A list of indicators calculated as proxies for illegal and predatory landlord behavior.  
The first phase in the pre-ownership phase (i.e. before the prospective landlord purchases a certain building) is property identification. Our hypothesis is that landlords looking to take advantage of vulnerable tenants to achieve higher returns on investment in a property will look for buildings with a significant amount of rent stabilized units in gentrifying neighborhoods. These buildings present maximal opportunity, as the eviction of rent stabilized tenants and subsequent rent increases (and eventually, unit deregulation) results in higher rents per individual, and the presence of the building in a gentrifying neighborhood results in higher market rate rents (which the deregulation of rent stabilized units allows for). Furthermore, a property with a high proportion of rent stabilized units may be sold at a lower price than a predominantly market rate building, as the anticipated returns with rent stabilized tenants are less than a building in the same neighborhood with market rate tenants. In sum, this type of property maximizes return because it may be sold at a lower price than its market rate counterpart, can potentially achieve the same returns as a market rate property after rent stabilized units are deregulated (often through the harassment of rent stabilized tenants), and has the potential to generate further returns based on anticipated rent increases of the neighborhood at large.
When attempting to measure the presence of this type of building in a landlord's portfolio, we divided the relevant characteristics into two categories: neighborhood characteristics and building characteristics.
Methodology: Neighborhood Characteristics
At the neighborhood level, we used two indices of gentrification to measure how the median level of gentrification for the landlord's portfolio compared to the median level of gentrification for the city at large.  The indices used (the Urban Displacement Project Index and the Renovation Index) capture different phenomena significant in measuring the extent of gentrification, resulting in unique insights generated from both. The Urban Displacement Project Index measures the social impacts of gentrification such as displacement and demographic change, whereas the Renovation Index measures the impacts of gentrification on the property market. Taken together, these indices provide an understanding both of how much new development and demographic changes (two of the critical components of gentrification) the portfolio being analyzed has seen.
Methodology: Building Characteristics
At the building level, the critical indication that a property is susceptible to landlord predation is the percent of units in the building that are rent stabilized. The higher the number of rent stabilized unit count, the larger the financial gain the landlord can net through harassing rent stabilized tenants out of their units. Our team measures these indicators in two ways: statically, through the percent of units in a portfolio that are rent stabilized, and dynamically, by measuring the decline in rent stabilized units over the past ten years for a given portfolio. The static indicator should be interpreted as the extent of the financial incentive a landlord has to harass rent stabilized tenants out of their apartments at present: the higher the amount of rent stabilized units, the greater the incentive. The dynamic indicator should be interpreted as the extent to which a landlord has already taken advantage of this financial incentive to deregulate rent stabilized units: the larger the percent change, the more the landlord may have already harassed rent stabilized tenants out of their apartments. This dynamic indicator is calculated using the landlord's current portfolio, and therefore does not account for changes in a portfolio that may occur over time.
When applying these indicators, it is important to apply some sort of baseline to understand the extent to which the indicator is an outlier from the standard case. Because our team does not have a full list of residential landlord portfolios, the baseline we used revolved around the full set of apartment buildings in NYC. For the static indicator, the baseline to compare to was the amount of units that are rent stabilized across the city. For the dynamic indicator, the baseline was the change over the last ten years in the amount rent stabilized units per building for all rent stabilized buildings in NYC; the standard deviation of this statistic was calculated as well to get a sense of variance.
It must be noted that these building level indicators on their own are not evidence of illegal or predatory behavior, as any number of unrelated reasons could generate these statistics (e.g. natural flow in and out of units, the presence of subsidies mandating a certain amount of rent stabilized units, etc...). These numbers must be combined with further indicators (such as the ones detailed below) and an on-the-ground understanding gleaned through conversations with stakeholders involved organizing tenants to conclude with more certainty that these indicators truly are indicative of a portfolio being taken advantage of for the landlord's financial gain.
Methodology: Listings coded for predatory landlords
Through conversations with stakeholders involved in tenant advocacy, we learned that the property listings of buildings of the type described in this section (i.e. in a gentrifying neighborhood with a lot of rent stabilized units) will often use specific descriptors such as "value add," "high upside," and "upside potential" to indicate to buyers that the property has potential financial value for a landlord willing to harass tenants out of units. Our team generated a list (REFERENCE APPENDIX) of rental properties currently for sale on Zillow and StreetEasy that are advertised using these search terms. While this list cannot support research into existing portfolios as the data collected is on properties currently for sale and therefore not in any portfolio, it is a list of properties tenant advocates and organizers should be aware as potential targets for future landlord predation.
Should historical data be acquired, the percent of buildings within a portfolio that were marketed using this coded language for predatory landlords can be generated, indicating the extent to which a landlord purchased properties marketed with the intention of harassing and removing rent stabilized tenants: the higher this amount, the more likely it is that a landlord has as central to their financial strategy the conversion of units through the harassment and displacement of rent stabilized tenants. As with the building level indicator, this indicator alone is not evidence of a landlord intent on harassing tenants out of rent stabilized units. However, taken with further indicators and evidence from tenants past and present, it can be a compelling piece of evidence in an argument about the extent of a landlord's predatory behavior.