Tuesday, October 30, 2007

The worst Test opening batsmen

I spent yesterday afternoon looking for inept batsmen who had nevertheless opened the batting for their country in Test cricket. Here's some of what I found. Some of it is boring but most of it's pretty interesting.

20. Howard Francis (South Africa). First-class average: 12.90.

Francis is one of the luckiest players to have ever played Test cricket, luckier even than Nathan Hauritz. A very occasional wicket-keeper, he was picked as a specialist batsman by South Africa for their two Tests against England in 1898/9. Why he was picked is not clear to me - he had made a score of 38 in the previous season for Western Province, but the only half-century (55) of his career had come whilst in England nearly five years earlier. He batted at 3 in both innings of the first Test, making 29 in the second innings. He was promoted to open the innings in the second Test but got out for 1. Dropped down to number five for the second innings, his score of 2 was actually not that bad in an innings where South Africa were skittled for 35.

He played only one more first-class match, against the touring Australians in 1902/3.

19. John Hartley (England). First-class average: 12.89.

Picked as a legspinner for England in two Tests against South Africa in 1906, he opened the batting in the second innings of his debut Test when Plum Warner partially reversed the batting order.

18. Harry Butt (England). First-class average: 12.83.

A wicket-keeper in the days when they weren't expected to bat, he was number eleven in three of his four Test innings. He opened the batting in the second innings of his debut Test as a nightwatchman. He was dismissed off the last ball of the day for a duck. South Africa were bowled out for 30 in their second innings.

Why George Lohmann opened the batting for England in their first innings is a mystery to me.

17. Goofy Lawrence (South Africa). First-class average: 12.63.

For no obvious reason, he was promoted from his usual spot of number nine to open the batting in the first innings of what proved to be his last Test, against New Zealand in 1961/2. He did well, top scoring with 43, before being banished back to number eight for the second innings.

16. Alfred Shaw (England). First-class average: 12.44.

Shaw was one of England's opening bowlers in the first Test series. In the second Test, he was promoted to open in what looks like a measure to protect the batsmen from a sticky. He was dismissed, along with regular opener Jupp, before stumps on day 1, to leave England 2/7.

15. Arthur Mailey (Australia). First-class average: 12.33.

Mailey is best remembered as Australia's best legspinner in the 1920's. All of his innings for Australia were at either number ten or number eleven, except in this match. With South Africa taking two innings to match Australia's first innings score, Herbie Collins sent out Mailey and wicket-keeper Sammy Collins to knock off the run. Mailey did so off his first ball.

14. Edwin Evans (Australia). First-class average: 12.26.

England made 434 in their first innings in the third Test in 1886. But perhaps then it started to rain? Whatever the cause, Australia were all out for a boring 68 in 60.2 four-ball overs in reply. Following on, Evans was sent in to open the batting as a nightwatchman. He and McIlwraith batted out the day, but Australia lost by an innings and 217 anyway.

13. Ciss Parkin (England). First-class average: 11.77.

England had forty minutes to bat out on the last day of this dull draw in the 1921 Ashes series. Parkin got to open, where he feasted on the non-regular bowlers, scoring 23 in 27 minutes.

12. Azeem Hafeez (Pakistan). First-class average: 11.39.

Another case of teams having fun to end a drawn match. Against such bowlers as Dilip Vengsarkar, Azeem scored 18.

11. Ian Meckiff (Australia). First-class average: 11.27.

Same deal. This time the draw was against India.

10. Danny Morrison (New Zealand). First-class average: 10.94.

Having been set a target of 2, John Wright sent out his number ten (Martin Snedden) and number eleven (Morrison) to open. They each scored a single.

9. Bert Strudwick (England). First-class average: 10.88.

England's wicket-keeper was sent out to open the batting as a nightwatchman in the third Test against South Africa in 1909/10. He saw out the day and even made five runs on day five. England reached their target of 221 with three wickets in hand.

8. Percy Hornibrook (Australia). First-class average: 10.77.

The fifth Test of the 1928/9 Ashes series was a dead rubber, England having won the previous four. On a flat track, England eventually set Australia a target of 286. Australia started their innings late on day five, and Jack Ryder sent out Bert Oldfield and Hornibrook (on debut) as nightwatchmen. They saw out the day and a good deal of day six, as they put on 51 runs in 97 minutes. Ryder and a young Don Bradman saw Australia through to victory, on day eight, by five wickets.

7. Harry Boyle (Australia). First-class average: 10.24.

In the Test in England in 1880, England batted first and, led by WG Grace's 152, they made 420. In reply, Australia's batsmen either got out early or couldn't go on after making a start. Boyle, batting at number eight, top-scored with an unbeaten 36 in a total of 149. England enforced the follow-on, and it looks like Billy Murdoch told Boyle to keep the pads on. Unfortunately, Boyle lasted only seven balls in the second innings, making 3.

Australia were headed for a big innings defeat before the ninth and tenth wicket partnerships, which yielded 52 and 88 runs respectively. England lost five wickets on their way to the target of 57.

6. Albert Powell (South Africa). First-class average: 9.86.

Albert Powell's only Test was the same one in which Howard Francis opened the batting (see above). That two batsmen of such ineptness opened the batting in one match makes this Test pretty special. But let's not be too harsh on Powell - his 11 was the highest score in South Africa's 35.

Powell also bowled sometimes, which makes his selection a little bit more fathomable than Francis's.

5. Johnny Hayes (New Zealand). First-class average: 9.54.

Another tame draw. Hayes plundered the likes of Denis Compton and Cyril Washbrook for 19 at better than one a ball.

4. Allen Hill (England). First-class average: 8.94.

In the first Test, Hill batted at 9 in the first innings, scoring a quick 35 not out. In the second innings, Lillywhite jumbled the batting order, letting Hill open the innings. He made a second ball duck. But a 49 in the second Test helped his Test average to 50.5, well above his first-class average.

3. Ken Farnes (England). First-class average: 8.32.

The first Test against the West Indies in 1934/5 was played on a sticky. George Headley's 44 led the West Indies to 102 in their first innings. England replied with a fairly quick 7dec/81, led by Wally Hammond's 43. The Windies made it tit-for-tat declarations as they declared at 6/51, setting England a target of 73.

Bob Wyatt then ordered Farnes and Jim Smith (another tailender) to open the batting for the run chase. Errol Holmes, a competent lower-order bat, batted at 3. One can't say for sure whether the tactic was effective, but England won, scoring the required runs at four and a half per over.

2. Albert Rose-Innes (South Africa). First-class average: 7.77.

Rose-Innes played his two Tests in the series against England in 1888/9. Charles Davis describes the second Test as being the match the least worthy of Test status of all matches now regarded as Tests. Apparently Rose-Innes could bat well at club level, but he averaged just 3.5 in his four innings opening for South Africa, and less than eight in his seven first-class matches.

1. Chuck Fleetwood-Smith (Australia). First-class average: 7.34.

This match has gone into folklore in Australia, but perhaps there are some of you unfamiliar with it.

Australia batted first and made 9dec/200. The innings finished on day two; England were caught on a sticky and made 9dec/76, the declaration coming just before stumps. Bradman sent Bill O'Reilly and Fleetwood-Smith to open. O'Reilly got out caught and bowled to Voce first ball. Frank Ward was sent in as nightwatchman, and he and Fleetwood-Smith survived the rest of the day.

The next day was a rest day, and a day of dry weather was good for the pitch. Fleetwood-Smith promptly got out when play resumed on Monday, and then the regular batsmen went in. Bradman batted at seven and made 270, a record for a number seven which still stands. He added 346 for the sixth wicket with Jack Fingleton, another record that still stands.

Chasing an unlikely 689 for victory, England made only 323.

And for those of you who prefer big-dataset statistical analyses rather than anecdotes, I'll get around to it.

Monday, October 29, 2007

A list of all-rounders

After the embarrrassment that was never having heard of Aubrey Faulkner, I decided to rank all players by (batting average - bowling average), and see what turned up. I put no qualifications on, so that I could look for any interesting stories. It's no surprise who comes in at number one, but number two was new to me.

The number indicates the player's rank.

2. Arthur Hill: He played three Tests for England in South Africa in 1895/6. He did well with the bat, averaging 62.75, and then took 4/8 to wrap up South Africa's second innings in the third Test, giving England an innings victory. His only spell of bowling in Test cricket went 8-4-8-4, giving him a bowling average of 2. (Edit added much later: what makes that spell more remarkable was that he was the wicket-keeper, and didn't bother taking his pads off to bowl.)

7. Mick Malone (!): In his one Test before joining World Series Cricket, he made 46 and took 6 wickets, including 5/63 in the first innings. His bowling average finished at 12.83. He was not usually such a good batsman - that 46 was his highest first-class score.

8. Ken Hough: He played two Tests for New Zealand against England in 1958/9. In his first Test innings, coming in at number eleven, he slogged 31 runs in 28 balls without being dismissed, dominating the 40-run tenth wicket stand. Unfortunately for him, this only brought NZ to 142, in reply to England's 374. Hough finished his short Test career with 62 runs for once out, and six wickets at 29.16. I don't know why he didn't play more Tests - he averaged less than 21 with the ball in first-class cricket. (Edit: Thanks to Fiery of the Cricket Fans' Forum, who tells me that Hough also played international soccer.)

10. Allen Hill: Hill was the outstanding all-rounder of the first Test series, scoring 101 runs at 50.5 and taking 7 wickets at 18.57. In the first Test, he batted at 9 and made 35 not out. In what is reminiscent of some club matches, he was promoted to open the batting in the second innings, where he was promptly dismissed second ball. His batting was not usually as good as it was in this series - his first class average was 8.94. This must surely be one of the lowest averages for an opening batsman in Test history. (I'll do a thorough search later for this record. The lowest I've been able to find is Chuck Fleetwood-Smith, whose first-class average was 7.34.)

13. Anthony McGrath (!): I was shocked to see this name here: 40.2 and 14 are his averages. But they're inflated by two of his four Tests coming against Zimbabwe.

16. Jacques Kallis

17. Basil Butcher: He was a solid middle-order bat for the West Indies from 1958 to 1969. In 44 Tests, he bowled in only six innings. In one of those, he took 5/34. Batting average: 43.11; bowling average: 18.

18. Mark Boucher: Yes, he's taken a Test wicket. He got his wicket (for 6 runs) in this tame draw.

22. Garry Sobers

23. Albert Trott: He played five Tests for Australia in the 1890's. He didn't get dismissed until his fourth innings in Test cricket, and when he did get out his batting average was 205, and his bowling average was 6.5, him having taken 8/43 on debut. For some reason, he was overlooked for the next England tour, but he came back in South Africa for his two last Tests, doing very little with the bat but taking lots of cheap wickets. 228 runs at 38; 26 wickets at 15.

34. Wally Hammond: We all know that he was one of the greatest batsmen in history, but he was also a useful partnership breaker. In 85 Tests, he took 83 wickest at 37.8. Not bad when it goes with a batting average of 58.45.

35. Alan Fairfax: His career was a lot shorter than I thought it was. He played just 10 Tests for Australia, being extremely consistent with the bat - his lowest score was 9, his highest 65, and his average was 51.25. (His first class average was only 29.) He also took 21 wickets at 30.71.

37. Doug Walters: He didn't take many wickets, but his averages get him onto this list. In 74 Tests, he averaged 48.26 with the bat, but he was a handy part-time bowler, taking 49 wickets at just 29.08.

47. Stanley Jackson: He played 20 Tests for England around the turn of the century. He scored 1415 runs at 48.79, and took 24 wickets at 33.83.

48. Bob Cowper: I didn't know he bowled. He took 36 wickets in 27 Tests at an average of 31.63, to go with the batting average of 46.84 and the first Test triple-century in Australia.

49. Zulfiqar Ahmed: He started his career as a specialist batsman, but he took 11 wickets in his sixth Test (albeit against New Zealand in 1955/6). Three of his nine Tests were against NZ, which helped his stats a bit. 200 runs at 33.33; 20 wickets at 18.3.

52. Imran Khan

54. Allan Steel: He was the outstanding all-rounder for England in the 1880's, out-performing WG Grace at Test level, though his batting was weaker than Grace's at county level. He played 13 Tests, scored 600 runs at 35.29, and took 29 wickets at 20.86.

56. Charles Macartney: As well as being a great attacking batsman (2131 runs at 41.78), he was a useful, if somewhat occasional, left-arm finger spinner, taking 45 wickets in his 35 Tests at an average of 27.55. His best bowling came in the 1909 Ashes series, in which he took 11 wickets in the third Test.

57. Aubrey Faulkner: 25 Tests, 1754 runs at 40.79; 82 wickets at 26.58.

59. Keith Miller

61. Steve Waugh

67. Ted Dexter: He's another one I didn't know bowled. 62 Tests, 4502 runs at 47.89; 66 wickets at 34.93.

73. Eddie Barlow: His career was cut short by Apartheid, but he still managed 30 Tests, scoring 2516 runs at 45.74, and taking 40 wickets at 34.05.

77. Billy Bates: Another great all-rounder for England in the 1880's, his career ended when a straight drive hit him in the eye. 15 Tests, 656 runs at 27.33; 50 wickets with his off-spin at 16.42.

79. Frank Worrell: His record with the ball wasn't great, but he did open the bowling at times for the West Indies, as Stuart informs us in his profile on Roy Gilchrist. 51 Tests, 3860 runs at 49.48; 69 wickets at 38.72.

80. Asif Iqbal: His Test career came in two parts - he played his first few Test series as a bowling all-rounder, before concentrating on his batting for the latter part. 58 Tests, 3575 runs at 38.85; 53 wickets at 28.33.

Friday, October 19, 2007

The concept of form in cricket has predictive value, though its sign varies with the batsman.

One of Martin's suggested blog topics has been abandoned, taken up by me instead. He wondered, for instance, "If Ricky Ponting has averaged 80 in his last 5 innings does this mean he is more likely to score above his career average in the next innings?"

The answer in this specific case is "Yes", but the overall picture is much cloudier.

To study this problem I took the innings by innings list of various batsmen's Test careers, and modified it so that each "innings" actually meant the number of runs between dismissals. (I couldn't be bothered making my code airtight; if a batsman ended his career with a not out, that innings doesn't count). I then plotted the value of each innings (from the sixth innings onwards) against the average from the previous 5.

Below the cut are the resulting scatterplots, with three curves drawn on them. The blue curve is a least squares linear fit to the data. (Yes, I know that cricket innings aren't remotely nice as far as statistical distributions go, but it gives some indication nonetheless.) The red curve is the expected least squares fit for the batsman's innings in random order. I did this with a bit of Monte Carlo: shuffle the innings a hundred times, do a least squares fit for each one and take the average. The black curve is perhaps the most useful one.

Define 'lead-up form' as the average over the last five innings. The y-value of the black curve is then the average of all innings such that the lead-up form is greater than the x-value.

So it tells you what Ponting averages in his next innings, given that the average over the previous five was greater than, e.g., 80.

























I'll restrict myself to a few comments. Justin Langer got most of his big scores after mediocre runs. The big drop-off at the high end of Greg Blewett's graph shows how inconsistent he was. When Greg Chappell's form was dynamite, it was really dynamite. When Bradman had a rut for five innings, England must have been very, very worried. Conversely, if he'd made 750 runs in his last five dismissals, they would have been confident that he was due for a low score.

Victor Trumper's is the most surprising for me.

Now that I've got the code working, it only takes a minute or two to make one of these graphs. So feel free to make requests, especially for players where the notion of form might be interesting.

I will also try using three innings rather than five to measure form, when I summon the energy to do so.

Edit added later: If you take all batsmen who average over 40, and scale each batsman's innings so that all of them average 40, and throw all of the data points onto the graph, the slope of the regression curve becomes a rather flat-looking 0.01.

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]