Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 106 views

    Michael Rees BestPractices - Oct09 - New - Handouts

    Uploaded by

    Abhishek Gupta
    This document provides best practices for modelling. It recommends that models be objectives-driven and designed based on sensitivity analysis thinking. Key aspects include selecting relevant variables, ensuring a logical flow, and using an appropriate level of detail. The document also emphasizes building error-free models with robust formulae, and implementing models in a compact and modular structure to minimize complexity. Sensitivity analysis techniques like data tables, random values, and risk analysis are presented as ways to test models.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Michael Rees BestPractices - Oct09 - New - Handouts

    Uploaded by

    Abhishek Gupta
    106 views10 pages
    This document provides best practices for modelling. It recommends that models be objectives-driven and designed based on sensitivity analysis thinking. Key aspects include selecting relevant variables, ensuring a logical flow, and using an appropriate level of detail. The document also emphasizes building error-free models with robust formulae, and implementing models in a compact and modular structure to minimize complexity. Sensitivity analysis techniques like data tables, random values, and risk analysis are presented as ways to test models.

    Original Description:

    Michael Rees FM

    Original Title

    Michael Rees BestPractices.oct09_new.handOuts

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document provides best practices for modelling. It recommends that models be objectives-driven and designed based on sensitivity analysis thinking. Key aspects include selecting relevant variables, ensuring a logical flow, and using an appropriate level of detail. The document also emphasizes building error-free models with robust formulae, and implementing models in a compact and modular structure to minimize complexity. Sensitivity analysis techniques like data tables, random values, and risk analysis are presented as ways to test models.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    106 views10 pages

    Michael Rees BestPractices - Oct09 - New - Handouts

    Uploaded by

    Abhishek Gupta
    This document provides best practices for modelling. It recommends that models be objectives-driven and designed based on sensitivity analysis thinking. Key aspects include selecting relevant variables, ensuring a logical flow, and using an appropriate level of detail. The document also emphasizes building error-free models with robust formulae, and implementing models in a compact and modular structure to minimize complexity. Sensitivity analysis techniques like data tables, random values, and risk analysis are presented as ways to test models.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 10

    Modelling Best Practices

    October 2009

    By Michael Rees
    mrees@palisade.com
    Michael Rees
    Professional Experience
    20 years
    Strategy consultant
    Equity analyst
    Independent consultant
    Palisade Director of Training and
    Consulting

    Qualifications
    B.A. in Mathematics and D.Phil. in
    Mathematical Modeling from Oxford
    MBA from INSEAD
    Certificate in Quantitative Finance
    Financial Modelling in Practice (John
    Wiley & Sons, October) [see Wiley.com
    for free .pdf of Ch 1; without the models]

    2
    The Two Schools of Everything
    Alan Greenspan Einstein

    I guess I should warn you if I Everything should be made as


    turn out to be particularly clear simple as possible, but no simpler
    you've probably misunderstood
    what I've said

    3
    What is a good model?
    Objectives-driven: Suitable for decision-making
    Allows scenarios and sensitivities that align with mind-set/thought-processes of
    decision-makers
    Appropriate logical flow (definition of outputs versus inputs!)
    Appropriate level of detail: one that decision-makers relate to
    Requires minimal time to understand
    Clear logical flow (e.g. left-to-right, top-to-bottom)
    Modular structure, and with input values separate to calculations
    Compact (related items close together, total audit path through is minimized, as
    few worksheets as possible)
    Use of formatting (borders, colours, bold, underline) to create transparency of
    structure and flow
    Avoids unnecessary complexity (macros, circular references etc)
    Error-free
    Correct base level calculations/formulae
    Correct calculations as inputs are varied
    No unacceptable implicit assumptions
    4
    Hot controversies
    Circular references

    Multiple worksheets (workbooks)

    Named ranges

    Global/local placement of inputs/modularity issues

    Macros/user-defined functions

    5
    Key Theme: Sensitivity Analysis
    Thinking (SAT)
    Selection of appropriate variables

    Appropriate level of model detail

    Building robust formulae

    Error-checking

    6
    Using SAT to Build Robust
    Formulae and Error-checking
    Basic Methods Using sensitivity analysis
    Robust formulae: Excel
    Test a trial formula through the full range of Manual approach
    values, and modify as necessary DataTables
    Error-checking: RANDs in place of fixed inputs
    Calculate same quantity in two or more ways
    Build the difference as an error
    DTS
    Apply conditional format to highlight if non-
    zero TopRank to quickly vary all relevant inputs
    individually
    Sometimes, group these error cells together
    to get quick overview @RISK to vary input combinations

    7
    @RISK Models: Robust ranges and
    relationships
    Distributions Relationships
    See other talk!!! Independence

    Parameter dependence (semi-


    dependence)

    Correlated sampling

    Variations
    Parameter linkage through lookup functions
    Correlated time series
    Contextual dependencies (e.g. continuous
    uncertainty within discrete scenarios)

    8
    Summary
    Model Design Model Implementation
    Objectives-driven: Error-checks
    Clear view of key decisions to be made, and Dual calculation routes
    their information and sensitivity-analysis Sensitivity analysis (Excel, TopRank and
    requirements @RISK)
    As simple as possible, but no simpler Conditional formatting of error cells
    Compact
    As few worksheets as possible (no linked
    Designed around sensitivity analysis workbooks ideally)
    thinking Modular structure, clear logical flow
    Selection of variables Related calculations together
    Logical flow and dependencies Clear formulae
    Level of detail Either numbers or calculations
    Short, transparent calculation steps
    Formatting >50% of modelling time!
    Named ranges: selective use
    Avoid: circular references and macros
    as far as possible

    You might also like