Best-Ball Events in Golf: An Application of the Multinomial Distribution

Probability models are developed for two-man teams in best-ball events in golf. The multinomial distribution is used to develop the theoretical probabilities for possible team scores. Here each individual player is assigned a set of probabilities for making a certain score on each hole of an 18-hole round of golf. The two-man team results are compared with the 1980-1982 Bing Crosby Pro-Ams held in Monterey, California.     

Non-parametric procedure for knockout tournaments

In a seeded knockout tournament, where teams have some preassigned strength, do we have any assurances that the best team in fact has won? Is there some insight to be gained by considering which teams beat which other teams solely examining the seeds? We pose an answer to these questions by using the difference in the seeds of the two players as the basis for a test statistic. We offer several models for the underlying probability structure to examine the null distribution and power functions and determine these for small tournaments (less than five teams). One structure each for 8 teams and 16 teams is examined, and we conjecture an asymptotic normal distribution for the test statistic.     

A comparison of scientific and engineering criteria for Bayesian model selection

Given a set of possible models for variables X and a set of possible parameters for each model, the Bayesian estimate of the probability distribution for X given observed data is obtained by averaging over the possible models and their parameters. An often-used approximation for this estimate is obtained by selecting a single model and averaging over its parameters. The approximation is useful because it is computationally efficient, and because it provides a model that facilitates understanding of the domain. A common criterion for model selection is the posterior probability of the model. Another criterion for model selection, proposed by San Martini and Spezzafari (1984), is the predictive performance of a model for the next observation to be seen. From the standpoint of domain understanding, both criteria are useful, because one identifies the model that is most likely, whereas the other identifies the model that is the best predictor of the next observation. To highlight the difference, we refer to the posterior-probability and alternative criteria as the scientific criterion (SC) and engineering criterion (EC), respectively. When we are interested in predicting the next observation, the model-averaged estimate is at least as good as that produced by EC, which itself is at least as good as the estimate produced by SC. We show experimentally that, for Bayesian-network models containing discrete variables only, the predictive performance of the model average can be significantly better than those of single models selected by either criterion, and that differences between models selected by the two criterion can be substantial.     

Paired comparison models applied to the design of the Major League baseball play-offs

This paper presents an analysis of the eff ect of various baseball play-off configurations on the probability of advancing to the World Series. Play-off games are assumed to be independent. Several paired comparisons models are considered for modeling the probability of a home team winning a single game as a function of the winning percentages of the contestants over the course of the season. The uniform and logistic regression models are both adequate, whereas the Bradley-Terry model (modified for within-pair order eff ects, i.e. the home field advantage) is not. The single-game probabilities are then used to compute the probability of winning the play-off s under various structures. The extra round of play-off s, instituted in 1994, significantly lowers the probability of the team with the best record advancing to the World Series, whereas home field advantage and the diff erent possible play-offdraws have a minimal eff ect.     

The Role of the Statistician: Scientist or Shoe Clerk

A carnival game is described in which a player rolls eight dice or eight marbles into holes. To win, the player must obtain a favorable outcome on a number of plays. The probability distribution for the outcome of a single play is discussed. The distribution of a winning combination of outcomes is discussed and approximated. These probabilities show that the game is essentially unbeatable.     

A modified subset selection formulation with special reference to one-way and two-way layout experiments

The usual formulation of subset selection due to Gupta (1956) requires a minimum guaranteed probability of a correct selection. The modified formulation of the present paper includes an additional requirement that the expected number of the nonbest populations be bounded above by a specified constant when the best and the next best populations are ‘sufficiently’ apart. A class of procedures is defined and the determination of the minimum sample size required is discussed. The specific problems discussed for normal populations include selection in terms of means and variances, and selection in terms of treatment effects in a two-way layout.     

Supersaturated design including an orthogonal base

Recently, many supersaturated designs have been proposed. A supersaturated design is a fractional factorial design in which the number of factors is greater than the number of experimental runs. The main thrust of the previous studies has been to generate more columns while avoiding large values of squared inner products among all design columns. These designs would be appropriate if the probability for each factor being active is uniformly distributed. When factors can be partitioned into two groups, namely, with high and low probabilities of each factor being active, it is desirable to maintain orthogonality among columns to be assigned to the factors in the high-probability group. We discuss a supersaturated design including an orthogonal base which is suitable for this common situation. Mathematical results on the existence of the supersaturated designs are shown, and the construction of supersaturated designs is presented. We next discuss some properties of the proposed supersaturated designs based on the squared inner products.     

AN ANALYSIS OF CLASSICAL AND TIE-BREAKER TENNIS

Expressions for the probability that a player wins a game, set or match of classical or tie-breaker tennis are obtained. Also, expressions for the distribution, the mean and the variance of the number of points in a game, set or match of classical or tie-breaker tennis are obtained. These results explain the long matches often previously observed in classical tennis between two players each with very effective serves on fast grass-court surfaces. The methodology used to derive these expressions could be used to obtain corresponding expressions for other "nested-type" scoring systems. For example, tennis is 3-nested (game, set, match) whereas squash is only 2-nested (game, match).     

On lower confidence bounds for pcs in truncated location parameter models

We are concerned with deriving lower confidence bounds for the probability of a correct selection in truncated location-parameter models. Two cases are considered according to whether the scale parameter is known or unknown. For each case, a lower confidence bound for the difference between the best and the second best is obtained. These lower confidence bounds are used to construct lower confidence bounds for the probability of a correct selection. The results are then applied to the problem of seleting the best exponential populationhaving the largest truncated location-parameter. Useful tables are provided for implementing the proposed methods.     

Simultaneous prediction intervals for multiple comparisons with a standard

k normal populations having common variance are used to construct two-sided and one-sided simultaneous prediction intervals for the differences between the future means of independent random sample from each of these populations compared to a standard. These prediction intervals are particularly useful if one has sampled the performance of several products and wishes to simultaneously predict the differences between future sample mean performance of these products and a standard with a predetermined joint probability. Methods on sample size determination are also given. The procedures are illustrated with a numerical example. Received: February 25, 2000; revised version: February 6, 2001     

Action rate models for predicting actions in soccer

AStA Advances in Statistical Analysis - We present a data-driven approach to predict the next action in soccer. We focus on passing actions of the ball possessing player and aim to forecast the...     

Cubic lattices

In some experiments, the problem is to compare many unstructured treatments in small blocks, the classical example being the study of new plant varieties on variable land. A common method is to use lattice designs, i.e. block designs based upon rows and columns of a square format, with ficrther replicates being formed, required, from orthogonal squares applied to the format. It has been known for some time that cubes can be used instead; this paper sets out to explore the possibilities. There are two cases. In one case, the blocks are formed from the planes of the cube and, in the other case, from its lines. The cubic lattice basically has three replicates-one from each dimension-but, two or four replicates are required, a design can be found by omitting or duplicating one of the dimensions. Where standard treatments need to be introduced, -a useful device is to reinforce, i.e. supplement each block with additional plots of standards, with each block of a replicate being supplemented in the same way. These possibilities are examined. It emerges that cubic lattices with two or three replicates usefully extend the range of available designs, but that those with four replicates are disappointing. However, there is the alternative of using designs based upon Latin cubes. This matter is not taken far but it is shown that, where the Latin cube exists, it gives a better design. A quick way of calculating an approximate analysis of variance is given, which is applicable in a wide range of cases.     

Program of the Section on Economic and Business Statistics

Suppose that in the situation of a paired t test natural pairing, such as the use of twins, is not possible. Reduction in variability is then often achieved artificially, for example by pairing animals of similar birth weight. This article points out that, unless such pairing is ineffective, the usual assumptions underlying the paired t test are violated. Nevertheless, simulation indicates that, with randomization in the allocation of treatments, the standard procedure gives good results. Our bivariate normal model provides the factor by which the length of the confidence interval for the mean treatment difference is reduced as a result of the pairing. Another form of pairing sometimes used is shown to be incorrect. Nonparametric analogs are also briefly considered.     

Control-variate methods for comparison with a standard

Comparison with a standard is a general multiple comparison problem, where each system is required to be compared with a single system, referred to as a ‘standard’, as well as with other alternative systems. Screening procedures specially designed to be used for comparison with a standard have been proposed to find a subset that includes all the systems better than the standard in terms of the expected performance. Selection procedures are derived to determine the best system among a number of systems that are better than the standard, or to select the standard when it is equal to or better than the other alternatives. We develop new procedures for screening and selection through the use of two variance reduction techniques, common random numbers and control variates, which are particularly useful in the context of simulation experiments. Empirical results and a realistic example are also provided to compare our procedures with the existing ones.     

Brain of Britain: brain or chance?

The paper describes a simulation of the popular BBC Radio general knowledge quiz 'Brain of Britain'. The relative levels of knowledge of the 64 contestants in any annual competition are selected from an appropriate distribution and ranked accordingly. The relative difficulties of the questions are selected from three distributions in accordance with the deliberate intentions of the question-setter. By Monte Carlo simulation of the complete annual quiz the relationship between a contestant's rank and his fortune has been established.

The best contestant is found to have only a 20% probability of becoming 'Brain of Britain'. Nevertheless it is demonstrated that there are no practicable modifications to the quiz format which would result in any worthwhile improvement.     

A Bayesian method for safety signal detection in ongoing blinded randomised controlled trials

Sponsors have a responsibility to minimise risk to participants in clinical studies through safety monitoring. The FDA Final Rule for IND Safety Reporting requires routine aggregate safety evaluation, including in ongoing blinded studies. We are interested in estimating the probability that the true adverse event rate in the experimental arm exceeds that in the control arm. We developed a Bayesian approach that specifies an informative meta-analytic predictive prior on the event probability in the control arm and an uninformative prior on that in the experimental arm. We combined these priors with a mixture likelihood that considers each patient in the ongoing blinded study may belong to the experimental or control arm. This allowed us to estimate the quantity of interest without unblinding. We evaluated our method by simulation, pairing scenarios that differed only in whether a safety signal was present or missing, and quantifying the ability of our model to discriminate using signal detection theory. Our approach shows benefit. It detects safety signals more reliably with greater sample sizes and for common rather than rare events. Performance does not deteriorate markedly when historical studies exhibit heterogeneous hazards or non-constant hazards. Our method will allow us to monitor safety signals in ongoing blinded studies with the goal of earlier identification and risk mitigation. Our method could be adapted to use informative priors on both arms or predictive covariates where pertinent data exist. We stress that ongoing safety monitoring should involve a multi-disciplinary team where statistical methods are paired with medical judgement.     

Multinomial Selection for Comparison with a Standard

The multinomial selection problem is considered under the formulation of comparison with a standard, where each system is required to be compared to a single system, referred to as a “standard,” as well as to other alternative systems. The goal is to identify systems that are better than the standard, or to retain the standard when it is equal to or better than the other alternatives in terms of the probability to generate the largest or smallest performance measure. We derive new multinomial selection procedures for comparison with a standard to be applied in different scenarios, including exact small-sample procedure and approximate large-sample procedure. Empirical results and the proof are presented to demonstrate the statistical validity of our procedures. The tables of the procedure parameters and the corresponding exact probability of correct selection are also provided.     

Ranking paired contestants

The ranking of paired contestants (players) after a series of contests is difficult when every player does not play every other player. In the 1975 JASA Mark Thompson presented a maximum likelihood solution based on the assumption that the probability of any one player defeating any other is a function only of the difference in their ranks. Here the linear approximation to that likelihood is shown to lead to a nonparametric measure of the efficacy of the ranking, called the net difference in ranks (NDR) , which is the sum of the differences in ranks of the paired players in the observed contests that agree with the ranking minus the sum of the differences in ranks in the observed contests that disagree with the ranking (upsets) . The subject is part of a large literature that has been consolidated by H.A. David in The Method of Paired Comparisons (1963, 1988). The method was introduced by the psychophysicist Fechner in 1860 and has been widely applied to sensory testing,     

The Two-Dimensional Asymmetric Gambler's Ruin Problem with n Players and Simulations

In this article, the asymmetric n-player gambler's ruin problem is considered, when the players use equal initial fortunes of d dollars and d euros, 1 ≤ d ≤ n + 1. In each round an unfair coin is tossed to decide the currency. The expected ruin time and the individual ruin probabilities are computed. It is proved that the ruin time and which player is ruined are independent events. Finally, some special games are simulated. The simulation results verify the validity of the proposed formulas. As an innovation, the present study makes a combination of the n-player and multi dimensional games which can be viewed as a starting point for future studies.     

Some Geometric Methods for Constructing Decision Criteria Based On Two-Dimensional Parameters

This paper reviews two types of geometric methods proposed in recent years for defining statistical decision rules based on 2-dimensional parameters that characterize treatment effect in a medical setting. A common example is that of making decisions, such as comparing treatments or selecting a best dose, based on both the probability of efficacy and the probability toxicity. In most applications, the 2-dimensional parameter is defined in terms of a model parameter of higher dimension including effects of treatment and possibly covariates. Each method uses a geometric construct in the 2-dimensional parameter space based on a set of elicited parameter pairs as a basis for defining decision rules. The first construct is a family of contours that partitions the parameter space, with the contours constructed so that all parameter pairs on a given contour are equally desirable. The partition is used to define statistical decision rules that discriminate between parameter pairs in term of their desirabilities. The second construct is a convex 2-dimensional set of desirable parameter pairs, with decisions based on posterior probabilities of this set for given combinations of treatments and covariates under a Bayesian formulation. A general framework for all of these methods is provided, and each method is illustrated by one or more applications.