Confidence calibration in a multi-year geopolitical forecasting competition (CAC)

2017 ◽  
Author(s):  
Don A Moore ◽  
Elizabeth R. Tenney
Keyword(s):  
Forecast Accuracy ◽  
Data Set ◽  
Team Collaboration ◽  
Unique Data ◽  
Confidence Calibration ◽  
And Training ◽  
Over Time ◽  
Good Calibration

This research examines the development of confidence and accuracy over time in the context of forecasting. Although overconfidence has been studied in many contexts, little research examines its progression over long periods of time or in consequential policy domains. This study employs a unique data set from a geopolitical forecasting tournament spanning three years in which thousands of forecasters predicted the outcomes of hundreds of events. We sought to apply insights from research to structure the questions, interactions, and elicitations to improve forecasts. Indeed, forecasters’ confidence roughly matched their accuracy. As information came in, accuracy increased. Confidence increased at approximately the same rate as accuracy, and good calibration persisted. Nevertheless, there was evidence of a small amount of overconfidence (3%), especially on the most confident forecasts. Training helped reduce overconfidence and team collaboration improved forecast accuracy. Together, teams and training reduced overconfidence to 1%. Our results provide reason for tempered optimism regarding confidence calibration and its development over time in consequential field contexts.

scholarly journals Confidence Calibration in a Multi-year Geopolitical Forecasting Competition

2016 ◽  
Author(s):  
Don A Moore
Keyword(s):  
Forecast Accuracy ◽  
Data Set ◽  
Team Collaboration ◽  
Unique Data ◽  
Confidence Calibration ◽  
And Training ◽  
Over Time ◽  
Good Calibration

This research examines the development of confidence and accuracy over time in the context of forecasting. Although overconfidence has been studied in many contexts, little research examines its progression over long periods of time or in consequential policy domains. This study employs a unique data set from a geopolitical forecasting tournament spanning three years in which thousands of forecasters predicted the outcomes of hundreds of events. We sought to apply insights from research to structure the questions, interactions, and elicitations to improve forecasts. Indeed, forecasters’ confidence roughly matched their accuracy. As information came in, accuracy increased. Confidence increased at approximately the same rate as accuracy, and good calibration persisted. Nevertheless, there was evidence of a small amount of overconfidence (3%), especially on the most confident forecasts. Training helped reduce overconfidence and team collaboration improved forecast accuracy. Together, teams and training reduced overconfidence to 1%. Our results provide reason for tempered optimism regarding confidence calibration and its development over time in consequential field contexts.

Marketplace Lending, Information Aggregation, and Liquidity

2020 ◽  
Author(s):  
Julian Franks ◽  
Nicolas Serrano-Velarde ◽  
Oren Sussman
Keyword(s):  
Information Aggregation ◽  
Data Set ◽  
Information Efficiency ◽  
Liquidity Shocks ◽  
Unique Data ◽  
Retail Investors ◽  
Over Time

Abstract Lending marketplaces aimed at directly connecting retail lenders and borrowers retreat from auctions and, instead, set prices and allocate credit on their own, despite evidence that retail investors possess valuable soft and nonstandard information. We investigate this puzzle by analyzing a unique data set of 7,455 auctions and 34 million bids from a leading British peer-to-business platform. We find that the platform was vulnerable to liquidity shocks, resulting in sizable deviations from information efficiency. Deviations increased over time because of a growing role played by noncrowd players, particularly large investors and algorithms.

scholarly journals The Measure of Man and Older Age Mortality: Evidence from the Gould Sample

2004 ◽  
Vol 64 (1) ◽  
pp. 1-23 ◽  
Author(s):  
DORA L. COSTA
Keyword(s):  
Heart Disease ◽  
White Men ◽  
Abdominal Fat ◽  
Older Age ◽  
Data Set ◽  
Frame Size ◽  
Unique Data ◽  
The Past ◽  
Civil War Soldiers ◽  
Over Time

This article documents differences in body size between white, black, and Indian mid-nineteenth-century American men and investigates the socioeconomic and demographic determinants of frame size using a unique data set of Civil War soldiers. It finds that over time men have grown taller and heavier and have relatively less abdominal fat, implying that “modern” chronic diseases such as ischemic heart disease were common in the past. Changes in frame size explain almost half of the mortality decline among white men between 1914 and 1988 and predict even sharper declines in older age mortality between 1988 and 2022.

scholarly journals The Success of Business Forecasting: Comparisons across Industries, Countries and Time

2020 ◽  
Vol 13 (10) ◽  
pp. 108
Author(s):  
Tobias F. Rötheli
Keyword(s):  
Forecast Accuracy ◽  
Data Set ◽  
Electric Machinery ◽  
Forecasting Performance ◽  
Business Forecasting ◽  
The Stability ◽  
Changes Over Time ◽  
German Data ◽  
Historical Periods ◽  
Over Time

This study assesses the accuracy of forecasts by industry branches. Such an investigation provides a perspective on the relative benefits of forecasting in different industries. Accuracy of forecasting is assessed by econometrically investigating expectations data on firms’ production drawn from surveys covering manufacturing. Such data is available for only few countries and few historical periods. We study U.S. data covering the 1980s and German data over the period from 1991 to 2018. We first present rankings of industries according to forecast accuracy for both countries. Then the historical gap between the two countries’ data set is put to use to assess the stability and the dynamics in the relevance of forecasting in different branches of industry. We identify several industries that – across time and place – are among the most (e.g., electric machinery) and least accurate forecasters (e.g., the food industry). By contrast in some industries forecasting performance appear to undergo noticeable changes over time: the reported evidence suggests that forecasting has lost some of its potential in the printing and textile industries while gaining over time in the nonelectric machinery and in the metals industry. The findings can help management to make decisions regarding the allocation of resources to forecasting.

scholarly journals The The Shift in Firms’ Reliance on Debt Sources

2019 ◽  
Vol 8 ◽  
pp. 31-35
Author(s):  
Tu Cam Ho
Keyword(s):  
Capital Market ◽  
Public Debt ◽  
Structural Changes ◽  
Common Source ◽  
Bank Debt ◽  
Data Set ◽  
Private Placement ◽  
Financial Firms ◽  
Unique Data ◽  
Over Time

Structural changes in capital market and information innovations have altered characteristics of debt sources, make them more or less favourable to firms. This could possibly lead to a shift in firms' reliance on debt sources. Using a unique data set of debt mix of 1,100 U.S. non-financial firms, I conduct data analysis to reveal changes in firms' preference for different debt sources over a decade from 2004 to 2014. I find that bank debt remains the most common source of borrowing, followed by public debt and finally private placement debt. In addition, over time, firms have become more reliant on bank and public debt while less reliant on private placement debt. This pattern is consistent across different industries.  

Cloning Safe Driving Behavior for Self-Driving Cars using Convolutional Neural Networks

2019 ◽  
Vol 12 (2) ◽  
pp. 120-127 ◽  
Author(s):  
Wael Farag
Keyword(s):  
Gradient Descent ◽  
Autonomous Driving ◽  
Driving Behavior ◽  
Training Data ◽  
Stochastic Gradient Descent ◽  
Data Set ◽  
Safe Driving ◽  
Processing Pipeline ◽  
Self Driving Cars ◽  
And Training

Background: In this paper, a Convolutional Neural Network (CNN) to learn safe driving behavior and smooth steering manoeuvring, is proposed as an empowerment of autonomous driving technologies. The training data is collected from a front-facing camera and the steering commands issued by an experienced driver driving in traffic as well as urban roads. Methods: This data is then used to train the proposed CNN to facilitate what it is called “Behavioral Cloning”. The proposed Behavior Cloning CNN is named as “BCNet”, and its deep seventeen-layer architecture has been selected after extensive trials. The BCNet got trained using Adam’s optimization algorithm as a variant of the Stochastic Gradient Descent (SGD) technique. Results: The paper goes through the development and training process in details and shows the image processing pipeline harnessed in the development. Conclusion: The proposed approach proved successful in cloning the driving behavior embedded in the training data set after extensive simulations.

scholarly journals Microfinance and Vulnerability to Seasonal Famine in a Rural Economy: Evidence from Monga in Bangladesh

2020 ◽  
Vol 0 (0) ◽  
Author(s):  
Claudia Berg ◽  
M. Shahe Emran
Keyword(s):  
Food Security ◽  
Rural Economy ◽  
Screening Process ◽  
Data Set ◽  
Estimation Strategy ◽  
Positive Effects ◽  
Unique Data ◽  
Primary Mechanism ◽  
Poor Households ◽  
Microfinance Institution

AbstractThis paper uses a unique data set on 143,000 poor households from Northern Bangladesh to analyze the effects of microfinance membership on a household’s ability to cope with seasonal famine known as Monga. We develop an identification and estimation strategy that exploits a jump and a kink at the 10-decimal land ownership-threshold driven by the Microfinance Institution screening process to ensure repayment by excluding the ultra-poor. Evidence shows that microfinance membership improves food security during Monga, especially for the poorest households who survive at the margin of one and two meals a day. The positive effects on food security are, however, not driven by higher income, as microcredit does not improve the ability to migrate for work, nor does it reduce dependence on distress sale of labor. The evidence is consistent with consumption smoothing being the primary mechanism behind the gains in food security of MFI households during the season of starvation.

scholarly journals Microfinance and Vulnerability to Seasonal Famine in a Rural Economy: Evidence from Monga in Bangladesh

2020 ◽  
Vol 20 (3) ◽  
Author(s):  
Claudia Berg ◽  
M. Shahe Emran
Keyword(s):  
Food Security ◽  
Rural Economy ◽  
Screening Process ◽  
Data Set ◽  
Estimation Strategy ◽  
Positive Effects ◽  
Unique Data ◽  
Primary Mechanism ◽  
Poor Households ◽  
Microfinance Institution

AbstractThis paper uses a unique data set on 143,000 poor households from Northern Bangladesh to analyze the effects of microfinance membership on a household's ability to cope with seasonal famine known as Monga. We develop an identification and estimation strategy that exploits a jump and a kink at the 10 decimal land ownership-threshold driven by the Microfinance Institution (MFI) screening process to ensure repayment by excluding the ultra-poor. Evidence shows that microfinance membership improves food security during Monga, especially for the poorest households who survive at the margin of one and two meals a day. The positive effects on food security are, however, not driven by higher income, as microcredit does not improve the ability to migrate for work, nor does it reduce dependence on distress sale of labor. The evidence is consistent with consumption smoothing being the primary mechanism behind the gains in food security of MFI households during the season of starvation.

scholarly journals Back in Time for Breakfast: An Analysis of the Changing Breakfast Cereal Aisle

Nutrients ◽  
2021 ◽  
Vol 13 (2) ◽  
pp. 489
Author(s):  
Emilie Croisier ◽  
Jaimee Hughes ◽  
Stephanie Duncombe ◽  
Sara Grafenauer
Keyword(s):  
Dietary Fibre ◽  
Saturated Fat ◽  
Health Claims ◽  
Whole Grain ◽  
Breakfast Cereal ◽  
Full Data ◽  
Data Set ◽  
Time Points ◽  
Total Fat ◽  
Over Time

Breakfast cereal improves overall diet quality yet is under constant scrutiny with assertions that the category has not improved over time. This study aimed to comprehensively analyse the category of breakfast cereals, the nutritional values, and health claims across eight distinct sub-categories at four time points (2013, 2015, 2018, and 2020). An audit of products from four major supermarkets in metropolitan Sydney (Aldi, Coles, IGA, and Woolworths) collected ingredient lists, nutrition information, claims and Health Star Rating (HSR) for biscuits and bites; brans; bubbles, puffs, and flakes; granola and clusters; hot cereal flavoured; hot cereal plain; muesli; breakfast biscuits. The median (IQR) were calculated for energy, protein, fat, saturated fat, carbohydrate, sugars, dietary fibre, and sodium for comparisons over time points by nutrient. Data from 2013 was compared with 2020 (by sub-category and then for a sub-section of common products available at each time point). Product numbers between 2013 (n = 283) and 2020 (n = 543) almost doubled, led by granola and clusters. Whole grain cereals ≥ 8 g/serve made up 67% of products (↑114%). While there were positive changes in nutrient composition over time within the full data set, the most notable changes were in the nutrition composition of cereals marketed as the same product in both years (n = 134); with decreases in mean carbohydrate (2%), sugar (10%) and sodium (16%) (p < 0.000), while protein and total fat increased significantly (p = 0.036; p = 0.021). Claims regarding Dietary Fibre and Whole Grain doubled since 2013. Analysis of sub-categories of breakfast cereal assisted in identifying some changes over time, but products common to both timeframes provided a clearer analysis of change within the breakfast category, following introduction of HSR. Whole grain products were lower in the two target nutrients, sodium and sugars, and well-chosen products represent a better choice within this category.

scholarly journals Investigating the Broken-Heart Effect: a Model for Short-Term Dependence between the Remaining Lifetimes of Joint Lives

2012 ◽  
Vol 7 (2) ◽  
pp. 236-257 ◽  
Author(s):  
Jaap Spreeuw ◽  
Iqbal Owadally
Keyword(s):  
Mortality Rates ◽  
Short Term ◽  
Numerical Work ◽  
Data Set ◽  
Multiple State ◽  
Multiple State Model ◽  
Insurance Data ◽  
Over Time ◽  
Assurance Contracts

AbstractWe analyze the mortality of couples by fitting a multiple state model to a large insurance data set. We find evidence that mortality rates increase after the death of a partner and, in addition, that this phenomenon diminishes over time. This is popularly known as a “broken-heart” effect and we find that it affects widowers more than widows. Remaining lifetimes of joint lives therefore exhibit short-term dependence. We carry out numerical work involving the pricing and valuation of typical contingent assurance contracts and of a joint life and survivor annuity. If insurers ignore dependence, or mis-specify it as long-term dependence, then significant mis-pricing and inappropriate provisioning can result. Detailed numerical results are presented.

Sign in / Sign up

Export Citation Format

Share Document