DoG Top 100 Test Batsmen : How much importance should be given to strike rate?

[DrWolverine]

Please vote on the following for the next 24 hours:

A: Keep the same weighting (Average x4, Runs per innings x 2, Strike-rate x1)
B: Less emphasis on strike-rate (Average x6, Runs per innings x 3, Strike-rate x1)
C: No strike-rate (Average x2, Runs per innings x1)

 

[capt_Luffy]

This exercise has been a gargantuan effort from DoG's end, who has worked for nearly 2 decades now trying to perfecting the formula. I personally think career SR isn't a reliable metric to measure situational batting, which is the important quality in a batsman rather than just going fast or slow. So if I ever tried something similar, I won't be using it. Hence, my preferred choice is C. Buuuut, it's DoG's formula and if he thinks SR should play a role, then it should imo. So I won't mind at all, if DoG just chooses to ignore the poll results and go with his own metric weightage, because this exercise belongs to him, and him only.

 

[Patience and Accuracy+Gut]

I would go with 2nd option. I think slight consideration is fine. It does help from time to time when a batsman takes off the pressure of the non- striker.

 

[Coronis]

Ah yes I know DoG is very interested in constructive criticism on his list.

However at the end of the day it is his list of course.

Whilst we all have our own strong opinions on this specific stat I’d probably say to DoG to just calculate with the other two weightings compared to the current one and pick which list he likes better lol.

Anyway, you all know my stance.

 

[ataraxia]

Whilst we all have our own strong opinions on this specific stat I’d probably say to DoG to just calculate with the other two weightings compared to the current one and pick which list he likes better lol.

Ew. Tailoring a statistical ranking to one's subjective opinions defeats its point (and removes the fun).

 

[OverratedSanity]

Ew. Tailoring a statistical ranking to one's subjective opinions defeats its point (and removes the fun).

That's like every statistical analysis ever my guy.

 

[centurymaker]

That's like every statistical analysis ever my guy.

Awta

 

[ataraxia]

That's like every statistical analysis ever my guy.

No. If you do an objective ranking then alter it based on your own player-by-player beliefs, it becomes fraudulent. Unless you think your own opinions more accurate to the statistics than the analysis. The same applies if you do two analyses and check which one you like the most.

In my opinion, the point of a ranking isn't "here's what parameters most accurately reflect CW consensus", it's "here's a list based on this set of parameters which most people like, look how X is underrated and Y overrated – my, maybe some people are biased on this forum".

 

[capt_Luffy]

No. If you do an objective ranking then alter it based on your own player-by-player beliefs, it becomes fraudulent. Unless you think your own opinions more accurate to the statistics than the analysis. The same applies if you do two analyses and check which one you like the most.

In my opinion, the point of a ranking isn't "here's what parameters most accurately reflect CW consensus", it's "here's a list based on this set of parameters which most people like, look how X is underrated and Y overrated – my, maybe some people are biased on this forum".

The thing is simply, there's no hardset importance to put on the parameters, and the amount you do is bound to be reflective of your personal taste. Or you will end up like Anatha Narayan

 

[ankitj]

That's like every statistical analysis ever my guy.

No. I don't manipulate my analysis.

 

[ataraxia]

The thing is simply, there's no hardset importance to put on the parameters, and the amount you do is bound to be reflective of your personal taste. Or you will end up like Anatha Narayan

There are better ways to find errors with your methods out than going "oops, Barrington's higher than Viv" – which is obviously going to cloud one's judgment, given how most people think test stats severely overrate Barrington and underrate Viv – e.g. coming up with fake players and evaluating "is someone who averaged 50 over 18 years better than someone who averaged 55 over 10?"

 

[capt_Luffy]

There are better ways to find errors with your methods out than going "oops, Barrington's higher than Viv", e.g. coming up with fake players and evaluating "is someone who averaged 50 over 18 years better than someone who averaged 55 over 10?"

Individual players like them, I can see. But if Voges is beating Hill, you clearly did something wrong.

 

[ataraxia]

Individual players like them, I can see. But if Voges is beating Hill, you clearly did something wrong.

That's a case where you're unearthing big methodological problems (your longevity adjustment doesn't work in edge-cases, and/or you forgot to standardise for era). It's completely different to carefully deciding which list looks better.

 

[capt_Luffy]

That's a case where you're unearthing big methodological problems (your longevity adjustment doesn't work in edge-cases, and/or you forgot to standardise for era). It's completely different to carefully deciding which list looks better.

There are no hard lines there imo.

 

[Ali TT]

No. I don't manipulate my analysis.

You presumably make decisions on what you are going to use in your analysis though

 

[Johan]

Z E R O
E
R
O

 

[ankitj]

You presumably make decisions on what you are going to use in your analysis though

Yes. And then don't change the formula to get a ranking I like.

 

[OverratedSanity]

No. If you do an objective ranking then alter it based on your own player-by-player beliefs, it becomes fraudulent. Unless you think your own opinions more accurate to the statistics than the analysis. The same applies if you do two analyses and check which one you like the most.

In my opinion, the point of a ranking isn't "here's what parameters most accurately reflect CW consensus", it's "here's a list based on this set of parameters which most people like, look how X is underrated and Y overrated – my, maybe some people are biased on this forum".

Oh if that's what you mean by subjective opinions, that's fine. I thought you meant one's subjective opinion about which parameters are important and which aren't when judging the greatness of a player.

However, I do think people subconsciously do somewhat rate their favourites higher partly due to those same reasons. For example, the reason I rate VVS Laxman is probably because I highly value the ability to play innings at important/clutch times. If I make an attempt to make a statistical formula for batsmen, it's likely I'll try to include that as a parameter somehow (if possible). There's still some huge personal subjectivity in these exercises. It's completely unavoidable. Note that I don't mean that means these exercises from @Days of Grace are useless, much the opposite. I think the individual parameter ratings for batsmen may probably even more useful and revealing than just the overall ranking.

 

[RossTaylorsBox]

Individual players like them, I can see. But if Voges is beating Hill, you clearly did something wrong.

Not really, if I make a model with certain parameters and that's the outcome, then I can say that for this particular model certain players outrank others. There's no right or wrong answer because there's no objective truth.

 

[capt_Luffy]

Not really, if I make a model with certain parameters and that's the outcome, then I can say that for this particular model certain players outrank others. There's no right or wrong answer because there's no objective truth.

Yes there aren't. But you have to make some metric to judge a model's performance. There are bound to be some trial and error there. Like, when we make a ML model, it performs. But if can't measure at all how well the model is doing, it's not very useful.