Batting Average vs Batting RPI

[Majestic]

Some examples where looking at RPI is more important than batting averages.

Lara vs Kallis
Stokes vs Jadeja
Botham vs Imran
Kapil vs Pollock

In all these cases, the former is a better batsman than latter.

 

[Prince EWS]

Red ink innings should be treated like Grandi's series and counted as 0.5 of a dismissal.

I'm actually not even joking. I may play with this method if I ever do an update to the standardised averages formula.

Failing that, averages are clearly better. RPI is easier to calculate and conceptualise as a a adage, these a reason we've gone out our way to do it the way we do despite the complications for hundreds of years.

Jadeja is someone who comes up a lot when discussing this issue.

Batting avarage: 37.72
Runs per innings: 30.36
Grandi's sequence method: 33.64

It's always going to be halfway between the two, with some funky Ramanujan summation nerdery to justify it.

Like the sequence 1-1+1-1+1 etc... the answer to whether it's an innings is binary but it "flickers" between 0 and 1 when it's a not out. The assigned value of this infinite series is 0.5.

 

[sayon basak]

It's always going to be halfway between the two, with some funky Ramanujan summation nerdery to justify it.

Like the sequence 1-1+1-1+1 etc... the answer to whether it's an innings is binary but it "flickers" between 0 and 1 when it's a not out. The assigned value of this infinite series is 0.5.

If in an exam, you are asked to evaluate the sum of 1-1+1-1+..... or 1-2+3-4+5-...... and you answer 1/2 or 1/4 instead of stating that these are actually divergent sums, you'll get exactly 0 marks. Without special contexts (like zeta function and Co.), these are meaningless.

 

[Prince EWS]

If in an exam, you are asked to evaluate the sum of 1-1+1-1+..... or 1-2+3-4+5-...... and you answer 1/2 or 1/4 instead of stating that these are actually divergent sums, you'll get exactly 0 marks. Without special contexts (like zeta function and Co.), these are meaningless.

That's why I said assigned values.

I'm gilding the lily a bit here but it's a fun way to justify why I'm just picking a value halfway between two things after too many beers watching Thursday night footy.

 

[Prince EWS]

Let's look at it another way.

Traditional averages treat an incomplete innings as zero innings.

RPI measures treat an incomplete innings as one innings.

In reality both are these are never true - the innings has always started but never finished. We don't know exactly where in between 0 and 1 this number should lie, but it's definitely greater than zero (after the start) and less than 1 (before the end).

Half seems like a good number to pick, unless some analysis could show incomplete innings are on average likely actually closer to the start or the end of an innings.

 

[sayon basak]

Let's look at it another way.

Traditional averages treat an incomplete innings as zero innings.

RPI measures treat an incomplete innings as one innings.

In reality both are these are never true - the innings has always started but never finished. We don't know exactly where in between 0 and 1 this number should lie, but it's definitely greater than zero (after the start) and less than 1 (before the end).

Half seems like a good number to pick, unless some analysis could show incomplete innings are on average likely actually closer to the start than the end of an innings.

Introduce some metric that can deal with "exactly how incomplete the not out innings were". Let's say, one batter comes to bat half an hour before declaration and he scores 18*, and another batsman comes to the crease and the team declares right after an over and he scores 1*. Giving both of them 0.5 weightage seems too simplistic imo. The former should count for a higher value than the later one.

 

[Prince EWS]

Introduce some metric that can deal with "exactly how incomplete the not out innings were". Let's say, one batter comes to bat half an hour before declaration and he scores 18*, and another batsman comes to the crease and the team declares right after an over and he scores 1*. Giving both of them 0.5 weightage seems to simplistic imo. The former should count for a higher value than the later one.

If the 18* is Bradman and the 1* is Martin then the former is probably less complete.

At the moment averages always assign the value of 0 and RPI always assigns the value of 1. These are also simplistic and not representative of these nuances, but IMO worse values to assign than 0.5.

I'm sure it could be improved further, but I think assigning a value of 0.5 is already some improvement on assigning a value of 0 or 1. Probably fits with the "averages sometimes but RPI sometimes" instinctive replies of some people in the thread too.

 

[Coronis]

Red ink innings should be treated like Grandi's series and counted as 0.5 of a dismissal.

I'm actually not even joking. I may play with this method if I ever do an update to the standardised averages formula.

Failing that, averages are clearly better. RPI is easier to calculate and conceptualise as an average, these a reason we've gone out our way to do it the way we do despite the complications for hundreds of years.

Why do you have to play with us like this?

What about only counting those not outs as 1 which happen in the final inns of a game (inns to draw, chase down totals etc), but considering the rest of the not outs as 0

So if a player’s innings is interrupted by a declaration that doesn’t count?

 

[Prince EWS]

Why do you have to play with us like this?





So if a player’s innings is interrupted by a declaration that doesn’t count?

I actually have three weeks off over Christmas this year. But I'll probably just have a three week nap.

 

[Thala_0710]

So if a player’s innings is interrupted by a declaration that doesn’t count?

Yeah, the point being that the player (in general), doesn't serve the team any purpose by staying not out here, might as well try and maximize his runs in whatever deliveries he's getting.
Staying not out till the end to draw/win a game in the final inns, would generally serve a purpose for the team.

 

[capt_Luffy]

Yeah, the point being that the player (in general), doesn't serve the team any purpose by staying not out here, might as well try and maximize his runs in whatever deliveries he's getting.
Staying not out till the end to draw/win a game in the final inns, would generally serve a purpose for the team.

What about a loss? Like Jadeja in the 4th Test?

 

[Thala_0710]

What about a loss? Like Jadeja in the 4th Test?

Ummm not sure about that one

 

[shortpitched713]

Introduce some metric that can deal with "exactly how incomplete the not out innings were".

Crystal Ball for mine

 

[Majestic]

What about a loss? Like Jadeja in the 4th Test?

Count his runs but staying not out was pointless there.

 

[capt_Luffy]

Count his runs but staying not out was pointless there.

A dismissal means a batsman was beaten, a NO means he wasn't. A pretty straightforward and core concept I would argue

 

[Xix2565]

What about a loss? Like Jadeja in the 4th Test?

Count his runs but staying not out was pointless there.

India drew the 4th Test because he stayed not out ffs. They lost the 3rd Test because of a lucky roll. Get the Tests right.

 

[ataraxia]

Let's look at it another way.

Traditional averages treat an incomplete innings as zero innings.

RPI measures treat an incomplete innings as one innings.

In reality both are these are never true - the innings has always started but never finished. We don't know exactly where in between 0 and 1 this number should lie, but it's definitely greater than zero (after the start) and less than 1 (before the end).

Half seems like a good number to pick, unless some analysis could show incomplete innings are on average likely actually closer to the start or the end of an innings.

Tendulkar had an RPI of 85 in not-out innings, so his expected runs on average from that point is significantly higher than his average (compare starting your innings at 0 to continuing your innings at 85). I assume that's similar for many proper batsmen.

He's more than halfway through his innings, but the remainder of his innings will on average produce more runs than an average innings. Meaning that average and an in-between value between average and RPI both ascribe a significantly lower value to his not outs than their intended purpose. If you're going to go down that route, there's that to consider.

 

[Majestic]

A dismissal means a batsman was beaten, a NO means he wasn't. A pretty straightforward and core concept I would argue

If he was the only batter left and his team gets bowled out in a losing cause, it means he was beaten too.

Everything a player does is linked to the course of the match situation. They can’t say that because the opposition didn’t got them out, they didn’t lose. If the team lost, it means every single player that played also lost.

 

[Majestic]

India drew the 4th Test because he stayed not out ffs. They lost the 3rd Test because of a lucky roll. Get the Tests right.

Yeah it was 3rd Test at Lords.

 

[capt_Luffy]

If he was the only batter left and his team gets bowled out in a losing cause, it means he was beaten too.

Everything a player does is linked to the course of the match situation. They can’t say that because the opposition didn’t got them out, they didn’t lose. If the team lost, it means every single player that played also lost.

Ah yes, he was beaten because he didn't single handedly chase down 200+ runs target from one End!!!!