How valuable is that wicket!

[Ikki]

Sorry, you're right I did misunderstand it. I thought it was comparing the average of the batsman when you got him out (i.e. Zaheer takes Ponting 3 times (40, 30, 20) @ 30 and compare that to Ponting's overall average).

Which is weird because I interpreted it the right way round before in the thread. It's late...

 

[Ikki]

Although a different weaker criticism can be made from that. For example, you may concede your runs to a batsman that averages 20 yet take the wicket of a batsman that averages 50 out of the blue.

 

[ankitj]

Although a different weaker criticism can be made from that. For example, you may concede your runs to a batsman that averages 20 yet take the wicket of a batsman that averages 50 out of the blue.

That's fine. For your team you still got the same value. As for 'out of the blue' - this method is blind to that. Tendulkar did get a value of 53 odd when he dismissed Lara even if by fluke.

 

[Ikki]

Maybe that's an idea for another thread, if it's possible to get it done. The average of the average of all dismissals of all batsmen dismissed by a bowler compared to the average of the average of all the batsmen faced.

 

[ankitj]

That was the whole point we were making earlier on while contrasting Steyn and Marshall's discount factors. In the '80s there were a lot of world class bowlers, bringing down batsman averages overall and cutting into each others' discount factors. Whereas nowadays there aren't many bowlers around at the level of Steyn, hence he picks up fantastic value for dismissing the likes of Gambhir and Samaraweera, who are considered more valuable wickets than Gavaskar in this exercise.

Even there I want to caution that we need to know if the higher averages these days are due to (1) poor quality bowlers or (2) flatter pitches and better equipments.

In case (1) the value of wicket will be inflated while the bowling average will be just right. This will give a better than actual discount factor

If (2) is the case, there's nothing to take away from someone like Steyn's figures because in that case both the value of wicket and bowling average are likely to go up, keeping the discount factor unchanged

 

[vcs]

Even there I want to caution that we need to know if the higher averages these days are due to (1) poor quality bowlers or (2) flatter pitches and better equipments.

In case (1) the value of wicket will be inflated while the bowling average will be just right. This will give a better than actual discount factor

If (2) is the case, there's nothing to take away from someone like Steyn's figures because in that case both the value of wicket and bowling average are likely to go up, keeping the discount factor unchanged

True, agree with that. As always, it's probably a bit of both.

 

[Ikki]

I remember it was said that playing lesser batsmen shouldn't matter because a bowler will have to average better than the batsmen's average if they wanted to improve on their discount %, just as if the batsmen were greats. I created a mini model to see if that was the case.

I placed 3 teams where the Team C and D both have the same discount % but one is stronger than the other. It seems the overall discount % will be in favour of bowler A, even though his discount % across 60 wickets is the same, due to Team C lowering his overall average enough.

Is the logic correct in the above example?

 

[vcs]

Yes, for any rational number, a/b < (a+x)/(b+x). You have certainly raised an interesting point though, it seems there is some advantage to playing minnows more often, however little that may be. Even minnow batsmen would usually average 20 odd whereas ATG bowlers can rack up sub-20 averages and loads of wickets against them. So some healthy discount to be gained there.

 

[Prince EWS]

Yes, for any rational number, a/b < (a+x)/(b+x). You have certainly raised an interesting point though, it seems there is some advantage to playing minnows more often, however little that may be. Even minnow batsmen would usually average 20 odd whereas ATG bowlers can rack up sub-20 averages and loads of wickets against them. So some healthy discount to be gained there.

Yeah, the same thing happened when I standardised bowling averages across opposition quality. The best bowlers seem to benefit more from playing poor oppositions than the standardisation method actually accounts for. For example, Bangladesh averaged about 19 with the bat per batsman innings in the 2000-09 period and Murali averaged just 13 against them. Now Murali's obviously going to average less than the overall batting averages of most teams because he was much better than the average bowler, but accounting for the quality of the Bangladesh, he had a standardised average of 20.89 against them and his overall standardised average was 22.28. That's just one example but it's something I noticed when I looked into all the players - there seems to be a more exponential trend with the better bowlers than straight division and multiplication methods account for. It obviously makes it a lot less of a factor than the real average, but it's still a small factor. The same thing would happen in this analysis as well.

 

[ankitj]

I remember it was said that playing lesser batsmen shouldn't matter because a bowler will have to average better than the batsmen's average if they wanted to improve on their discount %, just as if the batsmen were greats. I created a mini model to see if that was the case.

I placed 3 teams where the Team C and D both have the same discount % but one is stronger than the other. It seems the overall discount % will be in favour of bowler A, even though his discount % across 60 wickets is the same, due to Team C lowering his overall average enough.

Is the logic correct in the above example?

That's hardly surprising and hardly unfair. What's happening here is that both bowlers do poorer than their own standard (100% instead of 83.33%) against one team. The bowler that does poorer against the stronger team suffers more than the one who does poorer against the weaker team. This means that the outlier performances (by your own standard) against stronger teams have bigger influence on your ultimate discount factor. Exactly how it should be

Flip side of this is that if the two bowlers did better than their own standard, let's say 50% against teams C and D, then the the one turning out better performances against the stronger team (Team D) benefits more. Bowler A gets 78.6% and Bowler B gets 72.2%. Pretty fair. Precisely how it would work in PEWS' method.

Note that the key is that comparison happens with one's own standard.

 

[ankitj]

The conventional averages on the other hand are blind to who the outlier performances are against.

 

[MrIncredible]

Had the pleasure of reading thru this thread.

Very Informative

Ambrose WAG and who the hell was it who said he was overrated??

Same goes for Mr Marshall and MCgrath

Ps Good to see that Ikki is still up to his same old trickks of throwing a temper tantrum whenever one of Lillee or Warne isnt high on a list. Dont worry mate Mcgrath is still bearing the torch lol

 

[Ikki]

Well really, if I wanted to tout McGrath it would be very easy as he is statistically up there with anyone. FTR, I think there is a bit more to cricket and watching Warne alongside him I'm in less doubt than others that it was the wily spinner who was more important to his side. Lillee is something like the symbol of the fast bowler in Australia.

 

[MrIncredible]

I was kidding mate dont take ne thing I say above too seriously

I know about Lillee's status in Ozland

It may surprise u to know he is also revered in the Caribbean as well. I know my dad rates him above all and he's been following cricket since the 50s. For me its been early 70s to now.

 

[ankitj]

I found the original article that used this method:

So, how 'valuable' was that wicket?

It would be worth seeing how the bowlers fared back in 2003.

 

[Debris]

One other little flaw in this analysis is that sometimes the value of a batsman's wicket changes over his career. Steve Waugh's wicket was worth a lot more in 1995 than it was in 1987 as an extreme example. Mohammed Yousuf during his golden years much more worthwhile that outside that period. Not sure how you cater for that.

 

[kyear2]

Long time back I came across a comprehensive analysis of bowlers based on the "value"of the wickets they take. The approach was wonderfully elegant and very logical. The article is not online anymore.

Nevertheless I decided to do the exercise again. I am happy to share the results of my hard work over last 3-4 days with you

Shortlisting the bowlers

For this analysis I chose a total of 41 bowlers. All 24 bowlers in the 300+ club are chosen. And some of the greats from the sub 300 group are chosen too. This includes the 3 great West Indian fast bowlers - Garner, Holding and Roberts. 3 great Australian fast bowlers of 1950's - Lindwall, Miller, Davidson. The great spinners - O'Reilly, Grimmet, Laker and Verity. 3 Great fast bowlers from pre world war I era - Barnes, Spofforth and Lohman. Two recent bowlers out of curiosity - Gillespie and Steyn. And finally out of curiosity - Sobers and Kallis.

The averages of these bowlers are listed below:

[B]Bowler        Bowling Average[/B]
George Lohman    10.75
Sydney Barnes    16.43
Fred Spofforth    18.41
Alan Davidson    20.53
M Marshall    20.94
Joel Garner    20.97
Curtly Ambrose    20.99
Jim Laker    21.24
Fred Trueman    21.57
Glenn McGrath    21.64
Allan Donald    22.25
Richard Hadlee    22.29
Bill O'Reilly    22.59
M Muralitharan    22.72
Imran Khan    22.81
Keith Miller    22.97
Ray Lindwall    23.03
Shaun Pollock    23.11
Dale Steyn    23.13
Waqar Younis    23.56
Wasim Akram    23.62
Michael Holding    23.68
Dennis Lillee    23.92
C Grimmett    24.21
Hedley Verity    24.37
Courtney Walsh    24.44
Bob Willis    25.20
Shane Warne    25.41
Andy Roberts    25.61
Jason Gillespie    26.13
Ian Botham    28.40
Mkhaya Ntini    28.82
Lance Gibbs    29.09
Chaminda Vaas    29.58
Kapil Dev    29.64
Anil Kumble    29.65
Brett Lee    30.81
Jacques Kallis    31.59
Harbhajan Singh    31.66
Daniel Vettori    33.86
Gary Sobers    34.03

Value of Wickets

What the averages don't tell you however is how valuable were the wickets taken by a bowler. A wicket of Harbhajan Singh is not as valuable as that of Sachin Tendulkar. That fact is howerver hidden in the overall bowling averages. In order to put value to every wicket taken, therefore, we can use the career batting average of the batsman dismissed as a proxy. Therefore value of Tendulkar's wicket is about 57 and that of Harbhajan only 16.

Now we can average the values of all the wickets taken by a bowler. Thus bowler X who takes Tendulkar's wicket thrice and Harbhajan's twice in a series (and takes no other wickets) has an average value of wicket of 40.6.

Taking such averages over the whole careers of the chosen bowlers, we get the following ranking based on the average value of wickets:

[B]Bowler        Bol Ave     Avg value of wicket [/B]
Bill O'Reilly    22.59     31.56
Mkhaya Ntini    28.82     31.47
Dennis Lillee    23.92     31.27
Jacques Kallis    31.59      31.25
Brett Lee    30.81     31.24
Hedley Verity    24.37     31.23
Keith Miller    22.97     31.17
Glenn McGrath    21.64     31.02
Gary Sobers    34.03     30.93
Chaminda Vaas    29.58     30.74
M Marshall    20.94     30.48
Harbhajan Singh    31.66     30.47
Lance Gibbs    29.09     30.44
Anil Kumble    29.65     30.25
Curtly Ambrose    20.99     30.23
Shaun Pollock    23.11     30.16
Michael Holding    23.68     30.14
Alan Davidson    20.53     30.04
Jason Gillespie    26.13     29.97
Bob Willis    25.20     29.95
Imran Khan    22.81     29.91
Jim Laker    21.24     29.83
Kapil Dev    29.64     29.80
Courtney Walsh    24.44     29.74
Allan Donald    22.25     29.71
Ray Lindwall    23.03     29.59
Richard Hadlee    22.29     29.54
Daniel Vettori    33.86     29.44
Shane Warne    25.41     29.32
Ian Botham    28.40     29.30
Dale Steyn    23.13     29.29
Andy Roberts    25.61     29.01
C Grimmett    24.21     28.99
Fred Trueman    21.57     28.90
M Muralitharan    22.72     28.51
Waqar Younis    23.56     27.65
Joel Garner    20.97     27.15
Wasim Akram    23.62     26.50
Sydney Barnes    16.43     25.02
Fred Spofforth    18.41     20.01
George Lohman    10.75     15.92

Price paid for the value

A bowler however cannot be critiqued for only having a low average value of wicket. Some bowlers may play more games against weeker oppositions. Some may come on to bowl only as first or second change and may get less chance to bowl to top order batsmen.

However if a bolwer does bowl more to weeker batsmen and dismisses them more often, he must also take those wickets more cheaply i.e. must pay lesser price for taking those wickets. This price is the bowling average of the bowler.

Consider the bowler X in the previous example averaged 30 for each wickets (i.e. gave away 150 runs for his 5 wickets). He has effectively paid a price of 30 while getting rid of batsmen that valued 40.6. Another way to look at it is that bolwer X discounted the batting average of the batsmen he bowled to down to 30/40.6 = 73.8%. Lower the "disconut factor" more effective the bowler. A discount rate of 80% would mean that a batsman who averages 50 overall will average 40 against that bowler in duels.

Note that it is immaterial if that "price" paid were runs scored by batsmen the bowler X ultimately dismissed or by others. That's how many runs he gave away in search of those wickets.

With bowling averages and average value of each wicket available with us, we can calculate discount factors for each of the 41 bowlers. Doing this, we get the following ranking on the basis of discount factors:

[B]Rank    Bowler        Bow Ave     Avg value Discount Factor[/B]
1    Sydney Barnes    16.43     25.02     65.7%
2    George Lohman    10.75     15.92     67.5%
3    Alan Davidson    20.53     30.04     68.3%
4    M Marshall    20.94     30.48     68.7%
5    Curtly Ambrose    20.99     30.23     69.4%
6    Glenn McGrath    21.64     31.02     69.8%
7    Jim Laker    21.24     29.83     71.2%
8    W O'Reilly    22.59     31.56     71.6%
9    Keith Miller    22.97     31.17     73.7%
10    Fred Trueman    21.57     28.90     74.6%
11    Allan Donald    22.25     29.71     74.9%
12    Richard Hadlee    22.29     29.54     75.4%
13    Imran Khan    22.81     29.91     76.3%
14    Dennis Lillee    23.92     31.27     76.5%
15    Shaun Pollock    23.11     30.16     76.6%
16    Joel Garner    20.97     27.15     77.2%
17    Ray Lindwall    23.03     29.59     77.8%
18    Hedley Verity    24.37     31.23     78.0%
19    Michael Holding    23.68     30.14     78.6%
20    Dale Steyn    23.13     29.29     79.0%
21    M Muralitharan    22.72     28.51     79.7%
22    Courtney Walsh    24.44     29.74     82.2%
23    C Grimmett    24.21     28.99     83.5%
24    Bob Willis    25.20     29.95     84.2%
25    Waqar Younis    23.56     27.65     85.2%
26    Shane Warne    25.41     29.32     86.7%
27    Jason Gillespie    26.13     29.97     87.2%
28    Andy Roberts    25.61     29.01     88.3%
29    Wasim Akram    23.62     26.50     89.1%
30    Mkhaya Ntini    28.82     31.47     91.6%
31    Fred Spofforth    18.41     20.01     92.0%
32    Lance Gibbs    29.09     30.44     95.6%
33    Chaminda Vaas    29.58     30.74     96.2%
34    Ian Botham    28.40     29.30     96.9%
35    Anil Kumble    29.65     30.25     98.0%
36    Brett Lee    30.81     31.24     98.6%
37    Kapil Dev    29.64     29.80     99.5%
38    Jacques Kallis    31.59     31.25    101.1%
39    Harbhajan Singh    31.66     30.47     103.9%
40    Gary Sobers    34.03     30.93     110.0%
41    Daniel Vettori    33.86     29.44     115.0%

My key take aways

Around the themes of some of the recent discussions:

1. Ambrose is NOT over-rated
2. Murali vs Warne debate is settled
3. Wasim Akram as a test bowler is a level below some of ATG fast bolwers

Hope you enjoyed reading this

EDIT: Added Kallis.

Excellent work here. Have you ever checked to see where Steyn would rank with this metric?

 

[Spark]

39 Harbhajan Singh 31.66 30.47 103.9%
41 Daniel Vettori 33.86 29.44 115.0%

Real oofs there. Basically being below replacement value across a career.

I'd be very interested in seeing this done now after we've had a good half-decade plus of really spicy pitches around the world in general, as opposed to the flat pitch hell of the 2000s and early 2010s. This sort of simple, elegant analysis (if not taken too far in the interpretation stage) is more of what we need to see in this space.

 

[ankitj]

Excellent work here. Have you ever checked to see where Steyn would rank with this metric?

I think I did after few more years of Steyn excellence.

FWIW, I redid this calculation for Steyn.

Then, he had a bowling average of 23.13, average value of wicket of 29.29, and discount factor of 79.0%. He was ranked #20

Now, he has a bowling average of 22.67, average value of wicket of 30.57, and discount factor of 74.2%. This will put him at #10 (without recalculating for others). Among those with 200+ wickets, he is behind only Marshall, Ambrose and McGrath (the margin is fair bit between these three and Steyn though).

 

[Coronis]

Kallis > Sobers confirmed


And solid proof Davidson is underrated.