BOWLING STRATEGY BUILDING IN LIMITED OVER CRICKET MATCH : AN - - PowerPoint PPT Presentation

BOWLING STRATEGY BUILDING IN LIMITED OVER CRICKET MATCH : AN APPLICATION OF STATISTICS

Akash Adhikari Rishabh Saraf Rishikesh Parma

About the game

• Cricket is a bat and ball game, and is one of

those sports which has been evolving with time .

• The format of the game include test cricket which

can go as long as five days, One day International (ODI) comprising of 50 overs and T20 which limits to 20 overs.

• With the introduction of new rules like batting

and bowling powerplay (a field restriction where

• nly limited number of fielders are allowed
• utside the 30 yard circle) have made the game

an interesting one .

Introduction

• Implementation of new approach to analyze

bowler’s performance in limited over cricket match.

• Current format of the game largely depends on

bowling performance.

• Application of statistics.
• The statistics mainly used to decide bowler’s

performance are bowling average, economy rate, and strike rate. However these statistics are individually deficient as they do not adequately account for overs, wickets and runs respectively.

Data

• The data of every ball bowled by a bowler in his

career during the years 2007-2016 has been used in the analysis.

• R script programming is used to extract data from

www.espncricinfo.com.

• All the analysis and statistical operations has been

done in microsoft excel.

Parameters

Runs between wickets (ri) : runs conceded in between fall of consecutive wickets in a spell of a respective bowler . Balls between wickets (bi) : balls delivered in between fall of consecutive wicket in a spell of a respective bowler .

Bowling analysis

Analysis of bowler has been done separately for the first 30 overs (SET 1) and the last 20

• vers (SET 2) .

30 different bowlers were analyzed, out of which five best rated fast-arm seamers and slow-arm spinners (as per ICC ODI rankings 2016) is mentioned .

Fast Arm Seamers

• MA Starc (Australia)
• TA Boult (New Zealand)
• DW Steyn (South Africa)
• M Morkel (South Africa)
• TG Southee (New Zealand)

Slow Arm Spinners

• Imran Tahir (South Africa)
• SP Narine (West Indies)
• Shakib Hasan (Bangladesh)
• R Ashwin (India)
• RA Jadeja (India)

Probability model

The probability for a bowler to take wicket given the number of runs he may concede P(rn) is defined as, P(rn) = F(rn)÷wt + P(rn-1) Here, wt = total number of wickets taken by the bowler, rn = n {0 ≤ n ≤ max (ri) } Frequency of (rn), F(rn) = number of occurences of rn among all the values of ri .

The probability for a bowler to take wicket given the number of balls to be bowled by him P(bk) is defined as, P(bk) = F(bk)÷wt + P(bk-1) Here, wt = total number of wickets taken by the bolwer, bk = k {1 ≤ k ≤ max (bi) }, P(b0) = 0, Frequency of (bk), F(bk) = number of occurences of bk among all the values of bi .

Procedure

• The

value

• f

P(rn) and P(bk) are calculated independently, considering the effect of only one parameter at a time .

• Since the permitted balls for a bowler in ODI cricket is

60, so the value of P(bk) is calculated for the domain [0,60] .

• Assuming that the maximum runs a bowler may

concede in an ODI cricket match is 80, the value of P(rn) is calculated for the domain [0,80] .

R ASHWIN ANALYSIS

Analysis in SET 1 ( 1-30 overs)

Matches 92 Runs Conceded 2663 Overs 616.2 Wickets 67 Extras 110

Calculated values of P(bk)

bk P(bk) 10 0.1194 20 0.2537 30 0.3582 40 0.5074 50 0.6119 60 0.6865

Calculated values of P(rn)

rn P(rn) 0.0298 10 0.2089 20 0.4179 30 0.5522 40 0.5970 50 0.7611 60 0.8208 70 0.8656 80 0.8805

Analysis in SET 2 ( 31-50 overs)

Matches 95 Runs Conceded 1721 Overs 329 Wickets 76 Extras 80

Calculated values of P(bk)

bk P(bk) 10 0.2666 20 0.4800 30 0.6400 40 0.8266 50 0.8800 60 0.9600

Calculated values of P(rn)

rn P(rn) 0.0533 10 0.3600 20 0.5733 30 0.6666 40 0.8000 50 0.9066 60 0.9600 70 0.9733 80 0.9999

• Similar analysis has been done for all the

bowlers mentioned before .

• Separately for fast-arm seamers and slow-arm

spinners.

• The calculated probabilities is compared with

the help of dominance curve.

DOMINANCE CURVES

SLOW-ARM SPINNERS

Bowlers Indicators Imran Tahir SP Narine Shakib Al Hasan R Ashwin RA Jadeja

Dominance curve of different spinners for SET 1 and with P(bk) as ordinate and bk as abscissa. Dominance curve of different spinners for SET 1 and with P(rn) as ordinate and rn as abscissa.

Bowlers Indicators Imran Tahir SP Narine Shakib Al Hasan R Ashwin RA Jadeja

Dominance curve of different spinners for SET 2 and with P(bk) as ordinate and bk as abscissa. Dominance curve of different spinners for SET 2 and with P(rn) as ordinate and rn as abscissa.

FAST-ARM SEAMERS

Dominance curve for different seamers for SET 1 and with P(bk) as ordinate and bk as abscissa. Dominance curve for different seamers for SET 1 and with P(rn) as ordinate and rn as abscissa. Bowlers Indicators MA Starc TA Boult DW Steyn M Morkel TG Southee

Dominance curve for different seamers for SET 2 and with P(bk) as ordinate and bk as abscissa. Dominance curve for different seamers for SET 2 and with P(rn) as ordinate and rn as abscissa. Bowlers Indicators MA Starc TA Boult DW Steyn M Morkel TG Southee

Discussions

• The probability of a bowler to take a wicket

given the number of balls he can deliver P(bk)

• r the number of runs he can concede P(rn)

does not entirely support each other .

• There may be instances where an exclusive

preference for either of the parameter does not exists .

• So we defined a new function ‘Striking score’ to

include the effect of both the parameters.

STRIKING SCORE AND DOMINANCE FACTOR

Striking score (Y) of a bowler as a function

• f bk is defined as,

Y = P + bk÷rn Given bk , P is equal to P(bk), and rn is equal to that value for which P(bk) = P(rn).

Striking score (Y) of a bowler is a more accurate value to analyze his performance, which includes the effect of both bk and rn in line with the probability to take wickets .

• Striking score is discrete values which

depends on bk .

• For

analysing

• verall

performance, Dominance factor is defined, which is the weighted average of striking score (Y) with respect to bk . D.F. = {∑(Y* bk) } ÷ {∑bk}

Striking score and Dominance factor

• f R Ashwin

bk Striking Score (Y)

10 2.1194 20 1.7921 30 2.1229 40 1.8867 50 1.8314 60 1.8314 D.F. 1.9391

SET 1 SET 2 bk Striking Score (Y)

10

1.9333

20

1.8133

30

1.7511

40

1.7357

50

1.9216

60

1.9944

D.F.

1.8729

Striking score and Dominance factor for SET 1 (Spinners)

bk IMRAN TAHIR SP NARINE S HASAN R ASHWIN R JADEJA 10 2.6694 2.7244 2.7043 2.1194 1.3111 20 1.8079 2.6303 1.8264 1.7921 1.9589 30 1.9823 2.5102 2.2485 2.1229 1.9153 40 2.1589 2.5170 2.1158 1.8867 2.0461 50 2.3287 2.2563 2.2903 1.8314 1.9128 60 2.4264 2.4217 2.3208 1.8314 1.9535 D.F. 2.2414 2.4474 2.2353 1.9391 1.9259

Striking score and Dominance factor for SET 2 (Spinners)

bk IMRAN TAHIR SP NARINE S HASAN R ASHWIN R JADEJA 10 2.0797 1.5765 1.9166 1.9333 1.5664 20 1.6304 2.1360 1.6407 1.8133 1.8143 30 2.0000 2.4381 1.7476 1.7511 1.8329 40 1.8854 1.9315 1.8311 1.7357 1.7224 50 2.1413 2.2460 1.9814 1.9216 1.9476 60 2.1592 2.3001 1.9773 1.9944 1.9575 D.F. 2.0259 2.1867 1.8826 1.8729 1.8603

Striking score and Dominance factor for SET 1 (Seamers)

bk MA STARC TA BOULT DW STEYN M MORKEL TG SOUTHEE 10 5.2941 2.7820 2.1891 2.1645 2.2133 20 2.1372 2.0256 1.8204 1.9182 2.2315 30 2.0364 1.9790 2.1936 2.1603 2.1789 40 2.0655 1.8547 2.2477 2.3751 2.2666 50 2.2602 2.1025 1.8526 2.0977 2.1488 60 2.4985 2.0962 2.0528 2.0487 2.2634 D.F. 2.3920 2.0609 2.0467 2.1316 2.2192

Striking score and Dominance factor for SET 2 (Seamers)

bk MA STARC TA BOULT DW STEYN M MORKEL TG SOUTHEE 10 2.0866 1.4761 1.5069 1.8085 2.8148 20 1.9700 1.8730 1.6984 1.6423 1.5711 30 1.9000 2.2380 1.8214 1.8300 1.7222 40 1.8723 2.2364 1.9484 1.7895 1.8674 50 1.8064 2.1868 1.9579 1.9604 1.9675 60 1.9677 1.6185 2.1479 2.1120 2.0158 D.F. 1.9073 1.9775 1.9447 1.9150 1.9298

• Higher the Dominance factor higher will be the

chance to take wicket for a bowler, given the number

• f balls (here, bk ) .
• SP Narine has highest Dominance factor among the

considered Slow-arm spinners .

• MA Starc has the highest Dominance factor among

the Fast-arm seamers .

Conclusion

• Within the limits of this study , the paper seeks to

highlight the tremendous scope that exists to improve and develop on the measures currently used to describe the performances of cricket players in general and bowlers in particular.

• The attempt of the paper is not to arrive at a model to

rank the utility of the players, it is just an approach to have an efficient bowling strategy for a match.

Future Works

• Similar approach can be adopted for T20 matches

to predict the performances of bowlers.

• This results can be used in the auction of players

for the league matches .

• The results obtained using the same approach for

a certain period of data can be useful to decide a player to bet on and against.

References

• D. Attanayake and G. Hunter Probabilistic Modelling of Twenty-

Twenty (T20) Cricket : An Investigation into various Metrics of Player Performance and their Effects on the Resulting Match and Player Scores, Proceedings of 4th International Conference

• n Mathematics in Sport 2015.
• U. Damodaran. Stochastic dominance and analysis of ODI

batting performance : The Indian cricket team 1989-2005. Journal of Sports Science & Medicine, 5:503-508,2006.

• stats.espncricinfo.com