DoG's Top 100 Test Batsmen: Discussion thread

[OverratedSanity]

IMO putting RPI is a terrible idea

I think it's a decent idea. It's not like he's only considering RPI, he's taking both RPI and average.

 

[TheJediBrah]

I think it's a decent idea. It's not like he's only considering RPI, he's taking both RPI and average.

How does adding RPI do anything beneficial though when you've already got average? It just punishes players who get the job done and end up not out. It makes even less sense to use RPI in Tests than it does in limited-overs.

I'm all for shoving in random measurements to shake things up but I really don't see any logic behind this one.

 

[stephen]

How does adding RPI do anything beneficial though when you've already got average? It just punishes players who get the job done and end up not out. It makes even less sense to use RPI in Tests than it does in limited-overs.

I'm all for shoving in random measurements to shake things up but I really don't see any logic behind this one.

Nah it's the opposite. It makes more sense in tests because most of the time you're trying to score as many as possible and not just as fast as possible.

 

[Victor Ian]

Just as a matter of curiosity, id love to see the distribution of not outs for each batting position. With all batsmen being equal id expect it to increase with batting order. But things arent equal so my guess is it peaks somewhere around 6.

 

[TheJediBrah]

Nah it's the opposite. It makes more sense in tests because most of the time you're trying to score as many as possible and not just as fast as possible.

You're rating how good batsmen are. How does treating an innings where a batsman is not out the same as if they were out help rating how good batsmen are? Makes absolutely no sense, it does the opposite of that. Just skewing it toward batsmen that bat higher in the order.

 

[Days of Grace]

It works well in small samples. For example, Steve Waugh averaged 77 across 50 innings but actually scored less than 60 runs an innings.

 

[SillyCowCorner1]

I always love a good innings of one carrying his bat. Stuff of legends. Especially in a 40 all out.

 

[Days of Grace]

IMO putting RPI is a terrible idea

Then Chanderpaul is a greater test batsmen than Richards. Are you ok with this?

 

[Malcolm]

Who has the second-highest average across 50 innings?

 

[TheJediBrah]

It works well in small samples. For example, Steve Waugh averaged 77 across 50 innings but actually scored less than 60 runs an innings.

How does that work well at all? If Steve Waugh averaged 77 with 60 RPI then he batted better (assuming all other factors equal) than someone who averaged 65 with 60 RPI. I think you're example shows the opposite of it "working well".

The only thing adding RPI to the formula will do is make it less accurate at rating which batsmen are better.

Then Chanderpaul is a greater test batsmen than Richards. Are you ok with this?

Why wouldn't I be?

 

[TheJediBrah]

I'll throw you an analogy. Say you want to measure how fast a bowler is bowling, and you have a (hypothetical) new very accurate speed gun that is usually right that you've been using. Then someone comes along and suggests that you should use this old, less accurate speed gun that is wrong more often as well and average out the 2 speeds. You might get some slightly different numbers but they're going to be less accurate readings.

That's exactly what you're doing. RPI is just a less accurate way of measuring average.

 

[zaremba]

How does adding RPI do anything beneficial though when you've already got average? It just punishes players who get the job done and end up not out. It makes even less sense to use RPI in Tests than it does in limited-overs.

I'm all for shoving in random measurements to shake things up but I really don't see any logic behind this one.

Completely agree with this. RPI adds nothing, and just rewards those higher in the order.

A high proportion of not outs in some ways signifies a better player. A player who scores 35 (out) and 35 not out has had to start twice, whereas the player who gets 70 (out) has had to start once, then cashed in with some easier runs with his eye in.

The only thing RPI might usefully reward is the ability to concentrate over long periods, but that’s better recognised by a different measure such as (say) number of centuries.

 

[OverratedSanity]

Completely agree with this. RPI adds nothing, and just rewards those higher in the order.

This is a good thing.

 

[Days of Grace]

RPI awards batsmen who made contributions to their team’s total run tally. It is especially important in rating batsmen in ODIs.

I’ve decided to go with Average*4 RPI *2 Strike-rate *1

 

[TheJediBrah]

RPI awards batsmen who made contributions to their team’s total run tally. It is especially important in rating batsmen in ODIs.

I’ve decided to go with Average*4 RPI *2 Strike-rate *1

Nothing to do with how good they are at batting though, but fair enough

 

[Singh767]

There is no way you can do a statistical measurement of test batsman and not count 100s anywhere - and it is nothing like a 5fer

 

[Prince EWS]

Having done this sort of thing before, I will say that RPI is one of those measures that people don't like but tends to create rankings that conform to their instincts more. I think it's either something that we subconsciously value, or something that correlates with something we subconsciously value, even when we consciously attempt to reject it.

 

[srbhkshk]

RPI is a clearly flawed metric - put two absolutely equal batsman at #4 and #5 and #4 will have a better RPI.

 

[Prince EWS]

RPI is a clearly flawed metric - put two absolutely equal batsman at #4 and #5 and #4 will have a better RPI.

While this is true, and while I don't support including RPI as part of the analysis mix, I think in most cases #5 would end up with a better average and strike rate (although this is obviously a less uniform fact). If the other factors are going to be average and strike rate, maybe this proxy counterbalance isn't the worst thing in the world.

 

[mr_mister]

There is no way you can do a statistical measurement of test batsman and not count 100s anywhere - and it is nothing like a 5fer

Arguably Clem Hill's most important contribution in tests were consecutive matchwinning innings of 99, 97 and 98 in the 1902 ashes


My point being 100 is an arbitrary number for how good an innings is