A Simplistic Analysis of ODI Batsmen

[stephen]

Average, Strike rate and runs per innings are individually poor measures of a batsman's value to a team in ODIs. So I decided to sum up the values and use that combined figure to rate ODI batsmen.

Top ten batsmen (min 1500 runs):

Doeschate 203
Kohli 202
De Villiers 199
Bairstow 193
Buttler 188
Amla 185
Dhawan 183
De Kock 183
Maxwell 182
Warner 182

Viv came in 14th and was the second player on the list behind RTD that was not a modern player (had played in 2018).

This seemed to favour strike rates too much. So I divided the strike rate by two (for most batsmen this should be in the same ball park as average+rpi).

This gives a new list:

Doeschate 159
Kohli 156
De Villiers 148
Amla 141
Bairstow 141
Root 138
Azam 138
Dhawan 136
De Kock 135
Warner 133

Proving once and for all that the administrators have entirely ruined ODI stats by introducing the rule changes since the last world cup.

Top players not to have played in 2018 according to the list:

Rtd 159
Trott 133
Abbas 133
Richards 132
Tendulkar 129
Hussey 126
Bevan 126
Hayden 123
Sehwag 121
Turner 121

20 out of the top 30 on this list have played at least 1 match in 2018.

 

[TheJediBrah]

I don't know why you'd use Runs Per Innings as a measure when you've already got average. That doesn't really add anything to rating batsmen, it just gives more favour to top order batsmen.

 

[stephen]

Average+RPI top ten:

Rtd 115
Kohli 109
De Villiers 97
Amla 96
Azam 95
Trott 95
Root 95
Abbas 90
Williamson 89
Dhawan 89

Non 2018 top 10:

Rtd 115
Trott 95
Abbas 90
Bevan 89
Richards 87
Turner 87
Tendulkar 86
Greenidge 85
Taylor 85

I tried RPI + SR/2

Kohli 96
De Villiers 95
Bairstow 93
Rtd 92
Amla 91
Dhawan 90
De Kock 90
Warner 90
Buttler 90
Maxwell 90

Pre 2018:

Rtd 92
Sehwag 86
Richards 85
Abbas 85
Tendulkar 83
Gilchrist 83
Trott 82
Afridi 80
Watson 79
Gambhir 79

Now we're getting more interesting. I tried SR/3+RPI.

Kohli 80
De Villiers 77
Rtd 77
Amla 76
Bairstow 75
Dhawan 75
De Kock 74
Warner 74
Root 72
Azam 72

Pre 2018:

Rtd 77
Abbas 71
Richards 70
Tendulkar 70
Trott 69
Sehwag 69
Gilchrist 67
Hayden 66
Gambhir 65
Pieterson 64

 

[stephen]

I don't know why you'd use Runs Per Innings as a measure when you've already got average. That doesn't really add anything to rating batsmen, it just gives more favour to top order batsmen.

RPI shows more how valuable to an average team total a batsman is.

And like I said, it's all quite simplistic.

 

[ankitj]

You could just reuse this methodology: Stats from the Past: The best ODI batsmen from across eras | Cricket | ESPNcricinfo.

As good an approach as you can use to combine averages and SRs without complicating analysis beyond comprehension.

 

[TheJediBrah]

RPI shows more how valuable to an average team total a batsman is.

I don't think it does tbh. Might even do the opposite as your using it diminishes the influence of average which is a better indicator when paired with strike rate.

Finishing off innings, both batting first and second, can be more important to how valuable a batsman is than someone who bats at the top every game and gets heaps of runs per innings as a result.

 

[TheJediBrah]

Basically just saying that Avge + SR is a better indicator than Avge + SR + RPI. Adding RPI just rewards players who bat up the order, and don't finish off winning games for their team, and punishes those who are often there at the end.

 

[stephen]

You could just reuse this methodology: Stats from the Past: The best ODI batsmen from across eras | Cricket | ESPNcricinfo.

As good an approach as you can use to combine averages and SRs without complicating analysis beyond comprehension.

Yeah I was just looking at some simple formulas for fun.

 

[stephen]

Basically just saying that Avge + SR is a better indicator than Avge + SR + RPI. Adding RPI just rewards players who bat up the order, and don't finish off winning games for their team, and punishes those who are often there at the end.

I'm not really debating that. I think both factors are important. High RPI is like forwards in rugby league - gets you most of the way. High averages are like the backs - they're flashy and get the ball over the line.

Ultimately the RPI is a very good measure of how much a batsman contributes to a total while average tells you more about how often they're there at the end.

 

[srbhkshk]

RPI favors top order batsman, average doesn't favor any position(once you factor in the strike rate). It might be a fine metric in some cases, but using it to compare people who bat at different positions isn't a good way.

 

[TheJediBrah]

I'm not really debating that. I think both factors are important. High RPI is like forwards in rugby league - gets you most of the way. High averages are like the backs - they're flashy and get the ball over the line.

Ultimately the RPI is a very good measure of how much a batsman contributes to a total while average tells you more about how often they're there at the end.

Yeah, no, that's what I disagree with. It's not a good measure at all, and adding it in makes your rankings worse.

 

[Migara]

Use the standardized average * standardized SR statistic.

standardized average = (players average / average of batsmen played during players career) * global ODI average
standardized strike rate = (players strike rate / strike rate of batsmen played during players career) * global ODI strike rate

SAvg = (Avg / CAvg) * GAvg
SSR = (SR / CSR) * GSR

ln (SAvg * SSR) has a gaussian distribution, so very close to a normal population.

However these are raw numbers, and doesn't take length of the career in to account. However it will correct the effects brought on by rule changes.

 

[Pap Finn Keighl]

There's another simple method.
Ask me.

Sachin
Viv
Abdv
Kohli
Dhoni
Lara
Ponting
Beven
Abbas
Jones