Cribbage's Standardised Test Averages (UPDATED November 2018 - posts 753-755)

[Burgey]

Looks like you've put a lot of work into this Cribbage.

 

[NUFAN]

There's actually a third factor in there - "impact" - which is basically there to stop some **** batsman taking 10 jammy Test wickets @ 18 in a 15 year career and being rated as the best bowler of all time. It also inadvertently helped out spinners (or any bowlers who fit the bill, really) who were able to bowl a large percentage of their team's overs and have a really big impact on the overall quality of their team's attack, like Kumble, which I really liked. I wanted to ensure the batting value numbers were similar and comparable to the bowling value numbers so I did the same thing for the batting values although it didn't change much other than slightly penalise those who got stacks of not outs (again not that bad a result).

The whole thing works around the number 31.77 (which was the global all-time batting average when I first starting doing this IIRC) - so batting values as worked out as:

{[(StdAvg / 31.77) ^ 2] * Longevity * Impact} ^ {2/3}

Impact IIRC is something like standaridsedruns/innings/317 - how much you contribute to your team with the bat towards getting a par score on average.

The bowling one was basically along the same lines:

{[(31.77 / StdAvg) ^ 2] * Longevity * Impact} ^ {2/3}

Bowling impact, again without actually checking, was wickets/matches/15, IIRC.

Hey Prince, I love the work that has been put into this, I just have a question so I understand it better.

Why is the batting and bowling averages exactly the same - 31.77? With no balls and wides, shouldn't the bowling formula be a little higher? I know it will probably make a very small difference, but perhaps the all rounders will valued slightly less..

 

[Prince EWS]

Hey Prince, I love the work that has been put into this, I just have a question so I understand it better.

Why is the batting and bowling averages exactly the same - 31.77? With no balls and wides, shouldn't the bowling formula be a little higher? I know it will probably make a very small difference, but perhaps the all rounders will valued slightly less..

It actually doesn't matter what number you choose; it standardises it all to that value for a matter of reference - to make it easier for people to understand. If I standardised it all to 155 then the player values etc would all still be the same.

 

[NUFAN]

It actually doesn't matter what number you choose; it standardises it all to that value for a matter of reference - to make it easier for people to understand. If I standardised it all to 155 then the player values etc would all still be the same.

Ah okay, cheers for the explanation.

 

[Prince EWS]

Ah okay, cheers for the explanation.

Telling you that Murali's standardised average is 23.94 certainly holds more meaning to a cricket fan than telling you he's 1.327 times better than the mean, or that his scorecard average is 1.06 times too small.

 

[NUFAN]

Telling you that Murali's standardised average is 23.94 certainly holds more meaning to a cricket fan than telling you he's 1.327 times better than the mean, or that his scorecard average is 1.06 times too small.

Yeah true, I'm into stats though so I would probably prefer to read how much above or below the mean he is.

Just out of interest, are there many (or any) names that you feel are way out of place, just by watching their performances or your formula not being able to calculate the importance of some knocks/spells?

 

[Prince EWS]

Yeah true, I'm into stats though so I would probably prefer to read how much above or below the mean he is.

Just out of interest, are there many (or any) names that you feel are way out of place, just by watching their performances or your formula not being able to calculate the importance of some knocks/spells?

Vettori is one that sticks out for me as someone this system vastly over-rates. He was a decent Test bowler (and a rubbish tailender) for the first half of his career and a decent Test batsman (and an ever-declining holding bowler who eventually tapered off into near insignificance as a wicket taking option) for the second half, but the two didn't really overlap and he was never a special player in either discipline. Yet my system "thinks" he's a better cricketer overall than Rhodes, Sutcliffe, Lillee, Donald, Simpson etc which is really just flat out wrong. I don't think it accurately adjusts his minnow-bashing bowling, and it's given me cause to consider changing standardised wickets taken rather than runs conceded for bowlers (eg. if you play a team twice as bad as the average team it doubles your runs conceded for that game, but I've been thinking perhaps it should halve your wickets taken instead).

 

[NUFAN]

Vettori is one that sticks out for me as someone this system vastly over-rates. He was a decent Test bowler (and a rubbish tailender) for the first half of his career and a decent Test batsman (and an ever-declining holding bowler who eventually tapered off into near insignificance as a wicket taking option) for the second half, but the two didn't really overlap and he was never a special player in either discipline. Yet my system "thinks" he's a better cricketer overall than Rhodes, Sutcliffe, Lillee, Donald, Simpson etc which is really just flat out wrong. I don't think it accurately adjusts his minnow-bashing bowling, and it's given me cause to consider changing standardised wickets taken rather than runs conceded for bowlers (eg. if you play a team twice as bad as the average team it doubles your runs conceded for that game, but I've been thinking perhaps it should halve your wickets taken instead).

Hmm, that is interesting - it's bloody hard to get a system where you completely agree with it - it is hard to imagine how Vettori could be ahead of any of the players you've mentioned.

So if Vettori takes 6/100 against Bangladesh. would 6/200 really make much of a difference to 3/100 as per your possible suggestion?

 

[Prince EWS]

I think Gilchrist is significantly better than the 119th best batsman of all time as well. He fell victim a bit to me trying to keep batting and bowling comparable with the "impact" variable (which only really exists to stop part-time bowlers ranking really higher on the bowling lists if they play long careers with small averages, but hammered Gilchrist for low runs per match due to not having bat as often as his competitors) and relatively short longevity, but as a batsman he was huge for Australia. It's got him lower than Alec Stewart, which again is something I don't think anyone will agree with. That's more a problem with the longevity and value ratings than the standardising of averages though; his standardised average of 44 or so does accurately reflect his career IMO.

 

[Prince EWS]

Hmm, that is interesting - it's bloody hard to get a system where you completely agree with it - it is hard to imagine how Vettori could be ahead of any of the players you've mentioned.

So if Vettori takes 6/100 against Bangladesh. would 6/200 really make much of a difference to 3/100 as per your possible suggestion?

Well one game obviously wouldn't but Vettori's career is littered with such events, so changing the wickets would change how much each game counted for. In that example it'd count for half of what it did before, so it'd be like he did that once instead of twice. His failings against better opposition would be taking up more of a pie, so to speak.

 

[NUFAN]

I think Gilchrist is significantly better than the 119th best batsman of all time as well. He fell victim a bit to me trying to keep batting and bowling comparable with the "impact" variable (which only really exists to stop part-time bowlers ranking really higher on the bowling lists if they play long careers with small averages, but hammered Gilchrist for low runs per match due to not having bat as often as his competitors) and relatively short longevity, but as a batsman he was huge for Australia. It's got him lower than Alec Stewart, which again is something I don't think anyone will agree with. That's more a problem with the longevity and value ratings than the standardising of averages though; his standardised average of 44 or so does accurately reflect his career IMO.

So for GIlly and someone like Trescothick, longevity is actually career span not Tests played as both Cricketers played a stack of matches in a short time period and the fact they were never injured was very important to both Australia and England respectively.

Gilchrist is always going to be one of those Cricketers who statistically won't appear quite as good as how he actually played. We've all seen him perform under pressure, bat aggressively to change the momentum of a game and suffer a form slump at the fag end of his career. Definitely the ultimate team player.

Well one game obviously wouldn't but Vettori's career is littered with such events, so changing the wickets would change how much each game counted for. In that example it'd count for half of what it did before, so it'd be like he did that once instead of twice. His failings against better opposition would be taking up more of a pie, so to speak.

Cool, interesting once again. I feel like I'm conducting a Dilmah tea party interview with you mate. More tea?

 

[Prince EWS]

Well one game obviously wouldn't but Vettori's career is littered with such events, so changing the wickets would change how much each game counted for. In that example it'd count for half of what it did before, so it'd be like he did that once instead of twice. His failings against better opposition would be taking up more of a pie, so to speak.

I actually just tried doing this (was easier than I thought; robust code FTW) and it actually made **** all difference.

Standardising Runs Conceded - Vettori
Standardised average: 34.28
All-time bowling rank: 51st

Standardising Wickets Taken - Vettori
Standardised average: 34.64
All-time bowling rank: 52nd

It did *something* though.

 

[Prince EWS]

Looking at it, Vettori being a slightly below average bowler across his career and slightly above average batsman (including tailenders) seems about right; it's the way it incorporates longevity which has him (IMO anyway) too high, so it's like the reverse of the Gilchrist problem.

 

[weldone]

Well, my formula has Vettori at number 64 among bowlers and at number 237 among batsmen. Do you think my formula rates him more correctly than yours, PEWS?

 

[Spark]

I really struggle to have Vettori anywhere near the best 100 batsmen of all time tbh, let alone 52nd. Something's ****ed up there.

 

[Prince EWS]

I really struggle to have Vettori anywhere near the best 100 batsmen of all time tbh, let alone 52nd. Something's ****ed up there.

That's bowling dude.

 

[Spark]

Ah oops. Thought it was one for bowling and batting.

 

[Prince EWS]

Ah oops. Thought it was one for bowling and batting.

Nah two different systems to measure bowling. One which was designed specifically to bring him down, and still failed.

 

[Spark]

You should make a metric to measure troughs, and include that. IMO a bowler or batsman who has an extended period of non-performance should be penalised extra for it.

 

[weldone]

Why is the batting and bowling averages exactly the same - 31.77? With no balls and wides, shouldn't the bowling formula be a little higher?..

Run outs will be a more significant reason for the difference between batting and bowling averages IMO. But as PEWS says, it won't matter in the rankings; it will bump up or down the number by a constant multiplier for all.