What are the most important statistical measures for test bowlers?

[Days of Grace]

What weighting, though?

 

[cnerd123]

I wouldn't just weigh those two but if I had to, maybe 5:2?

 

[viriya]

Been rethinking my test bowling ratings recently. I can't decide whether to rate bowlers on a combination of bowling average/strike rate or strike rate/economy rate and what ratio to give each measure.

At the moment I am leaning towards a combination of strike-rate/economy rate with a weighting of 2:1 in favor of strike-rate. Conversely, I am thinking of it being 2:1 in favor of economy rate for ODI bowlers. Or would runs per wicket and strike-rate with a 2:1 weighting in favor of runs per wicket be better for test bowlers, since the main things we use to judge a bowler when looking at his stats are average and strike-rate?

Have I got the combination and ratio correct? I would appreciate any feedback.

In addition, how much more important is batting average than strike-rate in tests? At the moment I have it 6:1 in favor of average, but maybe I could give more weight to strike-rate?

In my setup for Tests just comparing the runs and SR factors it's more like 10:1. 6:1 is too generous imo. Especially since era plays a big part in the SR a batsman has.

For Test bowling the three major factors I use are wickets per run (average measure), wickets per ball (strike rate measure), runs per ball (econ measure). Of those Wkts/Run is given most importance then strike rate then econ. I can't check right now but SR is almost twice as important as econ as I recall. Average is more important than SR but not twice as much.

All this is ignoring about 10 other less important factors of course. In some cases (match status) those factors can be as important as these main factors.

 

[indiaholic]

Do you guys face problems with the variables you use being strongly correlated with each other? Where even though it seems like you are measuring different attributes, you effectively end up just giving more weight to a single attribute?

 

[viriya]

Do you guys face problems with the variables you use being strongly correlated with each other? Where even though it seems like you are measuring different attributes, you effectively end up just giving more weight to a single attribute?

tbh I think there hasn't been enough analysis on stuff like this. For example instead of a factor model I would really prefer to just come up with different variables and use ML methods to determine which variables matter most etc. The problem is it's unclear what the target variable should be. It can't just be the match result (win/draw/loss) because when you rate innings it's not just about that - there's a certain "je ne sais quoi" you're trying to capture especially when rating batting innings since how great an innings is mostly about how much people remember it.

Definitely something I have on my to do list to explore.

 

[Bijed]

Assuming equal averages, I'd take the bowler with a better SR than one with a better ER in general, though in a good, balanced test attack you should have someone with a good ER too, since you often see batsmen getting out to bowler x as a result of the pressure bowler y is building from the other end, not to mention there are times when you're going to want to restrict runs.

I guess if you're a bad team who only has access to bad bowlers and whose best hope is generally to draw a test, you'd probably want more economical bowlers, even if they all average in the 40s or something. But I guess a decent team looking to rise up the rankings would benefit more from the addition of a Mitchell Johnson than a Peter Siddle (sorry NUFAN).

And so ends another post where I'm not really sure if I've made point.

 

[Red]

Wickets per innings and SR.

 

[fredfertang]

Look I've had eight vodkas tonight. I can't do stats right now

Old school as ever - none of this six vodka nonsense - presumably you're on an uncovered wicket as well?

 

[wellAlbidarned]

SR is overrated. Good to watch but assuming you don't have a h4x lineup with 2-3 ATGs it's so much more important to be able to build pressure to support lesser bowlers in your side. Average and economy in 90% of cases.

In a very strong bowling side SR becomes more valuable as bowlers will be competing harder for wickets and runs will be hard to come by anyway.

 

[Burgey]

Old school as ever - none of this six vodka nonsense - presumably you're on an uncovered wicket as well?

Finished on ten, well five doubles anyway. Good start to a working week. Not.

 

[TheJediBrah]

Average is clearly by far the most important statistic. As to whether Strike-rate or economy is more important after that is a it more debatable.

Of course it depends on the rest of your bowling attack, game situations, teams requirements etc. so kind of hard to answer definitively

 

[thierry henry]

imo only average really matters as a measure of quality and e/r and s/r just tell you what style of bowler someone is.

In the real world I don't think people actually prefer bowlers who take 2/40 off 10 rather than 2/40 off 15 but for some reason strike rate gets preferential treatment in these types of conversations.

 

[thierry henry]

Also, obviously there tends to be a correlation between the two. If you're a good bowler you're probably going to take wickets AND be hard to score off. It's pretty rare for a bowler to take bags of wickets while simultaneously being easy to score off. Yet an economical bowler with a good average (i.e. a good bowler) will not have the best strike rate. An expensive bowler who also concedes more runs per wicket may nevertheless have the same strike rate but imo will be a clearly inferior bowler.

The more I think about it the more inclined I actually am to prefer e/r because being hard to score off is fundamentally the quality of a good bowler. But ultimately, it's just average that matters really and the rest is horses for courses, style of bowler for the situation etc.

 

[flibbertyjibber]

Size of their dong.

 

[SJS]

That's a top question and like all really good questions it does not have an easy answer - or a simple one. Without adding to the noise or getting into an argument let me say there is one which is very important but not often mentioned. . . .

Wickets per Test played.

I once did an exercise for comparing different statistical measures to rate bowlers. I took an eligibility criteria (say 100 Test wickets . . . it could well be 20 Test matches) and then ranked all the bowlers within that criteria in different lists by different criteria . . . I found myself most impressed with the way the list shaped up from top to bottom by the W/T criteria. Its not perfect but its very telling when comparing great bowlers with decent career lengths.

PS This may have been mentioned before in this thread. If so my apologies but I haven't read the whole thread

 

[stephen]

So how do you rate Davidson compared to say Waqar? Waqar with the way better strike rate but Davidson with the way better economy rate.

I'd have Davidson in my AT Aus XI with Lillee and McGrath. I think economy rate is hyper-important for 3rd seamers.

Basically you need at least 2 out of 3 seamers to have a good economy so that the captain can slow things down when needed. Against the best batsmen the ability to control the flow of the game is vital. Can't let them have easy runs.

 

[stephen]

That's a top question and like all really good questions it does not have an easy answer - or a simple one. Without adding to the noise or getting into an argument let me say there is one which is very important but not often mentioned. . . .

Wickets per Test played.

I once did an exercise for comparing different statistical measures to rate bowlers. I took an eligibility criteria (say 100 Test wickets . . . it could well be 20 Test matches) and then ranked all the bowlers within that criteria in different lists by different criteria . . . I found myself most impressed with the way the list shaped up from top to bottom by the W/T criteria. Its not perfect but its very telling when comparing great bowlers with decent career lengths.

PS This may have been mentioned before in this thread. If so my apologies but I haven't read the whole thread

The problem with this stat is that it punishes bowlers with good support like McGrath or Marshall. And Hadlee ends up as the best bowler.

 

[TheJediBrah]

The problem with this stat is that it punishes bowlers with good support like McGrath or Marshall. And Hadlee ends up as the best bowler.

Exactly what I was about to say. Wickets per Test played depends far too much on the strength of the rest of your bowlers.

By that measure people might mistakenly think Murali was the best spin bowler of the modern era, ridiculous as it may sound.

Also, obviously there tends to be a correlation between the two. If you're a good bowler you're probably going to take wickets AND be hard to score off. It's pretty rare for a bowler to take bags of wickets while simultaneously being easy to score off. Yet an economical bowler with a good average (i.e. a good bowler) will not have the best strike rate. An expensive bowler who also concedes more runs per wicket may nevertheless have the same strike rate but imo will be a clearly inferior bowler.

The more I think about it the more inclined I actually am to prefer e/r because being hard to score off is fundamentally the quality of a good bowler. But ultimately, it's just average that matters really and the rest is horses for courses, style of bowler for the situation etc.

I don't really agree with that. There seems to be a lot of bowlers that are quite expensive yet also take wickets more often, and vice versa.

example:

Malinga: SR: 51.5, Econ: 3.85
Vaas: SR. 66.6, Econ: 2.68

very different types of bowlers, which you would prefer depends on the make-up of your team, opposition etc.

 

[thierry henry]

I don't really agree with that. There seems to be a lot of bowlers that are quite expensive yet also take wickets more often, and vice versa.

example:

Malinga: SR: 51.5, Econ: 3.85
Vaas: SR. 66.6, Econ: 2.68

very different types of bowlers, which you would prefer depends on the make-up of your team, opposition etc.

I don't think that example really refutes my point. Malinga averaged 33 and Vaas 29 which is not a huge difference but not inconsequential either. To my mind it supports the point that Vaas was a generally better bowler and hence was both harder to score off and conceded less runs per wicket.

But then again, I see the circularity in my own logic- I already tend to think of Vaas as being better, which is partly due to his better average, therefore I pat myself on the back and tell myself s/r is not a mark of a good bowler.

However, I do think that you are wrong and that there is a correlation between good average and good economy (good economy is partly causative of a good average, of course) and bowlers who are perceived as good in general. But I cbf crunching the numbers.

 

[thierry henry]

Put it this way, I reckon the overall e/r of the great bowlers of all time would be lower than the average e/r for all bowlers. But I could be wrong.