Publications & Downloads

A selection of papers related to the technologies used by TFL.

Informative Statistics for Loan Portfolio Management
A Discussion Document

The Revised International Capital Framework, known as Basel II, has placed a regulatory responsibility upon lenders to manage current expected losses within their loan portfolios. Basel II also requires estimates of these for up to one year ahead.

"The paper identifies a methodology for minimising loan portfolio expected loss (EL), based on separable problems of minimising PD and maximising the performance of a collateral portfolio based on exposure, where exposure is generalised to include collateral “values” on a statistical distribution..."

Download the White Paper & Case Study

Marketing Decisions: Raising Standards
& the Marketing Response Modelling Tool (MRMT)

Published May 2006, this white paper outlines the TFL tool set developed to model the effects of all manufacturers’ promotional spending in an entire marketplace and optimise promotional budget spending in a competitive environment.

"Marketing issues in major organisations tend to be highly complex. They depend on market conditions, the actions of other participants, and are subject to non-linear effects such as media saturation. These complexities cannot be fully analysed using simple methodologies..."

Download the White Paper & Case Study

Forecasting Values of Commercial and Residential Property
Using Non-Linear Mathematical and Statistical Techniques

Presented at the 2002 Joint Glasgow Conference of the European Real Estate Society (ERES) and the Cutting Edge Conference of the Royal Institute of Chartered Surveyors (RICS):

Forecasting methods currently available to practitioners have been of limited use because of their inability to detect and model the past patterns underlying property data. Additional restrictions on accurate forecasts are imposed by the inherent linear characteristics of most models which fail to recognize the complexities of real property markets. Further complications arise from issues such as lack of reliable historical data, short data histories or temporally sparse data.

Property forecasts can be improved by employing 'associated' or parallel series in forecasting models. Methods are described for matching a model to the complexity of the target data series and combining forecasts from many models to reduce errors in final forecasts.

Accuracy statistics and examples of residential and commercial forecasts are included.

Read online

 

Forecasting
Theory and Practice

The aim of this paper is to explain some of the underlying techniques and background theory used by TFL to forecast data.

Property data series are frequently of short duration and alone, do not possess sufficient information for good forecasts. Successful property forecasts can be made from property data that is augmented by information from associated series. The combined series form a system known as a double embedding, which can be analysed by dynamical systems theory to achieve useful forecasts.

Read online

 

The Methods Employed by TFL
in Forecasting Processes

An academic background to appear in the book of the NATO ASI 2000 Multisensor Data Fusion Symposium:

Combining disparate information from multiple diverse sources, such as using multiple time series derived from economic or social factors to obtain better predictions of house prices, is a classic topic of Data Fusion. The topic of Data Fusion covers different levels of extracting information, from very low level signal analysis to high level strategic decision making. This paper demonstrates how techniques exploiting these different levels of data fusion analysis can be applied in the related sector of financial analysis. We use examples from options pricing, to market value prediction and portfolios, demonstrating how multiple sources, and multiple models can be combined to deliver better informed decisions.

Read online

Do you need help viewing these documents?

All TFL download documents are formatted in Adobe Acrobat. To get the latest version of Adobe Acrobat Reader, please click on the following link: