Welcome to the StratoDem Analytics Frequently Asked Questions page. After working through issues with many of the largest private equity funds, REITs, operators and developers, we've noticed that some questions just come up repeatedly!
Here you can find answers to the most frequently asked questions about StratoDem Analytics, data science and our services. Questions are grouped by topic area, from general questions about the company and our clients to more specific questions about methodologies and products.
If we missed including your question, please feel to ask your question on the Contact Us page, and we'll typically send an answer to your question within 24 hours.
StratoDem Analytics is a Boston-based data science firm deploying advanced statistical methodology and machine learning to deliver geographic market intelligence for investors, developers, operators and advisory services clients in the real estate sector.
StratoDem Analytics helps clients to prepare for economic and demographic forces shaping market outcomes by building predictive models on massive economic and demographic data. Clients typically either deploy the StratoDem Research Suite or request custom data science work. We’re partnering with industry leaders, investors and real estate developers to help them make decisions faster, earlier and better.
After 15 years of strategy, research and predictive analytics work, the StratoDem Analytics founders StratoDem Analytics spent 18 months painstakingly engineering data pipelines for 200+ governmental data feeds, deploying machine learning to identify patterns that humans simply cannot see, developing statistically rigorous models and algorithms that recalculate instantaneously with every new data feed, then testing and back-testing the outputs.
We rely on three pillars for modeling and analysis:
The core product is StratoDem Analytics Suite, a powerful data-science platform that allows clients to analyze historical trends, track current developments with our nowcasts, and build confidence in future market performance with forecasts of demographics and economics for any market or submarket across the US.
Pricing depends on a number of factors, including geographic coverage, level of customization/build time, license terms, and segment coverage.
To learn more, please see our Pricing Information page.
Yes. StratoDem Analytics frequently builds what we call semi-custom deliveries, where we do some of the following for our clients:
We have served a wide range of clients including:
StratoDem Analytics does not typically provide trial periods.
With that being said, most of our clients have found that the most helpful way to assess StratoDem Analytics has been by comparing prior data to what can be generated by StratoDem Analytics. The most productive way we've found to achieve that is when clients send some of the past data (e.g., previous market studies, demographic or economic analyses), and then we'll prepare an analysis around that same site and walk through points of greatest difference online with you.
StratoDem Analytics provides complete coverage for the entire US (except some remote regions of Alaska), including:
StratoDem Analytics provides nowcasts and forecasts for critical economic and demographic metrics, including:
This is just a starting point for what StratoDem Analytics provides for clients, and we also frequently add, build, or integrate new data sets for clients during the initial on-boarding process or as more comprehensive semi-custom builds.
Don't see something you need on the list above?
We may already have it, but we also frequently add, build, or integrate new data sets for clients during the initial onboarding process.
Yes. Our Portfolio Analysis Engine can answer questions about metros or rank performance across all properties in a portfolio. Our machine-learning-powered analysts provide desktop access for you to ask questions or pull market scorecards for any market of interest.
Yes. For example, Our investment management clients use our bulk analysis tools to create portfolio definitions on the Research Suite to for queries about acquisition and disposition decisions.
StratoDem Analytics uses the same definition for household net worth as used by the Federal Reserve: total household assets less total household liabilities.
Households are composed of all individuals living at an address who consider it their usual place of residence.
Total assets are broken up into financial and nonfinancial assets
Financial assets include checking accounts; savings accounts; money market accounts; call accounts at brokerages; certificates of deposit; directly-held mutual funds, excluding money market mutual funds; equities; government bonds, excluding bond funds and savings bonds; IRAs; thrift accounts; future pensions, including currently received benefits; savings bonds; cash value of whole life insurance; trusts; annuities; managed investment accounts in which the household has equity interest; loans from the household to someone else; future proceeds; royalties; futures; non-public stock; deferred compensation; oil, gas or mineral investments; and cash.
Nonfinancial assets include vehicles; principal residence; land contracts the household has made; properties other than a principal residence; timeshares; vacation homes; current value less tax basis of active and non-active businesses; luxury and household items (gold, silver, other metals, jewelry, gemstones, cars (antique or classic), antiques, furniture, art, photographs, rare books, collectibles, musical instruments, livestock, horses, crops, wine, computers, equipment, and tools).
Total debts include mortgages; home equity loans; home equity lines of credit (HELOCs); land contracts; debt for other residential or vacation properties; other lines of credit; credit card debt; vehicle loans; education loans; other installment loans; loans against pensions; loans against life insurance; margin loans; and miscellaneous other debts.
Home values are based on machine learning models with data from multiple open sources, including Census Bureau microdata and FHFA hyper-local home price indexes. We also use some other third-party non-governmental open data sources to improve and validate our housing machine learning models.
Nearly all of our households metrics are broken out by age, including:
Yes. Many of our demographic metrics range back to 2000, with nearly all available from at least 2005.
Only StratoDem Analytics can calculate current market depth accurately by ZIP code/radius/drive-time, age, income, net worth, home value, among other factors in combination with each other. (For example, while most legacy-generation data providers cannot identify net worth beyond $500,000, StratoDem Analytics calculates the actual market depth for specific segments most likely to move into higher-end communities. Some legacy-generation data providers cannot generate counts of households beyond age 75, but for senior housing developers, StratoDem Analytics calculates market depth for the specific segments most likely to move.)
Only StratoDem Analytics has the data granularity and models to calculate local-market alphas and betas to determine how much of a region’s growth is driven by structural factors versus cyclical factors, which regions will most likely sustain or compress in a downturn, and which of the larger employment sectors in the region are more likely to grow, contract, or remain stable in a downturn.
StratoDem Analytics can help clients slice and dice by age, income, net worth, and a number of other factors in combination with each other (e.g., households age 80+ meeting specific levels of retirement income, net worth, home value, insurance coverage, etc.). Then StratoDem Analytics can map them by Census tract or generate segmentation heatmaps to quickly identify segments with higher or lower concentration than the US or a comparison market. One way we analyze the data is through a location quotient.
A location quotient is the relative index of a segment's concentration in one market compared to another market. (For example, if households 80 to 84 years old with $200,000 or more in household income living in Dallas-Plano-Irving, TX MSA have a location quotient of 6.65 relative to the U.S., then the percentage of households in that segment in Dallas-Plano-Irving, TX MSA is 6.65 times greater than the percentage of households in that segment nationally.)
Nowcasting is the prediction of the present state of data before it is released by the Federal governmental statistical agencies.
Example: StratoDem Analytics can predict 2017 regional GDP numbers (Gross Metropolitan Product or GMP) in January 2018, when the Bureau of Economic Analysis will not release the 2017 data until November 2018. The Goldman Sachs Growth Tracker and the Federal Reserve Bank of Atlanta GDPNow deploy similar methods for national growth estimates, although only StratoDem Analytics can release this data at local-market level instead of national level.
Why does it matter?
Every data set is released on a lag. For example: The 2017 American Community Survey population data is released by the US Census Bureau at the end of 2018. 2016 county-level income data is released by the US Bureau of Labor Statistics at the end of 2017. Household net worth data is released by the Federal Reserve Bank once every three years. Nowcasting creates accurate estimates of where that data will be before it is released, giving clients dramatic leads in understanding local market conditions before competitors – providing true competitive advantage.
Bayesian modeling is an approach to statistics that uses the mathematics of probability to combine data with prior information.
Why use Bayesian modeling? Bayesian models drive inferences which are more precise than would be obtained by either of those sources of information alone. StratoDem Analytics uses Bayesian modeling to combine hundreds of data sets at varying frequencies and geographic levels in a statistically rigorous, robust way.
As one example, the StratoDem Analytics engine builds its core population forecasts using hierarchical Bayesian time-series modeling with data from national, state, metro, county, census tract data (and more!).
The downside of Bayesian modeling? The models typically come at extremely high computational cost, which is why complex implementations are only now becoming possible in this era of scalable data science.