into account in order to refine the results to be as accurate as possible: Indexes, Models, Source Data and Comparable Neighbourhood Segments. The model adjustments will yield significantly more accurate results – even during times of market volatility. We also studied in depth the most common Enterprise Resource Planning (ERP) systems to get a better understanding of how IMMazing could interface with them. After working with several banking partners, we noted that compatibility with a specific ERP system is not a necessity. The banks often work with a multitude of third-party service suppliers and have their own abstracted middleware that allows external software to communicate with their own internal systems. Our system would be another one that would plug into their overall corporate workflow interface but within a specific range of functionality.
In the commercial expansion we took time to analyze the pertinence of entering in different countries taking into account a set of factors: number of competitors, nature of them -if they are in a dominant position- size of the market, availability of data and the format in which data is portrayed, regulations, language requirements and presence of our current partners in the countries of interest. With all those factors in our matrix, we have decided to target Central European Eastern Countries (CEE) with a preference for Poland, Slovakia, Hungary and Czech Republic with a close observation to Germany and Romania as they look also very interesting for expanding our business. When doing the competitor benchmarking we did even get in contact with one competitor in Poland (cenatorium) to explore potential synergies as we approach to the market from different perspectives.
In the commercial expansion we took time to analyze the pertinence of entering in different countries taking into account a set of factors: number of competitors, nature of them -if they are in a dominant position- size of the market, availability of data and the format in which data is portrayed, regulations, language requirements and presence of our current partners in the countries of interest. With all those factors in our matrix, we have decided to target Central European Eastern Countries (CEE) with a preference for Poland, Slovakia, Hungary and Czech Republic with a close observation to Germany and Romania as they look also very interesting for expanding our business. When doing the competitor benchmarking we did even get in contact with one competitor in Poland (cenatorium) to explore potential synergies as we approach to the market from different perspectives.
Our main progress beyond the state of the art after carefully reviewing the state of the art, apart from the scientific edge that our connection to Academia entail, is the transparency factor. We have found several patents applications overlap some of the functional concepts of our product. However, there were none that matched the comprehensive functionality found in IMMazing, especially regarding the transparency, set as a requirement by Central European Bank and adopted for all the National central Banks across Europe. We deliberately give the users of IMMazing PRO access to an ‘appraiser’s mind and expertise’, and openly disclose the facts and assumptions under which our software runs.
Proof of the quality of accuracy our system render is that every quarter, we provide the official residential property price index to Austrian National Bank (OeNB) that is even communicated to European Central Bank (ECB), Organisation for Economic Co-operation and Development (OECD) and Bank for International Settlements (BIS). Before being published, our results were validated by the regulator, endorsing the quality of our data and the models used to generate the index. We have supplied this information since mid-2017. Previously, the OeNB indices did not allow for region-specific differentiation and could not provide. Consequently, it was impossible to derive any real estate market monitoring, market fluctuation results. In 2016, DSS was commissioned to completely revise the Residential Property Price Index for OeNB. Following the joint development of the OeNB project with Prof. Feilmayr, we began to run parallel system from Q1 2017 onwards, using IMMazing’s data and models. In Q3 2017, OeNB decided to accept and exclusively use our solution. We were chosen because DSS has the largest and most representative property dataset and our methodology was fully disclosed and examined by the OeNB.
Proof of the quality of accuracy our system render is that every quarter, we provide the official residential property price index to Austrian National Bank (OeNB) that is even communicated to European Central Bank (ECB), Organisation for Economic Co-operation and Development (OECD) and Bank for International Settlements (BIS). Before being published, our results were validated by the regulator, endorsing the quality of our data and the models used to generate the index. We have supplied this information since mid-2017. Previously, the OeNB indices did not allow for region-specific differentiation and could not provide. Consequently, it was impossible to derive any real estate market monitoring, market fluctuation results. In 2016, DSS was commissioned to completely revise the Residential Property Price Index for OeNB. Following the joint development of the OeNB project with Prof. Feilmayr, we began to run parallel system from Q1 2017 onwards, using IMMazing’s data and models. In Q3 2017, OeNB decided to accept and exclusively use our solution. We were chosen because DSS has the largest and most representative property dataset and our methodology was fully disclosed and examined by the OeNB.