MIT Center for e-Business Securities Trading Of Concepts (STOC) Joint work with Nicholas Chan, Adlar Kim Andrew Lo (MIT Sloan Finance), Tomaso Poggio (MIT Brain & Cognitive Sciences, AI Lab) Presented by Ely Dahan MIT Sloan Marketing December 12, 2001
Agenda – Predicting New Product Success • Why Test? A stochastic view of NPD • New testing tools – The Virtual Customer Initiative • STOC – A novel market research method • Internal validation • External validation
Research Questions • How does STOC function in real settings? Web-based interface with media-rich product information • How accurate is STOC? Good internal and external validity • What does STOC actually measure? Individual preferences AND group preferences • How do STOC traders decide when to buy and sell? Personal preference, expectations of group preference, learning • For what product/service categories is STOC effective? Seems broad: movies, candidates, bike pumps, cars, laptop bags • Does STOC work for Product Concepts? Attributes? Yes for both, but varies from case to case
Literature • Parallel & Sequential search: • Web-Based Market Research: Weitzman(‘ 79) Srinivasan, Lovejoy, Beach (‘ 97) Dahan and Mendelson (’ 01) Morgan & Manning(‘ 85) • R&D Options: Hauser & Zettelmeyer (‘ 97) Baldwin and Clark (‘ 00) • Organizing Design: Wheelwright & Clark (‘ 92) Des. Champs & Nayak (‘ 95) Leonard-Barton (‘ 95) Dahan and Hauser (’ 02) Willkie, Adams & Girnius (’ 99) Dahan and Srinivasan (’ 00) Dahan and Hauser (’ 02) Prelec (’ 02) • Simulated Securities Markets: Plott and Sunder (’ 82, ’ 88) Hanson (’ 95, ’ 99) Forsythe, et. al. (‘’ 90, ’ 92)
Resolving Uncertainty in NPD: What Type of Test? Marketing Attributes/Levels Price Sensitivity Brand Preference Design Aesthetics Architecture Design Manufacturing Number of Parts Mfg. Process Modularity Cost Ideation Price Brand Aesthetics Design Parts Process Unit Cost Concepts Price Design Price Revenues Mix Process Unit Cost Total Cost Design & Engineer Testing Profits Launch STOC VOC, Conjoint Concept Test DFMA Prototypes Test Market
The Three Extreme Value Distributions New products are the best (most profitable) of many potential ideas
Spending on Concept Tests Spending on NPD testing depends on shape of upside profit uncertainty
Impact of the WEB on Market Research Verbal m pu ta tio n Virtual Customer Research Adaptive STOC Co Conceptualization Media Rich Fixed Design Traditional Slow Market Research Communication Fast
STOC fits the Virtual Customer Initiative Fixed Experiment Adaptive Experiment Attributes Ely Dahan John R. Hauser Duncan Simester Olivier Toubia Ely Dahan Rob Hardy Limor Weisberg Full concepts Ely Dahan V. Seenu Srinivasan Leonard Lee CIPD N. Chan, Adlar Kim, Ely Dahan, Andrew Lo, Tomaso Poggio Ce. B
Are Web Respondents Representative? Game Markets • Iowa Electronics Markets (http: //www. biz. uiowa. edu/IEM) • Foresight Exchange (http: //www. ideosphere. com/FX) • Hollywood Stock Exchange (http: //www. HSX. com)
Multiple Experiments for multiple products Product Category Conjoint Features Bicycle Pumps Virtual Concepts STOC Concepts User Design 5 - - Crossover Vehicles 7 - Laptop Bags 10 (4) (3) Ranked STOC Actual Choices Attributes Purchases (2) Sources of data for internal and external validation
User Design (UD) Individual attribute choices lead to an ideal bundle
Web-based Conjoint Analysis (WCA) Bundles of attributes to measure tradeoffs
User Design as Conjoint Validation Attribute-by-Attribute “Hit Rates” CEB 1/16/01 and MBA’s 3/20/01 (n=130)
Virtual Concept Test (VCT) Dahan and Srinivasan, JPIM 2000 ABCD: Attributes + Brand + Cost + Design
Virtual Concept Test (VCT)
Gore vs. Bush Share of Vote Probability of Winning Source: Iowa Electronics Market (IEM), 11/7/00 at 2 pm
Hooray for Hollywood Source: HSX. com, December 7, 2001
Bragging rights to predicting the future … Source: http: //www. ideosphere. com/fx, December 10, 2001
Current game markets share some traits FX IEM HSX • All three predict actual future outcomes • Underlying reasons are not made explicit
Securities trading of Concepts (STOC)
STOC Outcomes vs. Virtual Concept Testing Median Stock Prices r 2 = 0. 79 Physical, 0. 87 Web r 2 = 0. 64 Physical, 0. 76 Web
Choice Out of a Set of Eight (Rank order) $24, 000 $49, 000 $37, 000 $42, 000 $29, 000 $36, 000 $30, 000
Securities trading of Concepts (STOC) Students valued BMW and Mercedes STOC highly, even though they “bought” them less frequently r 2 = 0. 60 r 2 = 0. 63
Securities trading of Concepts (STOC) r 2 = 0. 87
Securities trading of Concepts (STOC) r 2 = 0. 80
Three STOC Games for Laptop Bags Actual Choice 14% (n=330) 10% 32% 6% 5% 2% 8% 23% Five STOC Metrics Prediction was very bad in all three case - almost pure noise
Securities Trading Of Attributes (STOA) Red Handle PDA Cell Logo Flap Mesh Boot Large Will prediction improve if traders consider each attribute independently?
Prior Individual Estimates of Attributes 109 Survey 330 Actual Responses Choices 54% (± 19%) 61% 22% (± 20%) 25% 66% (± 22%) 71% 68% (± 19%) 83% 26% (± 16%) 25% 33% (± 18%) 34% 51% (± 22%) 58% 43% (± 22%) 59% 49% (± 22%) 75% What % of 330 first year Sloanies (in 2000) bought each of the nine upgrades?
Securities Trading Of Attributes (STOA) Large Red Logo Handle PDA Cell Mesh Flap Boot Did prediction improve if traders consider each attribute independently? YES!
How do individual traders rank? (laptop bag attributes) Inaccurate Winners Accurate Loser Accurate Winners High market-priced portfolios don’t always identify “accurate” traders
Do STOC sellers rank concepts lower than buyers?
Order Effect: Bottom stocks get traded less, at lower prices October 15, 2001, 1 pm (n=52) This suggests a need to have stocks randomly ordered by trader
Order Effect: Bottom stocks get traded less, at lower prices October 15, 2001, 10 am (n=57) This suggests a need to have stocks randomly ordered by trader
Which STOC metric is best?
Conclusions about STOC • Functions well with informed traders • Is accurate at identifying winners • Measures preferences in the aggregate • Traders behave heterogeneously, learn • Predicts well for many product categories • Can be effective with concepts AND attributes • Real outcomes are not absolutely necessary
Скачать презентацию MIT Center for e-Business Securities Trading Of Concepts
538d20d78ffd4c91b5d4dbcc202d6cf7.ppt