The Statistical Gap.... 185 against 234. The true gap is possibly 60 runs.

[Imtiaz]

We all refer from time to time the gap between us and the "big boys". I have tried to calculate this "gap" statistically.

Please note that this is not a serious exercise. So, it is not a scientific prediction or anything like that. This is only for idle amusement. Sadly, it is not very funny.

I have taken the ODI batting averages of my favoured XI. As batting average is for completed innings', I have multiplied the total of the averages by 10/11 to arrive at the "expected mean score". Here goes:

Bashar 18.01
Faisal 17.00
Ashraful 15.61
Kapali 21.43
Rana 34.00 [ 3 n.o. out of 8 inn.]
Mashud 17.50
Mushfique 16.05
Mahmud 13.48
Rafique 13.85
Tapash 9.47
Razzak 27.00 [ 2 n.o. out of 3 inn. ]

Total 203.40 x 10 / 11 = 185

This is my mean expected score. Have you noticed how closely bunched the averages are ? It does not matter almost which position they bat at.

Now for the bowling: I have taken the economy rate as the basis since you cannot forecast the number of wickets each person would take.

Tapash 5.48
Mushfiq 4.28
Mahmud 4.95
Razzak 3.20
Rafique 4.60
Rana 4.04

Average 4.425

Total for 50 overs = 4.425 x 50 = 221.

This , I admit is very optimistic. It is because, Razzak's and Rana's economy rate are better than they would be as they have played very few matches. In Razzak's case Hong Kong is 1/3 of his career. Rana played in Zimbabwe. If Razzak's rate is adjusted upward by 1.00 and Rana's by 0.50, the expected opposition total comes to 234.

As I have accorded equal probability to each bowler, it assumes that they will bowl 8.33 overs each. In reality, Tapash, Rafique and Razzak are almost certain to bowl 10 overs each if Bangladesh bowl first.

Remember, the batting averages and the economy rates are based on all the countries these batsmen and bowlers have played against - not against Sri Lanka only. The batting averages against the "big boys" are bound to be lower hence the expected score will also be lower than 185. The opposite for the bowling. Therefore, that is expected to be higher.

I believe the true gap is about 60 runs

If I have time I will calculate Sri Lanka's expected mean score separately from their own batting averages.

[Edited on 23-7-2004 by Imtiaz]

[AsifTheManRahman]

Great analysis. A bit more thinking and improvement will make it even better.

[Tintin]

Very interesting analysis from Imtiaz. As he has mentioned, the batting averages are inflated by the presence of Rana and Razzaq who have played very few matches. So I just tried to improve it by taking the actual average of the team. Similarly for bowling, Razzak has an unfairly low runrate.

The five bowlers who bowled on Wednesday have together bowled 1110 overs and 3 balls, conceded 5289 runs at an average of 41.64 and a run/over of 4.76 [with Mushfiq included it will be 41.88 and 4.70]. That means that the average total made against Bangladesh is 238 + b/lbs.

The 11 players who played on Wednesday have together scored 5231 runs and were dismissed 308 times. From this the batting average of the whole team is 16.98 It is a fair assumption that we lose all the wickets.

I don't know how to interpret these numbers. Does this mean that the average score is 170 (+ extras) or should we apply a multiplication factor here too ? So I pass it back to you

Anyway, this is quite close to the estimate of 60 runs.



[Edited on 22-7-2004 by Tintin]

[chinaman]

Very interesting Projection. I don't know if any recognized entity uses such calculations. But it really looks great.

For comparisn and better analysis may I suggest the followings:
[list=1][*]Take regular 14 players and calculate their collective batting average. Then we can divide it by 14 to get "Average Individual Run" which could likely be around 16 leading to an expected team total of 176.
[*]Take 10 most regular bowlers and calculate their collective bowling average and get the "Average Individual Economy". This could likely be around 4.8 leading to an expected opponent score of 240.
[*]Now, lets call the gap "Team Performace Gap". So, the TPG for Bangladesh is Minus 66 (-66)
[*]We can calculate similar TPG for Test as well as for other nations periodically to see where we stand in a bigger picture.[/list=1]


Note: The reason why I said to use 14 players and 10 most regular bowlers is that it represents a nation instead of a team only.

Since the test has no specific over limit, we can use average overs played by a nation. Better idea?

[Edited on 7-22-2004 by chinaman]

[Tintin]

Take regular 14 players and calculate their collective batting average....

It is getting late here but for a start, this is the stats of the current 11 :

Batting (Runs / Completed innings) :

HB 919 51
Ash 484 31
Rjn 422 17
Kp 793 37
Fa 34 2
Msh 1015 58
Rna 170 5
Rf 651 47
Suj 782 58
Rz 27 1
Az 24 1

5321 308 16.98

Bowling (Runs, Wkts, Overs) :

Azz 392 12 70.5
Raz 93 5 29
Suj 2282 54 457
Raj 2322 50 504.33
Ran 200 6 49.5

5289 127 1110.33 4.76


Mushfiq 784 18 183.17

Total : 6073 145 4.70 41.88

[chinaman]

Great Tintin.

So if we take 14 players, the contribution may actually go down from 16.98.

Similarly the Economy Rate may go up from 4.76 for the bowlers if we add few more regulars.

At any rate, our TPG seems to be around minus 60-70. God knows what would be the TPGs from other nations!

[Edited on 7-22-2004 by chinaman]

[chinaman]

Split this thread for relevency.

[Piranha]

Imtiaz, great post and excellent analysis.

[Piranha]

Imtiaz, great post and excellent analysis.

[Imtiaz]

I didn't realise there would be so many posts on this one. Usually, stats can be quite boring.

I made a glaring error in not including byes and leg byes. No wonder when I first tabulated the figures it seemed quite low. For Bangladesh, 185 appeared eminently sensible though ! Even a little on the high side. This shows there has been recent improvement. My hunch was it would be around 170 !

I think your contributions have enhanced the "value" of the calculations.

Chinaman, your TPG is a runner. May I suggest that rather than adding all the historical scores, we could have TPG10 [ i.e. the last 10 matches ], TPG20 etc. This will have more relevance as clearly recent results are more important.

Do you want to complicate matters further ? Take a weighted average of totals giving more weight to recent scores [ or, give more weight to stronger teams ]. I leave such "interesting !!" calculations to those of you prepared to burn the midnight oil !

By the way, byes and leg byes have to be factored in.

I still think, in the batting calculations, a multiplier of 10/11 is needed as all averages by definition are for completed innings'. Not out scores are included in the runs , so the Not Out batsman[men]'s contribution is built-in in every batsman's average.

I am prepared to be corrected.

[chinaman]

TPG10, TPG20 etc seem cool. As an example, we could say BD TPG10 before and after the Asia Cup was -60 and -50 respectively which shows some improvements.

About the 10/11 issue, I was trying to avoid it in favor of an "Average Individual Run" system because rarely a team continues to play beyond 5/6 matches without making some change in the line up. Besides, a larger pool is more reflective of the actual strength. By taking the average of the standard "Batting Average" we effectively avoid repeating the not out issue which is already in the batting average.

I wouldn't worry about the byes because more or less 10-15% runs come from it and since it is get added to both for and against scores, it cancels each other out.

Great going! Please keep it up.

[Imtiaz]

The following is a further refinement of the ideas set out above. The aim is to compute a score which expresses a team's performance capability - not just as a ranking but also as a relative measure. For example, ODI championship table says Sri Lanka has 106 points and Bangladesh has 1 point. It , of course, follows that Sri Lanka is better than Bangladesh but surely not 106 times better !! It quite accurately gives a ranking but does not aim to indicate performance capability.

Below, I have considered all 2004 ODI matches played by each country. The following are the team batting and bowlind summary data:

Pl. Inn. NO. Runs Ave. Exp.Score
[Pl.] [Balls] [Runs] [Ec/R] [Op.Sc.]

BDESH 9 92 13 1410 17.85 178.48
Opp. 9 2233 1545 4.15 207.57
TPG2004 -29.09

SL 13 106 22 2253 26.82 268.21
Opp. 13 3489 2370 4.08 203.78
TPG2004 64.43

IND 18 155 31 4143 33.41 334.11
Opp. 18 4890 4346 5.33 266.63
TPG2004 67.49

PAK 13 121 21 3107 31.07 310.70
Opp. 13 3501 3006 5.15 257.58
TPG2004 53.12

AUS 18 4280 36.27 362.71
Opp. 18 4770 3535 4.45 222.33
TPG2004 140.38

ENG 10 1754 26.58 265.76
Opp. 10 2265 1922 5.09 254.57
TPG2004 11.19

Note: Batting Expected Total = Team Total Runs / No.of completed innings

Opposition Expected Total = Economy Rate * 50 overs

Based on this measurement, Bangladesh is 29 runs worse off against our 2004 representative opposition [ which is ZIM3, WI3, HK1, PAK1,IND1 ]. Bangladesh has a very good economy rate. But, bear in mind, our 2004 opposition.

Sri Lanka's score is +64. India is slightly better than SL.

Australia by the way stands head and shoulders above everyone else. The ZIM tour may have inflated these figures.

Another way of looking at the same data is through the Stricking Rate. This has the attraction in that it is the opposite of the Economy Rate. The Bangladesh S/R for 2004 is: 59.44. Thus Expected Team Total = 59.44 /100 * 6 * 50 = 178.32
Opposition = 207.57
Net Striking Rate [NSR]= -29.25

Sri Lanka ETT 71.18*3= 213.54 Opposition = 203.78
Net Striking Rate [NSR]= 9.76

Just like the official points system, a ranking can be made here as well.

As they say, there are many ways to skin a cat !

[Edited on 23-7-2004 by Imtiaz]

[Imtiaz]

Aus 86.36
BDESH 59.44
ENG 77.64
HKG 41.83
IND 81.76
NZ 79.41
PAK 83.32
RSA 79.88
SL 71.18
UAE 41.63
WI 74.82
ZIM 64.45

Bangladesh's low scoring rate can be attributed to the fact that in the latter stages of an innings, unlike other countries, our focus is more towards finishing the 50 overs rather than increasing the scoring rate as we seldom have wickets in hand.

[Arnab]

Why don't you take the scoring rates of all teams for their past 25-30 matches. That's probably the average amount of ODIs teams play each year.

[Edited on 23-7-2004 by Arnab]

[chinaman]

To make it even more easier, we can use all the available standard data, the Team Average Score and Opponent Average Score.

At Present:
Team Average Score: 166
Opponent Average Score: 211
Statistical Gap: -45

Before The Asia Cup:
Team Average Score: 165
Opponent Average Score: 212
Statistical Gap: -47

Before Whatmore (May 2003):
Team Average Score: 164
Opponent Average Score: 215
Statistical Gap: -51

So, we made improvement of "2" Statistical Gap during the Asia Cup and "6" Statistical Gap by Whatmore. This proves Whatmore actually made improvements!

[chinaman]

To make it even more easier, we can use all the available standard data, the Team Average Score and Opponent Average Score.

For Bangladesh:

At Present:
Team Average Score: 166
Opponent Average Score: 211
Statistical Gap: -45

Before Asia Cup:
Team Average Score: 165
Opponent Average Score: 212
Statistical Gap: -47

Before Whatmore (May 2003):
Team Average Score: 164
Opponent Average Score: 215
Statistical Gap: -51

So, we made improvement of "2" Statistical Gap during the Asia Cup and "6" Statistical Gap by Whatmore. This proves Whatmore actually made improvements!

---

For Pakistan:

At Present:
Team Average Score: 208
Opponent Average Score: 199
Statistical Gap: +9

Before Asia Cup:
Team Average Score: 208
Opponent Average Score: 200
Statistical Gap: +8

---

For India:

At Present:
Team Average Score: 208
Opponent Average Score: 209
Statistical Gap: -1

Before Asia Cup:
Team Average Score: 208
Opponent Average Score: 209
Statistical Gap: -1

---

For Sri Lanka:

At Present:
Team Average Score: 202
Opponent Average Score: 210
Statistical Gap: -8

Before Asia Cup:
Team Average Score: 202
Opponent Average Score: 210
Statistical Gap: -8

---

So, during the Asia Cup, the Statistical Gap for India and Sri Lanka remains unchanged, Pakistan went up 1 and Bangladesh went up 2.

Please note: Before Asia Cup means upto May 31, 2004.

[Edited on 7-23-2004 by chinaman]

[chinaman]

Before Whatmore:

Batting Average: 18.5
Bowling Average: 58.2

At Present:

Batting Average: 20.00
Bowling Average: 53.3

So, our Batting Average is up 1.5 and Bowling Average is down 4.9 since May 31, 2003. This proves Whatmore actually made improvements in test area too!

Batting Average = batting average of runs per wicket.
Bowling Average = bowling average of runs conceded per wicket taken.

[Tintin]

CM,

The stats4.cricket.com has a problem that they are not taking the number of overs faced while calculating the team and opponent score.

For eg, in the three matches against England, BD batted first every time. If we check the stats for that our score is given as 153 and opposition's as 156, even though our RR is much inferior : Link

We may always have to go the way Imtiaz has done five posts ago.


[Edited on 23-7-2004 by Tintin]

[Tintin]

I have a suggestion about the Imtiaz method also that we should not assume that the teams lose all the wickets..

IND 18 155 31 4143 33.41 334.11
Opp. 18 4890 4346 5.33 266.63
TPG2004 67.49

India lost 124 wickets in the 18 matches or 124/18 wkts per match. So India's average score should be 4143 * (124/18) = 285.41. So the TPG will be +18.78. India won 10 and lost 8 this year, so this is reasonable. For Australia, the projected score will come down to 280.5 and the margin 58 runs.

But for Bangladesh, this method gives only 160.63. Is this because of the two 25 over matches that we played ?

This thread is making me acutely aware of my lack of knowledge of mathematics



[Edited on 23-7-2004 by Tintin]

[Imtiaz]

It's like looking at a cube from different sides.

[Tintin]

Thanks Imtiaz, but I keep getting stuck.

But for Bangladesh, this method gives only 160.63. Is this because of the two 25 over matches that we played ?

What is the issue here ?

[Imtiaz]

I am currently following a collapse on the net. Has anyone thought of Bashar playing at No.5 to protect him from the moving ball. Better, why not everyone play at No.5 as all the averages appear to be the same.

[Arnab]

Heh, the statistical gap with Srilanka seems to be TEN WICKETS.

[Imtiaz]

If the laws of cricket allowed, I am sure Sri Lanka's margin of victory would have even bigger ! It was an annihilation. We just have to chin it out.

[Arnab]

I have a question. Why isn't the statistical gap normalized for both

a.# of runs in 1 over. (BD and opponents)
b. # of runs for 1 wicket. (ditto)

It seems to me that it would give you a better picture of the relative strengths of the teams.