.........................................................................

........................................................................
........................................................................
........................................................................
........................................................................
Maiden Over: Analyzing two commonly held
beliefs in one-day international cricket matches

........................................................................
Projesh Banerjea

............................................................................
........................................................................
........................................................................
........................................................................
........................................................................
........................................................................
........................................................................
Abstract
With its known history being traced back to the 16th century, cricket
is the second most popular game in the world and over 150 nations are
aliated with the International Cricket Council (ICC). Though interna-
tional matches have been played since 1844, academics have only recently
taken an interest in studying dierent facets of the game and this paper
analyzes the article  Winning the Coin Toss and the Home Team Advan-
tage in One-Day International Cricket Matches by Basil de Silva and Tim
Swartz. The authors study widely held perceptions regarding the coin-
toss and home team advantage in one-day international cricket matches
using a dataset of over 400 games played in the 1990s. Mathematical
models and processes such as Bernoulli variables, Logit regressions, and
the Bayesian paradigm are applied to study whether winning the coin toss
or playing at home provides a team with an advantage. The appendix to
this paper includes data tables, the results of the authors' analysis, and a
brief explanation of cricketing rules and terms.

..............................................................................................................
........................................................................
........................................................................
........................................................................
........................................................................
........................................................................
........................................................................
........................................................................
........................................................................
........................................................................

1
 Introduction
Given the world-wide appeal of the sport and histories of the competing
countries, cricket tournaments are often wrought with tension and a lot of money
is bet on the outcomes of certain games. When Bangladesh beat Pakistan in the
1999 World Cup, the next day was declared a national holiday for Bangladeshis
to celebrate the feat. After India beat Pakistan in the 1996 World Cup semi-
nals, numerous TV sets were smashed in Pakistan while reworks were set o
in multiple Indian cities. More recently, when Pakistan crashed out early from
the 2007 World Cup, their coach, Bob Woolmer, was found strangled to death
in his hotel room. Though the game is followed closely by fans, however, it
has not received much academic interest from a statistical and mathematical
perspective. Consequently, there are a number of commonly held beliefs that
aect teams' strategic decisions but the relevance and accuracy of these beliefs
are yet to be tested.
....
This paper examines two such beliefs using various statistical models and
techniques to do so. The rst is the belief that winning the coin-toss in the one-
day game leads to an advantage that may aect the outcome of the match. The
second is the belief that the team playing at home has an advantage that can
impact the outcome of a match. The authors use a dataset of 427 matches played
during the 1990s among the nine nations of the ICC and tables from their paper
are included in the appendix along with a brief explanation of certain cricket
rules and terms.
....
Hypothesis 1: Winning the coin toss has an impact on the overall
result of the game
Cricket matches start with a coin toss and the team that wins is given the
option of batting or elding rst. A number of people believe that the team that
wins the coin toss should always bat rst as they set the score to be chased and
create a psychological hurdle for the team batting second. Others believe that
the team batting second has the advantage as they go in to bat knowing the
runs they need to score and can adjust their strategy accordingly. Yet another
group believes that the decision should depend on the playing conditions and
be sensitized to key players in both teams to help maximize or minimize their
impact accordingly. It can also be seen from Table 1 in the appendix that
dierent teams follow dramatically dierent strategies. For example, Australia
chooses to bat rst in 87% of the games they win the toss in and have an overall
record of winning 63% of their matches. West Indies, however, chooses to bat
rst only 36% of the time but has an overall record of winning 53% of their
matches. Thus, it is not clear that a single strategy dominates.
....
Ideally, the authors would like to compare proposed strategies for dierent
teams but in most cases that would depend on variables such as team morale and

2
coaching whims that cannot be singled out from the data. Using the example of
New Zealand specically, the authors say that the team chooses to bat rst in
roughly half their matches but these could be the ones the believe they will win
anyway. Consequently, it would be erroneous to say that batting rst is what
led to New Zealand winning those games. The authors are therefore resigned
to studying the observed strategies of the nine ICC teams and they do so by
analyzing a team's performance when it wins the coin toss as opposed to when
it loses the toss. They do so using the four approaches dened separately for
convenience as follows  a) Bernoulli trial and Binomial test, b) Logit model c)
Bayesian approach d) Empirical Bayesian model. However, before the models
are examined, it is necessary to understand how the data is coded and arranged.
....
Data coding
The authors use the notation (xi , yi ), i = 1,2, . . . .n where n is the number
of games played by the team of interest, to arrange the data. The variable xi =
1(0)if the team wins (loses) the coin toss in the ith game and yi = 1(0) if the team
wins (loses) the game. Thus, two Bernoulli variables pi and qi are dened where
pi = 1 if yi = 1|xi = 1 or yi |xi = 1 is Bernoulli (pi ). Similarly, qi = 1 if yi =
1|xi = 0 or yi |xi = 0 is Bernoulli (qi ). ............................................................
............................................................
xi yi

0 0 1 − qi
0 1 qi
1 0 1 − pi
1 1 pi
............................................................
It is important to note that the variables pi and qi refer to events and not
probabilities in this data coding table. It is also important to note that the
probability P (xi = 0) = P (xi = 1) =
Pn0.5 for all i. Thus, the authors are
P n
interested in comparing the strategy i=1 pi versus i=1 qi where n is the
number of games played. The rst Bernoulli variable is the number of wins
having won the coin toss and the second is the number of wins having lost the
coin toss.
............................................................
....................................................................
....................................................................
Method 1: Bernoulli trial and Binomial Test
A Bernoulli trial is an experiment with a randomized outcome that can be
one of two values. If the probability of one outcome occurring is p then that of
the other is 1-p and these probabilities are held constant across all the events or
iterations. A commonly used example is that of a coin toss where the outcome
can be either heads or tails. Correspondingly, a Bernoulli process, a sequence
of Bernoulli trials, is a discrete time process involving independent random
variables. Also, the number of successes in the rst n trials has a Binomial
distribution and this distribution can be used to conduct a Binomial test.
............................................................

3
 The rst method requires the simple assumption that pi = p and qi = q
for all i=1,2,. . . n. This assumption is in keeping with the requirements of a
Bernoulli trial as the probabilities are constant and the events random and
independent. Unfortunately, this method assumes that all opponents are of
equal strength and that playing conditions are constant over time. It does not
allow any team to change its strategy based on evidence and experience due
to the  memoryless property of the independent trials. As a starting point
however, it allows the authors to test the null hypothesis Ho : p ≤ q versus
the alternative H1 : p > q as the rst Binomial variable is the number of wins
having won the coin toss while the second is the number of wins having lost the
coin toss. The p-values for the each of the ICC nations are given in column 1 of
Table2 in the appendix and it is clear to see that these p values are very high
(over .10) and consequently no clear relationship is established.
....................................................................
....................................................................
Method 2: Logit model
The logit model is used in statistics primarily when the dependent variable
is binary and the outcome is the probability of an event. While a simple lin-
ear regression works in most cases, it has no method of limiting the outcomes
between 0 and 1and values below 0 or greater than 1 have no meaning as proba-
bilities. The logit model, however, is bounded between these two outcomes and
is consequently a better method for this purpose. Unfortunately, the coecients
of a logit regression cannot be interpreted directly as probability a probability
and must be converted using the following formula
.......................................................................
P = 1/1 + e−(a+bx)
.......................................................................
The proof for this interpretation involves an understanding of the concept
of odds as well as probability. The odds of an event can be interpreted as the
likelihood of it occurring as opposed to not or the ratio of the event occurring
to it not doing so. While probabilties are commonly calculated as the instances
in which an event occurs over the total sample space, odds are calculated as
the instances in which the event occurs over the instances when it doesn't.
Correspondingly, while the probability of picking Sunday out of the days in the
week is 1/7, the odds are 6 to 1 against as there are six instances in which
Sunday is not picked.
.......................................................................
Thus, the odds of an event are related to its probability by
.......................................................................
odds = P/1 − P
.......................................................................
However, this relationship isn't symmetric because if the probability of an
event occurring is .90, its odds are 9 to 1 for but the reverse or 1 to 9 against
gives us a probability of .11. The asymmetry can be controlled by using the
natural logarith function as the natural log of an event occurring is exactly
opposite to the natural log of it not happening. Thus,

4
 .......................................................................
log(odds) = logit(P ) = ln(P/1 − P )
...................................................
Let ln(P/1 − P ) = a + bx
...................................................
=> e(a+bx) = P/1 − P
...................................................
=> P = ea+bx /1 + ea+bx
...................................................
=> P = 1/1 + e−(a+bx)
............................................
The logit model used by the authors is as follows:
................................................................
logit(pi ) = aP
+ bxi + ti where ti = tk if the ith game is against opponentk, k
n
=1, 2. . . 8. and i=1 ti = 0.
.......................................................................
Here ti is a strength dierentiator and this model is more accurate than the
rst as it is sensitive to teams with dierent strengths. Unfortunately, however,
this model assumes that teams maintain their strengths consistently as it does
not allow forti , the strength dierentiating term, to change over time. This
assumption is clearly inaccurate as teams change constantly with new players
being added and old players being dropped from the roster.
.......................................................................
When we test the hypothesis using this model, the p-values are again found
to be too high and so we fail to reject the null hypothesis again. The team
that comes closest to obtaining a statistically signicant output is India with
a p-value of .13 and it is interesting to note that India chooses to bat rst in
approximately half its matches (Table 1).
....................................................................
....................................................................
Method 3: Bayesian approach
The Bayesian method to calculating probabilities uses evidence to update the
probability that a hypothesis is true. As opposed to the classical or frequentist
approach which is intrinsically linked to the law of large numbers and depends
on outcomes and sample spaces, Bayesians use evidence to update the degree of
belief in a hypothesis. The formula used to represent this calculation is:
........................................
P (H|E) = P (E|H).P (H)/P (E)
........................................
Here the probability of a hypothesis standing after evidence, or the posterior
probability, is calculated by examining the probability of nding the evidence
given the hypothesis, multiplying it with the prior probability or degree of faith
in the hypothesis, and nally dividing these two terms by the probability of
nding evidence. Starting with prior beliefs or prior probabilities, this method
calculates the posterior probability of the hypothesis after the evidence has been

5
observed. If the probability of obtaining evidence to support a belief is low, the
prior probability will decrease the degree of belief and vice versa.
........................................
The specic model used by the authors is given below where this function
eectively sorts the dierent outputs to update the assumptions over time. With
independent priors pi and qi , the Bayesian methods gives marginal posterior
distributions which are then used to calculate the posterior probability of the
null hypothesis. The formula used by the authors to calculate the probabilities
is:
........................................
n y (1−xi )
pyi i xi (1−pi )(1−yi )xi q i i (1−qi )(1−yi )(1−xi ) for i = 1, 2, ....n
Q
[y, x|p, q]α i=1

Where the independent priors pi and qi have the distributions Beta(ai , bi )

and Beta(ci , di ) respectively. The Bayesian paradigm gives the independent
marginal posterior distripbutions [pi | x, y] with Beta(ai + yi xi , bi + (1 − yi )xi )
and [qi | x, y] with Beta(ci + yi (1 − xi ), di + (1 − yi )(1 − xi )). It is interesting to
note that of the four terms in the equation above, only one is tabulated every
time as the other three go to one given the outcome of the toss and game. If a
team wins the toss and the game (i.e.yi =1 and xi = 1) then only pi is registered
as the other three powers are 0 resulting in the terms being 1. Thus, the degree
of belief in a certain hypothesis is updated after every game based on the results
of that game and this new evidence now serves as a prior probability for the
next iteration.
.......................................................................
.......................................................................
Of the three methods used so far, this is the most accurate as it updates
assumptions based on evidence (the outcomes of the matches played) and con-
sequently controls for teams with dierent strengths that change over time.
Unfortunately, however, the problem with this model is the need to specify
prior information and the authors use standard default priors that are used to
represent ignorance which takes away from the ecacy of the model. The p-
values, shown in column 3 of Table 2, are too large again so this method too
fails to establish a statistically signicant relationship between winning the coin
toss and the outcome of the game.
....................................................................
....................................................................
Method 4: Empirical Bayesian model
Acting on a suggestion by a referee, the authors used a fourth method where
the prior parameters were determined by the data as opposed to the standard
default priors used in method three. The authors use the information from
column two of Table 1, and adjust the mean prior probability of winning the ith
match to the observed overall winning proportions Wo . The second parameter
in the Beta distribution is set to prior variances based o the belief that three
standard deviations would adjust this parameter to the nearest endpoint (0 or
1). Thus, Wo is set to Wo = ai /(ai + bi ) for i = 1, 2, ...n and Wo = ci /(ci +
di ), i = 1, 2, ...n. The Bayesian model, as shown above, is again used to calculate

6
posterior probabilities after the priors have been adjusted. The p-values of this
method are listed in column 4 of Table 2 and it is clear that they are too high to
be statistically signicant. Thus this method too fails to highlight a relationship
between the coin toss and the overall outcome of the game.
.......................................................................
Conclusion for the rst hypothesis
Under all four methods of analysis, no statistically signicant relationship
was found between the coin toss and the result of the game and so we fail to reject
the null hypothesis. Though these methods do not exhaust every possible way
of testing this hypothesis, the authors state that it would be dicult for other
researchers to come up with more accurate models and consequently conclude
that winning the coin toss has no impact on the overall outcome of a match.
....................................................................
....................................................................
Hypothesis 2: The Home Team Advantage
In most physically demanding sports or spectator sports played under pres-
sure, the existence of a home team advantage is generally accepted. It is likely
that the crowd support boosts adrenalin levels which in turn boost performance
and the authors note a home team winning percentage of 57.4% in the 1996/7
season of the NBA. Though cricket isn't as physically demanding as soccer or
basketball, it is possible this advantage exists due to familiarity with weather
conditions and biased umpiring. A simple comparison of the data in columns
2 and four of Table 2 reects that every ICC team wins a higher proportion of
games at home than their overall winning record.
.......................................................................
Data coding
To study this relationship, the authors modify their notation to let pijk
represents the probability that team i defeats team j at site k wherei,j,k =
1,2,. . . .9 andk = 0 denotes a non-ICC site. Here, ti is a dierential strength
term similar to the one used in method two above and sums to 0. The term
gijk = g if team i is the home team, 0 if the game is played on a neutral site,
and −g if team j is the home team as the assumption is that the probability of
team i beating team j decreases if j is the home team.
.......................................................................
.......................................................................
Logit model
A logit model is used again to calculate the probability and the model specif-
ically is logit(pijk ) = ti − tj + gijk
.....................................................
Where ti − tj represents the log-odds of the advantage or disadvantage that
team i has over team j. For two equally strong teams, we note that this sums
to 0 and if the game is played on a neutral site then logit(pijk ) = 0 implying
pijk = 0.5.
.......................................................................

7
 The home team advantage is quantied on the log-odds scale as the non-
linearity of this method reects that weaker teams have a stronger relative
improvement when playing at home as opposed to teams that were likely to
win anyway. The answer is quantied to be .53 which can be interpreted to
mean that a team with a winning percentage = 50% would increase its winning
percentage to 63% when playing at home. This spread will be larger for weaker
teams and smaller for stronger ones. Thus, the home team advantage is found to
be statistically signicant and to provide a substantial edge to the team playing
at home.
.......................................................................
This model is then modied slightly to account for countries with large em-
igrant populations as they are likely to receive the advantages of crowd support
even when they are not playing at home. Similarly, there are countries that
are less popular as they do not play as frequently and so they lose out on the
support when playing on neutral grounds. Thus, countries such as the West
Indies and Pakistan, which have large emigrant populations and frequently play
on tour, are given a positive home team advantage g for neutral grounds while
countries like New Zealand and Sri Lanka are given g for neutral grounds. The
result is found to be almost the same.
.......................................................................
Conclusion for the second hypothesis
Based on this research, it can be seen that countries playing at home have a
signicant advantage over their opponents and this advantage is increased or de-
creased given the relative strength of the team. This result might have an impli-
cation for people betting on the outcome of a game because if two teams are per-
ceived to be comparable in terms of their relative strength, then the one playing
at home has a higher probability of winning. .......................................................................
.......................................................................
.......................................................................
Conclusion
From the results of this paper we deduce that no clear relationship is es-
tablished between winning the coin toss and the outcome of the match but the
team playing at home does indeed have an advantage. The methods used in
this paper can also be applied to study other factors of the game and perhaps
even formulate strategies for dierent teams. Factors such as the use of a white
ball versus red, playing only during the day as opposed to day and night, and
the optimal combinations of batsmen and bowlers can all be studied using these
techniques and the results could have a signicant impact on team selection and
planning. The results could also impact those betting on the outcomes of these
games.
.......................................................................
.......................................................................
.......................................................................
.......................................................................

8
 ........................................
.......................................................................
APPENDIX 1: Data Tables and Regression Results
.......................................................................

Figure 1:

Figure 2:

.......................................................................

9
 ....................................................................
....................................................................
.....................................................................
APPENDIX 2: An Explanation of Cricket Basics
.......................................................................
Cricket is a team sport for two teams of eleven players each. A formal game
of cricket can last anything from an afternoon to several days. Although the
game play and rules are very dierent, the basic concept of cricket is similar to
that of baseball. Teams bat in successive innings and attempt to score runs,
while the opposing team elds and attempts to bring an end to the batting
team's innings. After each team has batted an equal number of innings (either
one or two, depending on conditions chosen before the game), the team with
the most runs wins. (Note: In cricket-speak, the word  innings is used for both
the plural and the singular.  inning is a term used only in baseball.)
.......................................................................
Equipment
Cricket Ball: Hard, cork and string ball, covered with leather. A bit like a
baseball (in size and hardness), but the leather covering is thicker and joined in
two hemispheres, not in a tennis ball pattern. Traditionally the ball is dyed red,
with the stitching left white. Nowadays white balls are also used, for visibility
in games played at night under articial lighting.
.......................................................................
Cricket Bat: Blade made of willow, at on one side, humped on the other
for strength, attached to a sturdy cane handle.
.......................................................................
Wickets: There are two wickets - wooden structures made up of a set of
three stumps topped by a pair of bails. ...........................................
............................
Stumps: Three wooden posts that are just close enough together that a
cricket ball cannot pass between them. .......................................................................
............................
Bails: Two wooden crosspieces which sit in grooves atop the adjacent pairs of
stumps. A complete wicket looks like this: .......................................................................
............................
Protective Gear: Pads, gloves, helmet, etc for batsmen to wear to prevent in-
jury when struck by the ball. .......................................................................
............................
Shoes: Leather, usually with spiked soles for grip on the grass. ..............................................................
............................
Clothing: Long pants, shirt (long or short sleeved depending on the weather),
possibly a sleeveless or long-sleeved woollen pullover in cold weather. For games
played with a red ball, the clothing must be white or cream. With a white ball,
players usually wear uniforms in solid team colours. Add a hat or cap to keep

10
the sun o. There are no regulations regarding identifying marks or numbers
on clothing.
.......................................................................
The Field
A cricket eld is a roughly elliptical eld of at grass, ranging in size from
about 90 to 150 metres (100-160 yards) across, bounded by an obvious fence
or other marker. There is no xed size or shape for the eld, although large
deviations from a low-eccentricity ellipse are discouraged. In the centre of the
eld, and usually aligned along the long axis of the ellipse, is the pitch , a
carefully prepared rectangle of closely mown and rolled grass over hard packed
earth. It is marked with white lines, called creases.
.......................................................................
The Play
The order in which the teams bat is determined by a coin toss. The captain
of the side winning the toss may elect to bat or eld rst. All eleven players of
the elding team go out to eld, two players of the batting team go out to bat.
The remainder of the batting team wait o the eld for their turn to bat. Each
batsman wears protective gear and carries a cricket bat. The game progresses
by the bowling of balls. The sequence of events which constitutes a ball follows:
The elding team disperses around the eld, to positions designed to stop runs
being scored or to get batsmen out. One elder is the bowler. He takes the ball
and stands some distance behind one of the wickets (i.e. away from the pitch).
Another elder is the wicket-keeper, who wears a pair of webbed gloves designed
for catching the ball and protective pads covering the shins. He squats behind
the opposite wicket. .......................................................................
.......................................................................
The rest of the elders have no special equipment - gloves to assist catching
the ball are not allowed to anyone but the wicket-keeper. One batsman stands
behind each popping crease, near a wicket. The batsman farthest from the
bowler is the striker, the other is the non-striker. The striker stands before his
wicket, on or near the popping crease, in the batting stance. For a right-handed
batsman, the feet are positioned like this: The batsman stands with his bat held
down in front of the wicket, ready to hit the ball, which will be bowled from the
other end of the pitch. The batsman usually rests the lower end of the bat on the
pitch and then taps the bat on the pitch a few times as warm-up backswings.
The non-striker simply stands behind the other popping crease, waiting to run if
necessary. The bowler takes a run-up from behind the non-striker's wicket. He
passes to one side of the wicket, and when he reaches the non-striker's popping
crease he bowls the ball towards the striker, usually bouncing the ball once on
the pitch before it reaches the striker. (The bowling action will be described
in detail later.) The striker may then attempt to hit the ball with his bat. If
he misses it, the wicket-keeper will catch it and the ball is completed. If he
hits it, the two batsmen may score runs (described later). When the runs are
completed, the ball is also considered completed. The ball is considered to be
in play from the moment the bowler begins his run-up. It remains in play until
any of several conditions occur (two common ones were just described), after

11
which it is called dead. The ball is also dead if it lodges in the striker's clothing
or equipment. .......................................................................
............................
Once the ball is dead, it is returned to the bowler for the next delivery
(another name for the bowling of a ball). Between deliveries, the batsmen may
leave their creases and confer with each other. When one bowler has completed
six balls, that constitutes an over. A dierent member of the elding team is
given the ball and bowls the next over - from the opposite end of the pitch. The
batsmen do not change ends, so the roles of striker and non-striker swap after
each over. Any member of the elding team may bowl, so long as no bowler
delivers two consecutive overs. Once a bowler begins an over, he must complete
it, unless injured or suspended during the over. Another possibility during a
ball is that a batsman may get out. There are ten dierent methods of being
out - these will be described in detail later. If a batsman gets out, the ball is
dead immediately, so it is impossible to get the other batsman out during the
same ball. The out batsman leaves the eld, and the next batsman in the team
comes in to bat. The not out batsman remains on the eld. The order in which
batsmen come in to bat in an innings is not xed. The batting order may be
changed by the team captain at any time, and the order does not have to be
the same in each innings. When ten batsmen are out, no new batsmen remain
to come in, and the innings is completed with one batsman remaining not out.
The roles of the teams then swap, and the team which elded rst gets to bat
through an innings. When both teams have completed the agreed number of
innings, the team which has scored the most runs wins.
.......................................................................
Scoring Runs
Whenever a batsman hits the ball during a delivery, he may score runs. A
run is scored by the batsmen running between the popping creases, crossing
over midway between them. When they both reach the opposite crease, one
run is scored, and they may return for another run immediately. The elding
side attempts to prevent runs being scored by threatening to run out one of the
batsmen. If the batsmen are attempting to take runs, and a elder gathers the
ball and hits a wicket with it, dislodging one or both bails, while no batsman
is behind that wicket's popping crease, then the nearest batsman is run out.
Specically, the batsman must have some part of his body or his bat (provided
he is holding it) grounded behind (not on) the crease. The batsmen carry their
bats as they run, and turning for another run is accomplished by touching the
ground beyond the crease with an outstretched bat. The batsmen do not have
to run at any time they think it is unsafe - it is common to hit the ball and elect
not to run. If the batsmen run one or three (or ve! rare, but possible), then
they have swapped ends and their striker/non-striker roles are reversed for the
next ball (unless the ball just completed is the end of an over). In addition to
scoring runs like this, if a batsman hits the ball so that it reaches the boundary
fence, he scores four runs, without needing to actually run them. If a batsman
hits the ball over the boundary on the full, he scores six runs. If a four or six is
scored, the ball is completed and the batsmen cannot be run out. If a spectator

12
encroaches on to the eld and touches the ball, it is considered to have reached
the boundary. If a elder gathers the ball, but then steps outside or touches the
boundary while still holding the ball, four runs are scored. If a elder catches
the ball on the full and, either during or immediately after the catch, steps
outside or touches the boundary, six runs are scored. The batsmen usually stop
taking runs when a elder is throwing the ball back towards the pitch area.
If no elder near the pitch gathers the ball and it continues into the outeld
again, the batsmen may take more runs. Such runs are called overthrows. If
the ball reaches the boundary on an overthrow, four runs are scored in addition
to the runs taken before the overthrow occurred. Runs scored by a batsman,
including all overthrows, are credited to him by the scorer. The number of runs
scored by each batsman is an important statistic. If, while running multiple
runs, a batsman does not touch the ground beyond the popping crease before
he returns for the next run, then the umpire at that end will signal one short,
and the number of runs scored is reduced by one.
.......................................................................
Getting Out
Here is a full list of the ten dierent ways of getting out. But rst, a few
necessary denitions: The wicket is said to be broken if one or both of the
bails have been dislodged and fallen to the ground. If the bails have fallen o
for any reason and the ball is still in play, then breaking the wicket must be
accomplished by pulling a stump completely out of the ground. If the wicket
needs to be broken like this with the ball, the uprooting of the stump must be
done with the ball in contact with the stump. The eld is notionally split into
two halves, along a line down the centre of the pitch. The half of the eld in
front of the striker is called the o side, the half behind is called the leg side,
or sometimes the on side. Thus, standing at the bowler's wicket and looking
towards a right-handed striker's wicket, the o side is to the left and the leg
side to the right (and vice-versa for a left-handed striker). The stumps of the
striker's wicket are called o stump, middle stump, and leg stump, depending on
which side they are on. When a batsman gets out, no matter by what method,
his wicket is said to have fallen, and the elding team are said to have taken a
wicket.
.......................................................................
Now, the ways of getting out:
Caught: If a elder catches the ball on the full after the batsman has hit
it with his bat. However, if the elder catches the ball, but either during the
catch or immediately afterwards touches or steps over the boundary, then the
batsman scores six runs and is not out.
.......................................................................
Bowled: If the batsman misses the ball and it hits and breaks the wicket
directly from the bowler's delivery. The batsman is out whether or not he is
behind his popping crease. He is also out bowled if the ball breaks the wicket
after deecting from his bat or body. The batsman is not out if the wicket does
not break.
.......................................................................

13
 Leg Before Wicket: If the batsman misses the ball with his bat, but inter-
cepts it with part of his body when it would otherwise have hit the wicket, and
provided several other conditions (described below) are satised. An umpire
must adjudicate such a decision, and will only do so if the elding team ap-
peal the decision. This is a question asked of the umpire, usually of the form
How's that? (or Howzat?), and usually quite enthusiastic and loud. If the
ball bounces outside an imaginary line drawn straight down the pitch from the
outside edge of leg stump, then the batsman cannot be out LBW, no matter
whether or not the ball would have hit the stumps. If the batsman attempts
to play a shot at the ball with his bat (and misses) he may only be given out
LBW if the ball strikes the batsman between imaginary lines drawn down the
pitch from the outside edges of leg and o stumps (ie. directly in line with the
wicket). If the batsman does not attempt to play the ball with his bat, then he
may be given out LBW without satisfying this condition, as long as the umpire
is convinced the ball would have hit the wicket. If the ball has hit the bat before
the hitting the batsman, then he cannot be given out LBW.
.......................................................................
Stumped: If a batsman misses the ball and in attempting to play it steps
outside his crease, he is out stumped if the wicket-keeper gathers the ball and
breaks the wicket with it before the batsman can ground part of his body or his
bat behind his crease.
.......................................................................
Run Out: If a batsman is attempting to take a run, or to return to his crease
after an aborted run, and a elder breaks that batsman's wicket with the ball
while he is out of the crease. The elder may either break the wicket with a hand
which holds the ball, or with the ball directly. It is possible for the non-striker
to be run out if the striker hits the ball straight down the pitch towards the
non-striker's wicket, and the bowler deects the ball on to the wicket while the
non-striker is out of his crease. If the ball is hit directly on to the non-striker's
wicket, without being touched by a elder, then the non-striker is not out. If the
non-striker leaves his crease (in preparation to run) while the bowler is running
up, the bowler may run him out without bowling the ball. Batsmen cannot be
run out while the ball is dead - so they may confer in the middle of the pitch
between deliveries if they desire.
.......................................................................
Hit Wicket: If, in attempting to hit a ball or taking o for a rst run, the
batsman touches and breaks the wicket. This includes with the bat or dislodged
pieces of the batsman's equipment - even a helmet or spectacles!
.......................................................................
Handle The Ball: If a batsman touches the ball with a hand not currently
holding the bat, without the permission of the elding side. This does not
include being hit on the hand by a delivery, or any other non-deliberate action.
.......................................................................
Obstructing The Field: If a batsman deliberately interferes with the eorts
of elders to gather the ball or eect a run out. This does not include running
a path between the elder and the wicket so that the elder cannot throw the

14
stumps down with the ball, which is quite legal, but does include any deliberate
attempt to swat the ball away.
.......................................................................
Hit The Ball Twice: If a batsman hits a delivery with his bat and then
deliberately hits the ball again for any reason other than to defend his wicket
from being broken by the ball. If the ball is bouncing or rolling around near the
stumps, the batsman is entitled to knock it away so as to avoid being bowled,
but not to score runs.
.......................................................................
Timed Out: If a new batsman takes longer than two minutes, from the time
the previous wicket falls, to appear on the eld.
.......................................................................
These methods of getting out are listed in approximate order of how com-
monly they occur. The rst ve are reasonably common, the last ve quite rare.
The last three methods are almost never invoked. If a batsman is out caught,
bowled, LBW, stumped, or hit wicket, then the bowler is credited with taking
the wicket. No single person is credited with taking a wicket if it falls by any
other method.
.......................................................................
Ocials
The game is adjudicated by two umpires, who make all decisions on the
eld and whose word is absolutely nal. One umpire stands behind the non-
striker's wicket, ready to make judgements on LBWs and other events requiring
a decision. The other umpire stands in line with the striker's popping crease,
about 20 metres (20 yards) to one side (usually the leg side, but not always),
ready to judge stumpings and run-outs at his end. The umpires remain at their
respective ends of the pitch, thus swapping roles every over. If the technology
is available for a given match, a third umpire is sometimes used. He sits o
the eld, with a television replay monitor. If an on-eld umpire is unsure of a
decision concerning either a run out or a stumping attempt, he may signal for the
third umpire to view a television replay. The third umpire views a replay, in slow
motion if necessary, until he either reaches a decision or decides that he cannot
make a clear decision. He signals the result to the on-eld umpire, who must
then abide by it. If the equipment fails, the replay umpire signals no decision.
The replay umpire cannot be used for any decisions other than run outs and
stumpings. Whenever any decision is in doubt, the umpire must rule in favour
of the batsman. If the ball hits an umpire, it is still live and play continues. If it
lodges in an umpire's clothing, then it is dead. The game is also presided over
by a match referee, who watches from outside the eld. The referee makes no
decisions of relevance to the outcome of the game, but determines penalties for
breaches of various rules and misconduct. In professional games, these penalties
are monetary nes. Arguing with an umpire's decision is simply not tolerated.
Anything more than a polite question to the umpires is heavily frowned upon
and could attract a penalty from the referee. The most serious misconduct in a
cricket match is of the order of a rude gesture to an opponent or throwing the
ball into the ground in disgust. Such gross misbehaviour would attract large

15
nes and possibly match suspensions. Penalties for physical violence can only
be guessed at, but would possibly be a career suspension.
.......................................................................
The Bowling Action
The bowling action itself has to conform to several restrictions. The bowler's
arm must be straight when the ball is bowled (so no throwing is allowed). The
ball must be bowled overarm, not underarm. The dierence between `bowling'
and `throwing': When you throw the ball, the elbow is cocked and used to
impart energy to the ball by straightening. When a ball is bowled, the elbow
joint is held extended throughout. All the energy is imparted by rotation of
the arm about the shoulder, and possibly a little by wrist motion. For a right-
handed bowler, the action goes roughly as follows: After the run-up, the right
foot is planted on the ground with the instep facing the batsman. The right
arm is extended backwards and down at this stage. The left foot comes down
on the popping crease as the bowler's momentum carries him forward - he is
standing essentially left-side on to the batsman. As the weight transfers to the
left foot, the right arm is brought over the shoulder in a vertical arc. The ball
is released near the top of the arc, and the follow-through brings the arm down
and the right shoulder forward rapidly. Bouncing the ball on the pitch is not
mandatory. It's usually done because the movement of the ball o the pitch
makes it much harder to hit.
.......................................................................
Unbounced deliveries, or full tosses are almost always much easier to hit,
and mostly they are bowled accidentally. A full toss above hip height is no
ball, and an umpire who suspects that such a ball was deliberate will give the
bowler an ocial warning. A warning is also given if the umpire believes the
bowler is bowling at the body of a batsman in a deliberate attempt to injure
the batsman. After two warnings a bowler is barred from bowling for the rest
of the innings. If any rule governing the bowling action is violated, a no ball
results. Bowlers are allowed to polish the ball by rubbing it with cloth (usually
on their trouser legs) and applying saliva or sweat to it. Any other substance
is illegal, as is rubbing the ball on the ground. Usually one side of the ball is
polished smooth, while the other wears, so that the bowler can achieve swing
(curving the ball through the air). It is also illegal to roughen the ball by any
means, including scraping it with the ngernails or lifting the seam. A bowler
who illegaly tampers with the ball is immediately suspended from bowling for
the rest of that innings.
.......................................................................
The bowler may bowl from either side of the wicket, but must inform the
umpire and the batsmen if he wishes to change sides. Bowling with the bowling
arm closest to the wicket is called over the wicket, and is most common. Bowling
with the non-bowling are closest to the wicket is called around the wicket. The
bowler may abort his run-up or not let go of the ball if he loses his footing or
timing for any reason. The umpire will signal dead ball and the ball must be
bowled again. If a bowler loses his grip on the ball during the delivery action,
it is considered to be a live ball only if it is propelled forward of the bowler. If

16
such a ball comes to rest in front of the striker, but any distance to the side, the
striker is entitled to walk up to the ball and attempt to hit it with his bat. The
elding team must not touch the ball until the striker either hits it or declines
to do so. A delivery may also be aborted by the striker stepping away from his
stumps, if distracted by an insect or dust in the eye, for example.
.......................................................................
Umpires Signals
The umpires signal various events with gestures, as follows:
............................
Out: When a batsman is out, the umpire making the decision raises one
hand above his head, with the index nger extended.
.......................................................................
Not Out: There is no formal signal to indicate that a batsman is not out.
The umpire can either shake his head `no' or not signal at all.
.......................................................................
Four: A four scored by the ball reaching the boundary is signalled by an
arm extended horizontally and waved briey back and forth in a horizontal arc.
.......................................................................
Six: A six is signalled by raising both arms straight over the head.
.......................................................................
No Ball: A no ball is signalled by holding an arm out horizontally.
.......................................................................
Wide: A wide is signalled by holding both arms out horizontally.
.......................................................................
Byes: Runs scored as byes are signalled by raising one arm over the head,
palm open.
.......................................................................
Leg Byes: Leg byes are signalled by raising one leg and tapping the knee
with one hand.
.......................................................................
Dead Ball: If the umpire has to signal dead ball to prevent the players from
assuming that the ball is still alive, he waves both arms across each other in
front of his abdomen.
.......................................................................
One Short: One short is signalled by touching the tip of one hand to the
same shoulder.
.......................................................................
TV Replay: If an umpire wishes the third umpire to make a decision based
on a TV replay, he signals by drawing a large square shape in the air with both
hands, spreading them out high in the air in front of him, bringing them down,
and then together again.
.......................................................................
The Two Forms of International Cricket.
The rst is limited duration, in which a specic number of hours of playing
time are allocated and each team plays two innings. The second is limited overs,
in which each team plays one innings of a pre-determined number of overs. First

17
Class Cricket First class cricket matches are the most prestigious games, played
at a professional level. The top level games are international Test matches,
played betwen countries. There are also domestic rst class cricket competitions.
First class matches are of limited duration. Test matches will be described rst,
then any dierences for other rst class matches will be described. Test matches
are played over ve days, with six hours play each day. Each day's play is divided
into three sessions of two hours each, with a 40 minute break between the rst
two session for lunch, and a 20 minute tea break between the last two sessions.
A short drinks break is taken once an hour, or more often in very hot weather.
Play usually goes from 11:00 local time to 18:00, although this may be varied
if sunset occurs early. The scheduled close of play time is called stumps. Test
matches are never played under articial lighting.
.......................................................................
Each team has two innings, usually played in alternating order. Each innings
is over when either ten batsmen are out, or the captain of the batting side
declares the innings closed (for strategic reasons, more later). When all the
innings are completed, the team with the most runs wins. If there is a tie, the
result stands (this is rare - it has only ever happened twice). If by the end of the
nal day's play all the innings are not completed, the game is a draw, no matter
who appeared to be winning. Thus the strategic importance of sometimes
declaring an innings closed, in order to have enough time to dismiss the other
team and so win the game. The order of the innings alternates except when
the follow-on is enforced. This can occur if the second team to bat in the rst
innings scores 200 or more runs fewer than the rst team. The captain of the
rst team may then ask the second team to follow on, i.e. to bat its second
innings immediately, and defer his own team's second innings until afterwards.
Whenever a change of innings occurs during a session, a ten minute break is
taken. If the end of an innings occurs within ten minutes of the end of the rst
or second sessions, the ten minute break is lost and the scheduled interval is
shifted to begin immediately. If the end of an innings occurs within ten minutes
of stumps, the day's play ends early. Test matches are played with a red cricket
ball. A new ball is used for the beginning of each innings. The same ball must
be used throughout the innings, being replaced only in the following cases: 1.
The captain of the bowling team may elect to take a new ball at any time after
80 overs have been bowled with the previous ball. 2. If the ball is lost, it is
replaced. 3. If the ball is damaged, either by the stitching coming undone or
the ball becoming clearly non-spherical, it is replaced. In cases 2 and 3, the ball
must be replaced by a previously used ball of similarly worn condition to the
old ball, as chosen by the umpires. If the ball is ever hit so that a spectator
gathers it, the spectator must return it so that play can continue.
.......................................................................
On each day of play in a Test match, a minimum of 90 overs must be bowled.
If the bowling team has not bowled the required minimum by the scheduled
stumps time, play is extended until the required number of overs have been
bowled. Whenever an innings ends, the number of overs to be bowled is recal-
culated, disregarding the number of overs bowled so far during the same day.

18
The required minimum is calculated to be the number of minutes of play re-
maining, divided by 4 and rounded up. On the last day of play, this formula is
used up until one hour before stumps, then fteen overs are added to the result.
If extra overs are bowled before the time one hour before stumps on the nal
day, then there still must be a minimum of fteen overs bowled after the time
one hour before stumps. All of these conditions are recalculated for time lost
due to poor weather, at a rate of one over per 4 minutes of lost time. If a day's
play ends early because of poor weather conditions, all calculations are reset for
the next day. If there is heavy cloud cover, the umpires may decide that the
ambient light level is too low and that the batsmen may be in danger because
of diculty in sighting the ball. If so, they oer the light to the batsmen, who
may agree to leave the eld or may decide to play on. If the light deteriorates
further, the umpires will oer again. If the batsmen decide to leave the eld and
the light improves, the umpires make the decision to resume play. If a elder
leave the eld for any reason and then returns during the same innings, he may
not bowl until he has been on the eld again for as much time as he spent o the
eld. Test matches are played in Series between two of the ocial Test nations.
.......................................................................
A Test Series consists of a set number of matches, from one to six, all of
which are played to completion, even if one team gains an unassailable lead in
the Series. Series of three or ve matches are most common. Some pairs of
nations compete against one another for a perpetual trophy. If a Series between
two such nations is drawn, the holder of the trophy retains it. Non-Test rst
class cricket diers from Test cricket in only a few respects. A non-Test rst
class match is usually four days long, not ve. In a four-day game, the cut-o
gure for enforcing the follow-on is 150 or more runs behind the rst team. The
formula used to determine the minimum number of overs bowled in a non-Test
rst class match may be dierent to that used for a Test match; there is no
standard regulation. Non-Test rst class competitions are usually round-robins
amongst several domestic teams. Other rst class matches include single games
between visiting international sides and domestic rst class teams.
.......................................................................
The above information is taken largely from a Purdue website explaining
cricket rules and been reinterpreted and reorganized in parts. The link to the web-
site is <http://www.cs.purdue.edu/homes/hosking/cricket/explanation.htm>
.......................................................................
.......................................................................
.......................................................................
.......................................................................
.......................................................................
.......................................................................
.......................................................................
.......................................................................
.......................................................................
.......................................................................
.......................................................................

19
 Bibliography
.......................................................................
De Silva, Basil M., and Tim B. Swartz. "Winning the Coin Toss and the
Home Team Advantage in One-Day Inernational Cricket Matches."
<http://www.math.sfu.ca/~tim/papers/cricket.pdf>.

.......................................................................
Stock, James, and Mark Watson. Introduction to Econometrics.
Addison-Wesley, 2002.

.......................................................................
Wooldridge, Jerey. Introductory Econometrics: a Modern Approach.
South-Western College Pub, 2005.

.......................................................................
http://luna.cas.usf.edu/~mbrannic/les/regression/Logistic.html

.......................................................................
http://wikipedia.org/

.........................................................................
http://yudkowsky.net/bayes/bayes.html .

20