Saturday, 24 February 2018

Ranking GOAT in Cricket (May 2014)

1. Introduction

1.1 About Cricket
Cricket is a bat-and-ball game played between two teams of 11 players on a field at the centre of which is a rectangular 22-yard long pitch. Each team takes it in turn to bat, attempting to score runs, while the other team fields. Each turn is known as an innings.
The bowler delivers the ball to the batsman who attempts to hit the ball with his bat far enough for him to run to the other end of the pitch and score a run. Each batsman continues batting until he is out. The batting team continues batting until ten batsmen are out, at which point the teams switch roles and the fielding team comes in to bat.
In professional cricket the length of a game ranges from 20 overs of six bowling deliveries per side to Test cricket played over five days. The Laws of Cricket are maintained by the International Cricket Council (ICC) and the Marylebone Cricket Club (MCC) with additional Standard Playing Conditions for Test matches and One Day Internationals.
Cricket was first played in southern England in the 16th century. By the end of the 18th century, it had developed into the national sport of England. The expansion of the British Empire led to cricket being played overseas and by the mid-19th century the first international matches were being held. The ICC, the game's governing body, has 10 full members. The game is most popular in Australasia, England, the Indian subcontinent, the West Indies and Southern Africa.
1.2 Ratings in Cricket
Cricket is a team game, but the players are also individually ranked based on their performances. The ICC Player Rankings are a widely followed system of rankings for international cricketers based on their recent performances. The ratings were developed at the suggestion of Ted Dexter in 1987. The intention was to produce a better indication of players' current standing in the sport than is provided by comparing their averages. Career averages are based on a player's entire career and do not make any allowance for match conditions or the strength of the opposition, whereas the ratings are biased towards recent form and account for match conditions and the quality of the opponent using statistical measures.
Initially the rankings were for Test cricket only, but separate One Day International rankings were introduced in 1998. Both sets of rankings have now been calculated back to the start of those forms of the game. From 2003 onwards, David Kendix has been responsible for calculating and maintaining the rankings, with the t20 world cups even ranking for t20 is included. The ranking includes the top 10 test and ODI and even t20 batsmen, bowlers and alrounders based on the rating of each player. But there is no rating system by ICC that ranks players on their all time performance.
There are many aspects of Cricket, like batting, bowling, fielding, wicket keeping etc. A player can both bat and bowl as well. So mainly there are three kinds of ratings. First kind of ratings is for the batsman and it is mainly based on the number of runs they score. Second kind of ratings is for the bowlers and it is mainly based on the number of wickets they take. The third and last kind of rating is for the alrounders, i.e. for the players who are good at both batting and bowling and it considers overall performance of a player in both the areas.

1.3 Objective of the Project
The objective of this Project is to develop a rating system that can rank the players of all-time, that is not based on their recent performance. It should be based on the performance of the players over their entire cricket period. Their records should be seen in bulk and not in a particular phase. The ranking should be purely based on the records of a cricketer and not on their reputation of cricketing style.
In this project, a rating system will be developed for batsmen, bowlers and alrounders in all the three formats of Cricket, which are one day internationals, test matches and T20 matches.

1.4 Data for Records
All the required data for the players performance is taken from www.espncricinfo.com. The Players of only 10 countries will be considered. These countries are:
i)                  Australia
ii)                Bangladesh
iii)              England
iv)              India
v)                New Zealand
vi)              Pakistan
vii)             South Africa
viii)           Sri Lanka
ix)               West Indies
x)                Zimbabwe
The data is up to 4th may 2014.


2. Normalization of Records

2.1 Need of Normalization
The Game of Cricket has evolved a lot from time to time. To make the game more interesting, the rules have been modified in such a way that the game becomes more favourable for batsman. The introduction of fielding restrictions and power plays has made the game easier for batsman. So it is necessary to normalize the performance of players so that the effect of change of rules can be minimized.

2.2 Normalization in ODI Cricket

2.2.1     Method for Normalization

       For normalization, it is assumed that the overall average (the average no. of runs the batsman scores before he gets dismissed) and overall strike rate (the average no. of runs the batsman scores in every 100 balls) of batsmen in every year should be same.
       To calculate this, the data of a player is divided equally to every year in which he has played. For example, Sachin Tendulkar has played in a span of 1989-2012, i.e. for 23 years in one day international cricket. In this period he has scored 18426 runs, faced 21367 balls and has been dismissed 411 times. So it will be assumed that he has scored 801.31 runs (18426/23), faced 929 balls (21367/23) and has been dismissed 17.86 times (411/23) every year. After calculating this for all the players, the batting averages and strike rate for each year will be calculated. Then, the averages and strike rate of all the players will be modified so that the performance of players in every year becomes same. Similarly for bowlers, the balls they have delivered, runs conceded from them and wickets they took will be considered and their records will be modified too.
The method used is as follows:
Let us suppose that Cricket is played for three years 2011, 2012 and 2013. The averages of batsman in these years are 28.00, 29.00 and 30.00 respectively. We have a player who has played for 3 years from 2011 to 2013 and has scored 3000 runs and has been dismissed 60 times.
So, current average of player = 3000/60 = 50.00
His modified average will be = (3000/3) * ( (30/28) + (30/29) + (30/30) ) / 60
                                                 = 1000 * 3.105911 / 60
                                                 = 51.77
This will be done for all the players.

2.2.2     Observation in One Day International (ODI) Cricket

       The data of all the 1682 players who have played cricket for these 10 countries were collected and a program was made to do all the necessary calculations. The results obtained were very interesting.



A graph showing batting average of players in each year in ODIs

       This graph shows that how the batting averages of players have increased so much in the recent years with a range of 25.23 to 30.27. The major reason for this is fielding restrictions and introduction of new power play rules in cricket. But still the variation is small. We see an extreme variation in strike rates of cricketers. Here is a graph for that:


       This graph shows how the speed of making runs has increased rapidly ever since 1980. The range of batting strike rate is 64.01 to 82.12.
Similarly, we have the graphs for bowling averages (No. of average runs given to take one wicket), bowling Strike Rate (No. of balls delivered to take one wicket) and Economy Rate (Average no. of runs given in one over).



2.2.3 Modified records in ODIs
Using this information the batting average and batting Strike rates of players is modified. Here is a list of 20 batsmen in Cricket who have scored maximum runs with their real batting average, modified batting average, real batting strike rate and modified batting strike rate.
Name
Runs
Ave
Mod Ave
SR
Mod SR
SR Tendulkar
18426
44.83
46.87
86.23
96.00
RT Ponting
13589
41.81
43.52
80.19
87.36
ST Jayasuriya
13364
32.51
34.06
91.25
101.93
KC Sangakkara
12241
40.39
41.38
77.32
82.08
Inzamam-ul-Haq
11701
39.53
41.60
74.20
83.49
JH Kallis
11545
45.63
47.32
73.19
79.12
DPMD Jayawardene
11243
32.77
33.75
78.07
83.48
SC Ganguly
11221
40.95
43.11
73.65
82.60
R Dravid
10768
39.15
40.78
71.18
77.47
BC Lara
10348
40.90
43.11
79.62
89.89
AC Gilchrist
9595
35.93
37.67
96.89
106.44
Mohammad Yousuf
9554
42.08
43.74
74.91
81.18
M Azharuddin
9378
36.92
39.78
74.02
87.36
PA de Silva
9284
34.90
37.60
81.13
95.10
Saeed Anwar
8824
39.21
41.64
80.67
92.63
S Chanderpaul
8778
41.60
43.43
70.74
77.65
CH Gayle
8688
37.77
38.92
84.17
90.06
DL Haynes
8648
41.37
45.62
63.09
77.71
MS Atapattu
8529
37.57
39.61
67.72
76.45
ME Waugh
8500
39.35
41.91
76.90
88.97
Comparison of real and modified batting averages and strike rates of players

Same modification is applied on the bowling average, bowling strike rate and bowling economy rate of all the players. Here is a list of 20 bowlers who have taken maximum wickets in Cricket. The six columns on the right shows their real bowling average, modified bowling average, real bowling strike rate, modified bowling strike rate, real economy rate and modified economy rate respectively.

Name
Wkts
Ave
Mod Ave
Econ
Mod Econ
SR
Mod SR
M Muralitharan
523
23.07
22.45
3.92
4.22
35.20
31.95
Wasim Akram
502
23.52
22.86
3.89
4.44
36.20
30.95
Waqar Younis
416
23.84
23.16
4.68
5.26
30.50
26.46
WPUJC Vaas
399
27.45
26.71
4.18
4.53
39.40
35.44
SM Pollock
387
24.31
23.64
3.65
3.93
39.80
36.12
B Lee
380
23.36
22.76
4.76
4.97
29.40
27.49
GD McGrath
380
21.98
21.38
3.87
4.23
34.00
30.37
Shahid Afridi
376
33.91
33.01
4.62
4.87
43.90
40.72
A Kumble
334
30.83
29.94
4.29
4.74
43.00
37.98
ST Jayasuriya
320
36.67
35.64
4.77
5.21
46.00
41.16
J Srinath
315
28.08
27.30
4.44
4.96
37.80
33.08
SK Warne
291
25.82
25.13
4.25
4.71
36.40
32.06
AB Agarkar
288
27.85
27.06
5.07
5.43
32.90
29.95
Saqlain Mushtaq
288
21.78
21.21
4.29
4.72
30.40
27.00
DL Vettori
276
31.76
30.86
4.11
4.33
46.30
42.85
AA Donald
272
21.78
21.18
4.15
4.63
31.40
27.48
Z Khan
269
30.11
29.34
4.95
5.17
36.40
34.07
JH Kallis
269
31.85
30.95
4.83
5.10
39.50
36.44
Abdul Razzaq
268
31.53
30.69
4.67
4.97
40.40
37.10
M Ntini
265
24.53
23.86
4.51
4.79
32.60
29.90
Comparison of real and modified bowling averages, bowling strike rates and economy rates of players

2.2.4     Program used to Normalize the data in ODIs
The program used takes its data from a file that contains 7 columns, that are the time span a player played, no. of the times the player was dismissed, runs he scored, balls he faced, no. of balls he delivered, wickets he took and the no. of runs given by him respectively. After this it calculates the averages, strike rates and economy rates for each year and outputs that data to a file. Using this information the above shown 5 graphs have been drawn. After this it modifies the statistics of the players and writes them to another file.
The source code of the program is given in the end.

2.3    Normalization in Test Cricket

2.3.1 Observations in Test Cricket

Similar method is used for normalization of all the records in Test Cricket. There were no test matches played in the years 1915-1919 and 1940-1945. So we have done modification to exclude these years.
So the graphs obtained for the normalization over a period of 138 years of Test Cricket are as follows:
A graph showing batting average of players in each year in Tests

A graph showing bowling average of players in each year in Tests

2.3.2. Modified Records in Tests
Using this information the batting average and batting Strike rates of players is modified. Here is a list of 20 batsmen in Cricket who have scored maximum runs with their real batting average and modified batting average.
Name
Runs
Ave
Mod Ave
SR Tendulkar
15921
53.78
55.82
RT Ponting
13378
51.85
53.90
R Dravid
13265
52.63
54.65
JH Kallis
13206
55.25
57.32
BC Lara
11912
53.17
55.87
DPMD Jayawardene
11319
50.30
51.73
S Chanderpaul
11219
51.93
53.92
AR Border
11174
50.56
53.52
KC Sangakkara
11151
58.07
59.15
SR Waugh
10927
51.06
54.03
SM Gavaskar
10122
51.12
54.21
GC Smith
9253
48.70
49.22
GA Gooch
8900
42.58
45.10
Javed Miandad
8832
52.57
55.75
Inzamam-ul-Haq
8829
50.16
52.74
VVS Laxman
8781
45.97
47.73
ML Hayden
8625
50.73
53.11
IVA Richards
8540
50.23
53.22
V Sehwag
8503
49.43
50.59
AJ Stewart
8463
39.54
41.87

2.4    Normalization in T20Is

2.4.1 Observations in T20Is
The following graph shows the batting average of players in the last 10 years in T20Is. This shows that it is not necessary to normalize the performance of players in T20Is as it is for a very short period of time as compared to other formats of the game.
A graph showing batting average of players in each year in T20Is
So we conclude that there is no need of normalization of records in T20I Cricket.
  
3    Ranking of batsmen

3.1  Minimum Criteria for Players

There are 1682 players who have played Cricket for the 10 countries that we are considered. But we cannot consider all of them for the rankings. The reason is that it will give weird results. As an example, a batsman from England named KJ Barnett played only one ODI match in 1988 and scored 84 runs in it. So his career batting average is 84.00 which is highest ODI batting average by any player. But we cannot judge his performance from a single match. So we need a player to satisfy some minimum criteria so that he can participate in the rankings. Other such examples are CT Radley of England who played only 4 ODI matches and has an batting average of 83.33 and BE Congdon from New Zealand who played 11 matches, made 338 runs but was dismissed only 6 times, so his career batting average is 56.33.
The same problem is with batting Strike rates of players. As an example, AF Milne of New Zealand played only 5 ODIs, but faced only 3 balls and scored 12 runs in that, so his career batting strike rate is 400.00 that is way more than any other player. J Louw of South Africa played only 3 ODIs, got one innings and scored 23 runs off 7 balls in it. So his career batting strike rate is 328.57. It is interesting to know that, the player who is considered as the biggest hitter of cricket ball, Shahid Afridi of Pakistan has a career batting strike rate of 114.57 and there are 34 batsmen who have better strike rate than him, but all of them except AD Russell of West Indies have played less than 20 ODI matches.
So to remove such data, we make the following minimum criteria for a player to be eligible for rankings:
i)                  ODIs
Criteria
Minimum Requirement
Innings
25
Balls Faced
500

ii)                Test
Criteria
Minimum Requirement
 Innings
20
Runs Scored
500

iii)               T20Is
Criteria
Minimum Requirement
 Innings
20
Balls Faced
300

So by these criteria, the numbers of players we are considering are 750 for test cricket, 461 for ODI cricket and 71 for T20Is.

3.2 Ranking in ODI

3.2.1 Parameters Considered for Batsmen Rankings in ODI
This is the most important part of ranking. We have to identify the parameters that are important to rank a batsman. Some of the obvious choices are runs a batsman scored, 100s scored by him, 50s scored by him, his batting average, his batting strike rate etc.
So, the parameters considered here will be:
i)                  Runs: The no. of runs a batsman has scored is a very important parameter as it shows his dominance in cricket.
ii)                Modified batting average: It is the most important parameter.
iii)               Modified batting strike rate: It shows how fast a player can score runs. It is very important in ODI matches.
iv)               100s: It shows the temperament of a player to play long innings
v)                 50s: Shows the consistency of batsman.
vi)               50+ Scores/Inn: Shows the no. of times a batsman scores more than 50 runs per inning
vii)              Span of Years Played: More no. of years shows consistency
Each of these parameters should have a particular weight-age.

3.2.2 Normalization of Parameters in ODI
To apply the weight age easily we have to normalize all the parameters. For each parameter we will give points out of 100 to a batsman. These points should be relative. It is done as follows:
Relative Points for Runs:
Sachin Tendulkar has scored maximum 18,426 runs in ODI Cricket. So he will receive 100.00 points for it.
Any batsman who has scored “x” runs in ODI Cricket will get = (x*100)/18426 points
Similar process is followed for 100s, 50s and 50+Scores/Inn.
But the process for relative points for modified batting average and modified batting strike rate is different.
Relative points for Modified batting average:
Any batsman whose modified batting average is below 10.00 will be given 0 points.
MG Bevan has maximum modified batting average of 56.48. So he will get 100.00 points for it.
Any other batsman with a modified batting average of “x” will get = ((x-10)*100)/46.58 points
Relative points for Modified strike rate:
Any batsman whose modified batting strike rate is below 60.00 will be given 0 points.
BL Cairns has the highest modified batting strike rate of 128.83. So he will get 100.00 points for it.
Any other batsman who has a modified batting strike rate of “x” will get = ((x-60)*100)/68.83 points
So here is a list of given points for each of these parameters to 20 Indian batsman who has maximum runs in ODI matches.

Name
Runs
100s
50s
50+/Inn
Ave
SR
SR Tendulkar
100.00
100.00
100.00
75.16
78.86
50.19
SC Ganguly
60.90
44.90
73.96
73.36
70.81
31.50
R Dravid
58.44
24.49
85.42
70.14
65.84
24.35
M Azharuddin
50.90
14.29
60.42
49.44
63.68
38.14
Yuvraj Singh
44.70
26.53
53.13
56.58
58.37
45.62
V Sehwag
43.39
30.61
38.54
51.84
56.57
72.14
MS Dhoni
42.72
16.33
56.25
68.84
93.10
45.76
V Kohli
30.58
38.78
31.25
91.11
90.63
45.61
A Jadeja
29.08
12.24
31.25
47.12
63.74
28.28
G Gambhir
28.43
22.45
35.42
73.73
65.17
41.85
SK Raina
24.94
6.12
30.21
46.28
55.10
49.22
NS Sidhu
23.95
12.24
34.38
71.95
63.54
31.06
K Srikkanth
22.20
8.16
28.13
50.09
47.47
39.42
N Kapil Dev
20.53
2.04
14.58
17.75
34.71
79.61
DB Vengsarkar
19.04
2.04
23.96
46.86
61.09
33.69
RG Sharma
18.60
8.16
22.92
52.06
55.53
29.86
RJ Shastri
16.87
8.16
18.75
40.27
47.54
21.11
SM Gavaskar
16.78
2.04
28.13
64.31
62.78
25.22
M Kaif
14.94
4.08
17.71
40.47
49.66
25.33
VG Kambli
13.44
4.08
14.58
38.65
52.54
32.05

3.2.3       Weightage for Parameters

Now as we have seven parameters that we will use to decide the all time greatest ODI batsman, we need to give some weightage to these parameters. Weightage should be according to the importance of the parameters. More important parameters should have a higher weightage than the others. It is a very difficult task to assign such weightage that will give the desired results as it can be biased and it can vary the results significantly. To do so we give a range of weightage to each of the parameters and after that we will use Monte Carlo simulation to get the desired results. The range decided for each parameter is as follows:
Parameter
Range
Parameter
Range
Runs
80 – 180
100s
8 – 35
Mod Bat Average
60 – 150
50s
5 – 20
Mod Bat Strike Rate
40 – 110
50+Score/Inn
20    – 40

3.2.3 Monte Carlo Simulation

To use Monte Carlo Simulation to find the rank of batsmen, we take a random weightage for each parameter in the above mentioned ranges. After that we give point to each player according to this formula:
Points =             (weightage of runs * points for runs) + (weightage of average * points for average) + (weightage of strike rate * points for strike rate) + (weightage of 100s * points for 100s) + (weightage of 50s * points for 50s) + (weightage of 50+/Inn * points for 50+/Inn)
After calculating points for each player, we rank them according to the number of points they have. We carry on this process for 50,000 times and in the end the average rank of a player determines his final rank.

3.2.4       Final Rank List of all time Batsmen in ODI Cricket

After carrying out the Monte Carlo simulation using the program given at the end, the results that we got for the top 50 ODI batsmen of all time are as follows:

Rank
Name
Team
Avg Rank
Mat
Runs
Ave
SR
100
50
1
SR Tendulkar
Ind
1.00
463
18426
44.83
86.23
49
96
2
RT Ponting
Aus
2.19
374
13589
41.81
80.19
29
82
3
IVA Richards
WI
4.20
187
6721
47.00
90.20
11
45
4
JH Kallis
SA
4.49
320
11545
45.63
73.19
17
86
5
ST Jayasuriya
SL
4.98
441
13364
32.51
91.25
28
68
6
MS Dhoni
Ind
7.45
240
7872
52.83
88.67
8
54
7
KC Sangakkara
SL
7.67
362
12241
40.39
77.32
18
82
8
SC Ganguly
Ind
8.12
308
11221
40.95
73.65
22
71
9
BC Lara
WI
9.74
295
10348
40.90
79.62
19
62
10
AB de Villiers
SA
10.50
154
6181
50.25
93.92
16
35
11
Inzamam-ul-Haq
Pak
11.56
375
11701
39.53
74.20
10
83
12
AC Gilchrist
Aus
11.74
286
9595
35.93
96.89
16
55
13
V Kohli
Ind
12.58
134
5634
52.16
89.87
19
30
14
MG Bevan
Aus
14.75
232
6912
53.58
74.16
6
46
15
M Yousuf
Pak
17.03
281
9554
42.08
74.91
15
62
16
Saeed Anwar
Pak
17.14
247
8824
39.21
80.67
20
43
17
R Dravid
Ind
17.36
340
10768
39.15
71.18
12
82
18
HM Amla
SA
18.09
85
4054
53.34
90.14
12
23
19
ME Waugh
Aus
20.10
244
8500
39.35
76.90
18
50
20
V Sehwag
Ind
20.33
241
7995
35.37
104.44
15
37
21
DL Haynes
WI
20.49
238
8648
41.37
63.09
17
57
22
PA de Silva
SL
21.25
308
9284
34.90
81.13
11
64
23
Zaheer Abbas
Pak
24.31
62
2572
47.62
84.80
7
13
24
DM Jones
Aus
24.55
164
6068
44.61
72.56
7
46
25
CH Gayle
WI
24.86
252
8688
37.77
84.17
21
44
26
MJ Clarke
Aus
26.48
236
7683
44.66
78.76
8
55
27
DPMD Jayawardene
SL
27.82
407
11243
32.77
78.07
15
69
28
S Chanderpaul
WI
28.06
268
8778
41.60
70.74
11
59
29
M Azharuddin
Ind
28.55
334
9378
36.92
74.02
7
58
30
MEK Hussey
Aus
30.48
185
5442
48.15
87.16
3
39
31
Javed Miandad
Pak
31.23
233
7381
41.70
67.01
8
50
32
ML Hayden
Aus
31.25
160
6131
44.10
78.98
10
36
33
Yuvraj Singh
Ind
32.18
290
8237
36.28
86.98
13
51
34
HH Gibbs
SA
32.54
248
8094
36.13
83.26
21
37
35
CG Greenidge
WI
35.33
128
5134
45.03
64.92
11
31
36
TM Dilshan
SL
35.68
277
8025
37.67
85.70
17
34
37
G Kirsten
SA
36.06
185
6798
40.95
72.04
13
45
38
A Ranatunga
SL
39.23
269
7456
35.84
77.90
4
49
39
MS Atapattu
SL
39.71
268
8529
37.57
67.72
11
59
40
GC Smith
SA
41.48
196
6989
38.19
80.86
10
47
41
SR Watson
Aus
41.97
173
5256
41.06
90.20
9
30
42
A Symonds
Aus
42.30
198
5088
39.75
92.44
6
30
43
GJ Bailey
Aus
42.69
39
1647
53.12
91.39
2
12
44
G Gambhir
Ind
44.91
147
5238
39.68
85.25
11
34
45
Shahid Afridi
Pak
46.17
373
7582
23.69
115.54
6
36
46
IJL Trott
Eng
47.06
68
2819
51.25
77.06
4
22
47
A Flower
Zim
49.00
213
6786
35.34
74.59
4
55
48
DR Martyn
Aus
49.38
208
5346
40.80
77.73
5
37
49
Saleem Malik
Pak
49.56
283
7170
32.88
76.41
5
47
50
RR Sarwan
WI
50.09
181
5804
42.67
75.74
5
38

3.3 Ranking in Test Cricket

3.2.1 Parameters Considered for Batsmen Rankings in Test Cricket
Now we identify the parameters that are important to rank a batsman. Some of the obvious choices are runs a batsman scored, 100s scored by him, 50s scored by him, his batting average and 50+/Inn.
So, the parameters considered here will be:
i)                  Runs: The no. of runs a batsman has scored is a very important parameter as it shows his dominance in cricket.
ii)                Modified batting average: It is the most important parameter.
iii)               100s: It shows the temperament of a player to play long innings
iv)               50s: Shows the consistency of batsman.
v)                50+ Scores/Inn: Shows the no. of times a batsman scores more than 50 runs per inning
Each of these parameters should have a particular weight-age.
3.2.2 Normalization of Parameters in Test
To apply the weight age easily we have to normalize all the parameters. For each parameter we will give points out of 100 to a batsman. These points should be relative. It is done as follows:
Relative Points for Runs:
Sachin Tendulkar has scored maximum 15,921 runs in Test Cricket. So he will receive 100.00 points for it.
Any batsman who has scored “x” runs in Test Cricket will get = (x*100)/15921 points
Similar process is followed for 100s, 50s and 50+Scores/Inn.
But the process for relative points for modified batting average.
 Relative points for Modified batting average:
Any batsman whose modified batting average is below 20.00 will be given 0 points.
DG Bradman has maximum modified batting average of 101.869. So he will get 100.00 points for it.
Any other batsman with a modified batting average of “x” will get = ((x-20)*100)/81.869 points
3.2.5       Weightage for Parameters
Now as we have five parameters that we will use to decide the all time greatest Test batsman, we need to give some weightage to these parameters. Weightage should be according to the importance of the parameters. More important parameters should have a higher weightage than the others. It is a very difficult task to assign such weightage that will give the desired results as it can be biased and it can vary the results significantly. To do so we give a range of weightage to each of the parameters and after that we will use Monte Carlo simulation to get the desired results. The range decided for each parameter is as follows:
Parameter
Range
Parameter
Range
Runs
50 – 100
100s
10 – 40
Mod Bat Average
200 – 400
50s
10 – 25
50+Score/Inn
100 – 200






3.2.3 Monte Carlo Simulation
To use Monte Carlo Simulation to find the rank of batsmen, we take a random weight-age for each parameter in the above mentioned ranges. After that we give point to each player according to this formula:
Points =             (weightage of runs * points for runs) + (weightage of average * points for average) + (weightage of 100s * points for 100s) + (weightage of 50s * points for 50s) + (weightage of 50+/Inn * points for 50+/Inn)
After calculating points for each player, we rank them according to the number of points they have. We carry on this process for 50,000 times and in the end the average rank of a player determines his final rank.
3.2.6       Final Rank List of all time Batsmen in Test Cricket
After carrying out the Monte Carlo simulation using the program given at the end, the results that we got for the top 20 Test batsmen of all time are as follows:
Rank
Name
Team
Avg Rank
Match
Ave
100
50
1
DG Bradman
Aus
1.00
52
99.94
29
13
2
SR Tendulkar
Ind
2.09
200
53.78
51
68
3
JH Kallis
SA
3.26
165
55.25
45
58
4
KC Sangakkara
SL
4.61
122
58.07
35
45
5
KF Barrington
Eng
5.11
82
58.67
20
35
6
RT Ponting
Aus
6.92
168
51.85
41
62
7
H Sutcliffe
Eng
7.20
54
60.73
16
23
8
R Dravid
Ind
8.38
163
52.63
36
63
9
JB Hobbs
Eng
9.08
61
56.94
15
28
10
ED Weekes
WI
10.02
48
58.61
15
19
11
BC Lara
WI
10.10
130
53.17
34
48
12
GS Sobers
WI
11.87
93
57.78
26
30
13
S Chanderpaul
WI
13.78
153
51.93
29
62
14
SM Gavaskar
Ind
13.93
125
51.12
34
45
15
RG Pollock
SA
16.16
23
60.97
7
11
16
AR Border
Aus
16.32
156
50.56
27
63
17
L Hutton
Eng
16.43
79
56.67
19
33
18
CL Walcott
WI
18.32
44
56.68
15
14
19
Javed Miandad
Pak
19.97
124
52.57
23
43
20
SR Waugh
Aus
19.99
168
51.06
32
50




3.3    Ranking in T20I Cricket

3.2.1 Parameters Considered for Batsmen Rankings in T20I Cricket
              The parameters considered will be same as considered for ODI Cricket. But the weigthage for the parameters will be different as it is a fast game and requires a better batting strike rate.
So, the parameters considered here will be:
i)                  Runs
ii)                Modified batting average
iii)               Modified batting strike rate
iv)               100s
v)                50s
vi)               50+ Scores/Inn
Each of these parameters should have a particular weight-age.
3.2.2 Normalization of Parameters in T20I Cricket
To apply the weight age easily we have to normalize all the parameters. For each parameter we will give points out of 100 to a batsman. These points should be relative. It is done as follows:
Relative Points for Runs:
BB McCullum has scored maximum 2044 runs in T20I Cricket. So he will receive 100.00 points for it.
Any batsman who has scored “x” runs in Test Cricket will get = (x*100)/2044 points
Similar process is followed for 100s, 50s and 50+Scores/Inn.
But the process for relative points for batting average and strike rate is different.
 Relative points for batting average:
Any batsman whose batting average is below 10.00 will be given 0 points. Virat Kohli has maximum batting average of 48.14. So he will get 100.00 points for it. Any other batsman with a modified batting average of “x” will get = ((x-10)*100)/38.14 points
Relative points for batting average:
Any batsman whose strike rate is below 100.00 will be given 0 points. KA Pollard has maximum batting strike rate of 149.73. So he will get 100.00 points for it.
Any other batsman with a strike rate of “x” will get = ((x-100)*100)/49.73 points

3.3.3       Weightage for Parameters
Now as we have six parameters that we will use to decide the all time greatest ODI batsman, we need to give some weightage to these parameters. Weightage should be according to the importance of the parameters. More important parameters should have a higher weightage than the others. It is a very difficult task to assign such weightage that will give the desired results as it can be biased and it can vary the results significantly. To do so we give a range of weightage to each of the parameters and after that we will use Monte Carlo simulation to get the desired results. The range decided for each parameter is as follows:
Parameter
Range
Parameter
Range
Runs
80 – 180
100s
5 – 10
Bat Average
80 – 180
50s
5 – 10
50+Score/Inn
10 – 20
Bat Strike Rate
80-180




3.2.3 Monte Carlo Simulation
To use Monte Carlo Simulation to find the rank of batsmen, we take a random weight-age for each parameter in the above mentioned ranges. After that we give point to each player according to this formula:
Points =             (weightage of runs * points for runs) + (weightage of average * points for average) + (weightage of 100s * points for 100s) + (weightage of 50s * points for 50s) + (weightage of 50+/Inn * points for 50+/Inn)
After calculating points for each player, we rank them according to the number of points they have. We carry on this process for 50,000 times and in the end the average rank of a player determines his final rank.
3.3.4       Final Rank List of all time Batsmen in Test Cricket
After carrying out the Monte Carlo simulation using the program given at the end, the results that we got for the top 20 Test batsmen of all time are as follows:
Rank
Name
Team
Ave Rank
Mat
Runs
Ave
SR
1
BB McCullum
NZ
1.00
68
2044
35.85
135.81
2
KP Pietersen
Eng
2.10
37
1176
37.93
141.51
3
V Kohli
Ind
4.23
27
906
45.30
129.98
4
CH Gayle
WI
4.73
42
1239
32.60
135.55
5
DPMD Jayawardene
SL
4.84
55
1493
31.76
133.18
6
DA Warner
Aus
6.87
51
1391
28.38
138.96
7
AD Hales
Eng
7.02
31
956
36.76
136.37
8
Yuvraj Singh
Ind
7.07
40
968
31.22
144.69
9
SR Watson
Aus
8.14
45
1074
26.85
146.32
10
JP Duminy
SA
9.56
55
1342
37.27
124.48
11
MEK Hussey
Aus
11.01
38
721
37.94
136.29
12
SK Raina
Ind
11.95
43
922
32.92
135.78
13
CL White
Aus
14.23
44
918
31.65
133.81
14
MJ Guptill
NZ
15.63
47
1241
31.82
122.02
15
TM Dilshan
SL
15.84
61
1452
29.04
119.3
16
KC Sangakkara
SL
15.92
56
1382
31.40
119.55
17
Shahid Afridi
Pak
17.37
74
1112
19.17
144.04
18
GC Smith
SA
17.54
33
982
31.67
127.53
19
EJG Morgan
Eng
18.13
48
1071
28.94
129.5
20
KA Pollard
WI
19.63
37
569
22.76
149.73

  
4        Ranking of Bowlers

3.2  Minimum Criteria for Players
As described for the batsmen, similarly for bowlers we cannot consider all the players while we are developing the rankings for best bowlers in the history of cricket. There are many instances when a single bowler has taken just one wicket giving very less runs, making his bowling average the best in the world. As an example, South African wicket keeper Mark Boucher has taken only one wicket in his ODI career giving only 6 runs. So his career bowling average is 6.00. Similarly many bowlers have economy rate of 0.00. So it is important to set minimum criteria for the removal of such players.
So to remove such data, we make the following minimum criteria for a player to be eligible for rankings:
iv)               ODIs
Criteria
Minimum Requirement
Overs Bowled
250

v)                Test
Criteria
Minimum Requirement
 Overs Bowled
500

vi)               T20Is
Criteria
Minimum Requirement
 Innings
20
Balls Faced
300

So by these criteria, the numbers of players we are considering are 750 for test cricket, 461 for ODI cricket and 71 for T20Is.

3.2 Ranking in ODI
3.2.1 Parameters Considered for Batsmen Rankings in ODI
This is the most important part of ranking. We have to identify the parameters that are important to rank a batsman. Some of the obvious choices are runs a batsman scored, 100s scored by him, 50s scored by him, his batting average, his batting strike rate etc.
So, the parameters considered here will be:
viii)            Runs: The no. of runs a batsman has scored is a very important parameter as it shows his dominance in cricket.
ix)               Modified batting average: It is the most important parameter.
x)                 Modified batting strike rate: It shows how fast a player can score runs. It is very important in ODI matches.
xi)               100s: It shows the temperament of a player to play long innings
xii)              50s: Shows the consistency of batsman.
xiii)            50+ Scores/Inn: Shows the no. of times a batsman scores more than 50 runs per inning
xiv)            Span of Years Played: More no. of years shows consistency
Each of these parameters should have a particular weight-age.
3.2.2 Normalization of Parameters in ODI
To apply the weight age easily we have to normalize all the parameters. For each parameter we will give points out of 100 to a batsman. These points should be relative. It is done as follows:
Relative Points for Runs:
Sachin Tendulkar has scored maximum 18,426 runs in ODI Cricket. So he will receive 100.00 points for it.
Any batsman who has scored “x” runs in ODI Cricket will get = (x*100)/18426 points
Similar process is followed for 100s, 50s and 50+Scores/Inn.
But the process for relative points for modified batting average and modified batting strike rate is different.
Relative points for Modified batting average:
Any batsman whose modified batting average is below 10.00 will be given 0 points.
MG Bevan has maximum modified batting average of 56.48. So he will get 100.00 points for it.
Any other batsman with a modified batting average of “x” will get = ((x-10)*100)/46.58 points
Relative points for Modified strike rate:
Any batsman whose modified batting strike rate is below 60.00 will be given 0 points.
BL Cairns has the highest modified batting strike rate of 128.83. So he will get 100.00 points for it.
Any other batsman who has a modified batting strike rate of “x” will get = ((x-60)*100)/68.83 points
So here is a list of given points for each of these parameters to 20 Indian batsman who has maximum runs in ODI matches.
Name
Runs
100s
50s
50+/Inn
Ave
SR
SR Tendulkar
100.00
100.00
100.00
75.16
78.86
50.19
SC Ganguly
60.90
44.90
73.96
73.36
70.81
31.50
R Dravid
58.44
24.49
85.42
70.14
65.84
24.35
M Azharuddin
50.90
14.29
60.42
49.44
63.68
38.14
Yuvraj Singh
44.70
26.53
53.13
56.58
58.37
45.62
V Sehwag
43.39
30.61
38.54
51.84
56.57
72.14
MS Dhoni
42.72
16.33
56.25
68.84
93.10
45.76
V Kohli
30.58
38.78
31.25
91.11
90.63
45.61
A Jadeja
29.08
12.24
31.25
47.12
63.74
28.28
G Gambhir
28.43
22.45
35.42
73.73
65.17
41.85
SK Raina
24.94
6.12
30.21
46.28
55.10
49.22
NS Sidhu
23.95
12.24
34.38
71.95
63.54
31.06
K Srikkanth
22.20
8.16
28.13
50.09
47.47
39.42
N Kapil Dev
20.53
2.04
14.58
17.75
34.71
79.61
DB Vengsarkar
19.04
2.04
23.96
46.86
61.09
33.69
RG Sharma
18.60
8.16
22.92
52.06
55.53
29.86
RJ Shastri
16.87
8.16
18.75
40.27
47.54
21.11
SM Gavaskar
16.78
2.04
28.13
64.31
62.78
25.22
M Kaif
14.94
4.08
17.71
40.47
49.66
25.33
VG Kambli
13.44
4.08
14.58
38.65
52.54
32.05

4.2.3       Weightage for Parameters
Now as we have seven parameters that we will use to decide the all time greatest ODI batsman, we need to give some weightage to these parameters. Weightage should be according to the importance of the parameters. More important parameters should have a higher weightage than the others. It is a very difficult task to assign such weightage that will give the desired results as it can be biased and it can vary the results significantly. To do so we give a range of weightage to each of the parameters and after that we will use Monte Carlo simulation to get the desired results. The range decided for each parameter is as follows:
Parameter
Range
Parameter
Range
Runs
80 – 180
100s
8 – 35
Mod Bat Average
60 – 150
50s
5 – 20
Mod Bat Strike Rate
40 – 110
50+Score/Inn
21    – 40

3.2.3 Monte Carlo Simulation
To use Monte Carlo Simulation to find the rank of batsmen, we take a random weightage for each parameter in the above mentioned ranges. After that we give point to each player according to this formula:
Points =             (weightage of runs * points for runs) + (weightage of average * points for average) + (weightage of strike rate * points for strike rate) + (weightage of 100s * points for 100s) + (weightage of 50s * points for 50s) + (weightage of 50+/Inn * points for 50+/Inn)
After calculating points for each player, we rank them according to the number of points they have. We carry on this process for 50,000 times and in the end the average rank of a player determines his final rank.
4.2.4       Final Rank List of all time Batsmen in ODI Cricket
After carrying out the Monte Carlo simulation using the program given at the end, the results that we got for the top 50 ODI batsmen of all time are as follows:
Rank
Name
Team
Avg Rank
Mat
Runs
Ave
SR
100
50
1
SR Tendulkar
Ind
1.00
463
18426
44.83
86.23
49
96
2
RT Ponting
Aus
2.19
374
13589
41.81
80.19
29
82
3
IVA Richards
WI
4.20
187
6721
47.00
90.20
11
45
4
JH Kallis
SA
4.49
320
11545
45.63
73.19
17
86
5
ST Jayasuriya
SL
4.98
441
13364
32.51
91.25
28
68
6
MS Dhoni
Ind
7.45
240
7872
52.83
88.67
8
54
7
KC Sangakkara
SL
7.67
362
12241
40.39
77.32
18
82
8
SC Ganguly
Ind
8.12
308
11221
40.95
73.65
22
71
9
BC Lara
WI
9.74
295
10348
40.90
79.62
19
62
10
AB de Villiers
SA
10.50
154
6181
50.25
93.92
16
35
11
Inzamam-ul-Haq
Pak
11.56
375
11701
39.53
74.20
10
83
12
AC Gilchrist
Aus
11.74
286
9595
35.93
96.89
16
55
13
V Kohli
Ind
12.58
134
5634
52.16
89.87
19
30
14
MG Bevan
Aus
14.75
232
6912
53.58
74.16
6
46
15
M Yousuf
Pak
17.03
281
9554
42.08
74.91
15
62
16
Saeed Anwar
Pak
17.14
247
8824
39.21
80.67
20
43
17
R Dravid
Ind
17.36
340
10768
39.15
71.18
12
82
18
HM Amla
SA
18.09
85
4054
53.34
90.14
12
23
19
ME Waugh
Aus
20.10
244
8500
39.35
76.90
18
50
20
V Sehwag
Ind
20.33
241
7995
35.37
104.44
15
37
21
DL Haynes
WI
20.49
238
8648
41.37
63.09
17
57
22
PA de Silva
SL
21.25
308
9284
34.90
81.13
11
64
23
Zaheer Abbas
Pak
24.31
62
2572
47.62
84.80
7
13
24
DM Jones
Aus
24.55
164
6068
44.61
72.56
7
46
25
CH Gayle
WI
24.86
252
8688
37.77
84.17
21
44
26
MJ Clarke
Aus
26.48
236
7683
44.66
78.76
8
55
27
DPMD Jayawardene
SL
27.82
407
11243
32.77
78.07
15
69
28
S Chanderpaul
WI
28.06
268
8778
41.60
70.74
11
59
29
M Azharuddin
Ind
28.55
334
9378
36.92
74.02
7
58
30
MEK Hussey
Aus
30.48
185
5442
48.15
87.16
3
39
31
Javed Miandad
Pak
31.23
233
7381
41.70
67.01
8
50
32
ML Hayden
Aus
31.25
160
6131
44.10
78.98
10
36
33
Yuvraj Singh
Ind
32.18
290
8237
36.28
86.98
13
51
34
HH Gibbs
SA
32.54
248
8094
36.13
83.26
21
37
35
CG Greenidge
WI
35.33
128
5134
45.03
64.92
11
31
36
TM Dilshan
SL
35.68
277
8025
37.67
85.70
17
34
37
G Kirsten
SA
36.06
185
6798
40.95
72.04
13
45
38
A Ranatunga
SL
39.23
269
7456
35.84
77.90
4
49
39
MS Atapattu
SL
39.71
268
8529
37.57
67.72
11
59
40
GC Smith
SA
41.48
196
6989
38.19
80.86
10
47
41
SR Watson
Aus
41.97
173
5256
41.06
90.20
9
30
42
A Symonds
Aus
42.30
198
5088
39.75
92.44
6
30
43
GJ Bailey
Aus
42.69
39
1647
53.12
91.39
2
12
44
G Gambhir
Ind
44.91
147
5238
39.68
85.25
11
34
45
Shahid Afridi
Pak
46.17
373
7582
23.69
115.54
6
36
46
IJL Trott
Eng
47.06
68
2819
51.25
77.06
4
22
47
A Flower
Zim
49.00
213
6786
35.34
74.59
4
55
48
DR Martyn
Aus
49.38
208
5346
40.80
77.73
5
37
49
Saleem Malik
Pak
49.56
283
7170
32.88
76.41
5
47
50
RR Sarwan
WI
50.09
181
5804
42.67
75.74
5
38



1 comment:

  1. how much free time do we have!!
    you wrote this and i read it!!
    :) :)

    ReplyDelete