Saturday, April 14, 2012
Balls faced errors
Charles Davis, in his blog post titled "Balls Faced Blues" (31 March 2012), reports on errors in the previously published scorecards to this Test between the West Indies and Australia in 1983/4. He writes:
I had never thought to compare the balls faced with the total overs in an innings. I was surprised to find that substantial disagreements are quite common. Some discrepancies are to be expected – a single off a no-ball used to be an extra ball but not a sundry; many wides can come from one ball; a wide used to be counted as a ball faced in Australia; there have been plenty of 5- and 7-ball overs which won't show up in the team totals.
But still. Below is a list of innings where the discrepancy is at least 20 balls. It's enormous. I've defined "discrepancy" as the sum of the balls faced, minus the total number of balls in the innings, minus the extras. I had originally planned to manually check all of these anomalies, but I couldn't be bothered. I've made notes on a few of the more egregious cases.
I didn't output the dates of the Tests when making this table. The quickest way to pull up the scorecard is probably via CricketArchive's Scorecard Oracle. To bring up Test 104, enter t104 as the 'match id'. The Test that Davis talked about is t983 – you can see that all three innings of that match are on the list, as are two innings from the fourth Test of that series.
The following list is based on the CricketArchive scorecards as they were a few years ago. Probably some errors have been ironed out since. (I've also omitted one scorecard which had obviously flipped the balls-faced and minutes; hopefully it will be corrected in the coming days.)
A full re-score confirms Border’s runs and the sequence of strokes, yet the balls faced figures as published seem to be quite wrong. Border is given 314 balls for his 98* and 269 for the 100*. The figures I get are 283 and 285 balls, respectively. Viv Richards’ figures for his 76 are also very doubtful: I get 130 balls not 188. There are other discrepancies. Using the revised figures, the balls faced now reconcile with the bowling figures.
I had never thought to compare the balls faced with the total overs in an innings. I was surprised to find that substantial disagreements are quite common. Some discrepancies are to be expected – a single off a no-ball used to be an extra ball but not a sundry; many wides can come from one ball; a wide used to be counted as a ball faced in Australia; there have been plenty of 5- and 7-ball overs which won't show up in the team totals.
But still. Below is a list of innings where the discrepancy is at least 20 balls. It's enormous. I've defined "discrepancy" as the sum of the balls faced, minus the total number of balls in the innings, minus the extras. I had originally planned to manually check all of these anomalies, but I couldn't be bothered. I've made notes on a few of the more egregious cases.
I didn't output the dates of the Tests when making this table. The quickest way to pull up the scorecard is probably via CricketArchive's Scorecard Oracle. To bring up Test 104, enter t104 as the 'match id'. The Test that Davis talked about is t983 – you can see that all three innings of that match are on the list, as are two innings from the fourth Test of that series.
The following list is based on the CricketArchive scorecards as they were a few years ago. Probably some errors have been ironed out since. (I've also omitted one scorecard which had obviously flipped the balls-faced and minutes; hopefully it will be corrected in the coming days.)
Test no. Home Away inns sum BF inns balls nb wd discrepancy
104 England Australia 4 310 330 0 0 20
180 Australia England 4 832 805 0 0 27
290 Australia India 3 368 399 1 0 32
303 England Australia 1 277 253 0 0 24
470 New Zealand England 2 377 437 0 0 60 Judging by the minutes,
I think Bolton faced ~60
more balls than is given
537 Australia England 1 715 693 0 2 20
598 Australia England 1 883 861 1 0 21
613 South Africa Australia 2 736 763 0 0 27
628 Australia India 4 618 590 0 0 28
673 South Africa Australia 3 779 728 20 0 31
703 India England 1 458 508 0 0 50 Minutes make sense.
Probably Venkat faced more.
708 India England 3 522 636 3 0 117 Minutes make sense.
Probably Denness faced more.
730 Australia New Zealand 3 740 717 2 0 21
731 West Indies England 2 798 736 16 0 46
748 Australia England 3 645 729 11 1 96 Wisden says Titmus batted for
200 minutes, not 108; probably
faced a lot more. Old too?
757 New Zealand England 3 401 461 12 0 72
835 India West Indies 1 812 822 13 0 23
835 India West Indies 2 928 966 8 0 46
855 India Australia 2 831 783 23 0 25
855 India Australia 3 657 682 3 0 28
858 India Australia 1 932 866 24 0 42
858 India Australia 2 628 639 19 0 30
859 India Australia 2 787 754 8 4 21
863 India Pakistan 1 593 551 4 0 38
865 India Pakistan 3 399 419 5 0 25
869 India Pakistan 3 441 400 17 0 24
876 Pakistan Australia 2 664 713 10 0 59
876 Pakistan Australia 3 521 540 2 1 22
878 Pakistan Australia 1 1302 1266 3 0 33
878 Pakistan Australia 2 724 756 15 0 47
889 Australia New Zealand 3 316 289 5 2 20
891 Australia New Zealand 1 799 729 6 3 61
908 England Australia 3 658 626 7 1 24
914 India England 2 955 961 21 4 31
918 India England 2 917 935 12 0 30
922 New Zealand Australia 1 762 696 11 4 51
924 New Zealand Australia 4 138 178 0 0 40
937 Pakistan Australia 3 494 540 8 0 54
970 Australia Pakistan 1 722 734 4 4 20
983 West Indies Australia 1 605 553 17 0 35
983 West Indies Australia 2 815 756 1 2 56
983 West Indies Australia 3 634 673 14 1 54
985 Sri Lanka New Zealand 2 1079 1050 6 2 21
987 West Indies Australia 1 563 581 10 0 28
987 West Indies Australia 2 850 874 1 0 25
993 England West Indies 3 639 579 0 0 60
998 Pakistan New Zealand 1 392 448 0 0 56
998 Pakistan New Zealand 2 530 560 2 0 32
998 Pakistan New Zealand 3 441 498 4 1 62
998 Pakistan New Zealand 4 347 391 0 0 44
1000 Pakistan New Zealand 1 674 651 1 0 22
1000 Pakistan New Zealand 3 378 337 3 0 38
1010 New Zealand Pakistan 2 1002 882 6 0 114 Probably Wright faced a lot less.
1033 Australia India 2 930 890 17 0 23
1045 West Indies England 2 635 646 40 0 51
1052 India Australia 2 600 566 6 0 28
1052 India Australia 3 273 294 1 0 22
1058 Pakistan West Indies 2 566 661 2 1 98 Rameez faced much more.
1065 Australia England 2 598 564 2 2 30
1072 New Zealand West Indies 3 382 423 8 0 49
1087 Australia New Zealand 4 577 552 4 0 21
1097 West Indies Pakistan 3 546 563 14 0 31
1118 West Indies India 1 618 632 22 0 36
1118 West Indies India 2 667 671 22 0 26
1118 West Indies India 3 543 585 13 0 55
1118 West Indies India 4 271 288 3 0 20
1152 Pakistan New Zealand 2 782 750 12 0 20
1163 Australia England 1 509 400 5 1 103 Atherton faced much less.
1170 West Indies Australia 3 584 547 4 0 33
1180 Australia India 2 836 732 11 0 93 Probably Jones faced much less.
1186 Australia India 3 719 681 6 0 32
1195 Sri Lanka Australia 3 715 690 5 0 20
1220 West Indies Pakistan 1 222 230 11 2 21
1232 Sri Lanka South Africa 4 592 672 15 1 96 Probably Rhodes faced much more.
1237 Pakistan Zimbabwe 1 903 930 1 0 28
1238 Australia New Zealand 1 608 633 2 0 27
1239 Sri Lanka West Indies 3 102 133 1 0 32
1247 India Sri Lanka 1 350 383 0 0 33
1247 India Sri Lanka 3 608 627 1 0 20
1262 England New Zealand 1 855 879 2 1 27
1266 England South Africa 1 543 554 18 1 30
1268 Pakistan Australia 2 558 523 4 0 31
1278 India West Indies 1 855 856 14 5 20
1278 India West Indies 2 767 771 17 1 22
1282 South Africa New Zealand 3 643 667 6 1 31
1291 New Zealand South Africa 1 663 633 0 0 30
1293 New Zealand Sri Lanka 2 887 834 15 0 38
1325 West Indies New Zealand 1 369 372 17 1 21
1325 West Indies New Zealand 2 949 987 12 0 50
1415 Sri Lanka New Zealand 1 665 632 9 1 23
# posted by David Barry : 10:37
Comments:
<< Home
This is pretty crazy. I was thinking about how these errors come about so I decided to look at the distribution of the signed discrepancys. Sorted, the samples are:
-117 -98 -96 -96 -72 -62 -60 -59 -56 -55 -54 -54 -51 -50 -50 -49 -47 -46 -44 -40 -36 -33 -32 -32 -32 -31 -31 -31 -30 -30 -30 -28 -28 -28 -27 -27 -27 -26 -25 -25 -23 -22 -22 -22 -21 -21 -20 -20 -20 -20 -20 20 20 20 20 21 21 21 21 21 22 23 23 24 24 24 25 27 28 28 30 30 31 31 32 33 33 35 38 38 38 42 46 51 56 60 61 93 103 114
I couldn't come up with a graph that tells me anything too exciting so I'll just stick to the data.
The fact that there are 7 numbers with magnitude near 100 and none from 80-90 suggests that those are simple transcription errors in the 100s place.
This suggests that many of the number near 20,30,50 etc are the same problem in the 10s place.
When I was scoring in junior cricket and we bothered to count dot balls for batsmen, the balls faced for batsman was almost always less than the balls bowled (you definitely remember to put the dot in the bowler's analysis but it's easy to forget the dot for the batsmen). This suggests bad scoring gives negative discrepancies. Looking at the data there are more negative than positive.
Anything else you can think of that causes it or a better way to test which problems came from where?
-117 -98 -96 -96 -72 -62 -60 -59 -56 -55 -54 -54 -51 -50 -50 -49 -47 -46 -44 -40 -36 -33 -32 -32 -32 -31 -31 -31 -30 -30 -30 -28 -28 -28 -27 -27 -27 -26 -25 -25 -23 -22 -22 -22 -21 -21 -20 -20 -20 -20 -20 20 20 20 20 21 21 21 21 21 22 23 23 24 24 24 25 27 28 28 30 30 31 31 32 33 33 35 38 38 38 42 46 51 56 60 61 93 103 114
I couldn't come up with a graph that tells me anything too exciting so I'll just stick to the data.
The fact that there are 7 numbers with magnitude near 100 and none from 80-90 suggests that those are simple transcription errors in the 100s place.
This suggests that many of the number near 20,30,50 etc are the same problem in the 10s place.
When I was scoring in junior cricket and we bothered to count dot balls for batsmen, the balls faced for batsman was almost always less than the balls bowled (you definitely remember to put the dot in the bowler's analysis but it's easy to forget the dot for the batsmen). This suggests bad scoring gives negative discrepancies. Looking at the data there are more negative than positive.
Anything else you can think of that causes it or a better way to test which problems came from where?
I'm sure you're right that there are plenty of transcription errors. But others...? To take the examples from Davis's blog: how do you get from 283 to 314, 285 to 269, and 130 to 188?
Davis uploaded a scanned copy of the bowling figures from one of the 1990/1 Tests here. It's barely legible, and there are all sorts of bizarre aspects to it which he describes in the 5 September 2010 post here.
Post a Comment
Davis uploaded a scanned copy of the bowling figures from one of the 1990/1 Tests here. It's barely legible, and there are all sorts of bizarre aspects to it which he describes in the 5 September 2010 post here.
Subscribe to Post Comments [Atom]
<< Home
Subscribe to Posts [Atom]
