Sportonomics

Statistics. Don't you just love them? Especially when you make them say something convenient for your news story...

Hello Observer Sport. Goodbye Steve McClaren, England manager. Well, maybe, if England don't qualify for Euro 08. In their big piece entitled "How lucky can you get?" the Observer ran a table in which they showed that:

McClaren and Don Revie are the only England managers to record a 24 per cent defeat record. The best record belongs to Terry Venables - 4 per cent, representing one defeat in 23 games.

	P	W	D	L	win%	loss%
Steve McClaren	17	9	4	4	52.9	23.5
Don Revie	29	14	8	7	48.3	24.1
Kevin Keegan	18	7	7	4	38.9	22.2
Walter Winterbottom	139	78	33	28	56.1	20.1
Bobby Robson	95	47	30	18	49.5	18.9
Graham Taylor	38	18	13	7	47.4	18.4
Rob Greenwood	55	33	12	10	60.0	18.2
Glenn Hoddle	28	17	6	5	60.7	17.9
Alf Ramsey	113	69	27	17	61.1	15.0
Sven-Goran Eriksson	67	40	17	10	59.7	14.9
Terry Venables	23	11	11	1	47.8	4.3

(note: table was rounded up with NO decimal points)

Gosh, that's terrible, isn't it? Sack him now.

Perhaps he will get the sack, but first, a few pointers.

1) Putting McCLaren top of the table was disingenuous. McClarens loss percentage is 23.5 - Revie's is 24.1. Rounding up and down is fair enough, but Revie should be top of that table. Shame on the Observer for that one.

2) Venables is McClaren's number 2 and has been mooted in some circles as a possible replacement. Suggesting he has the BEST record is very misleading. The best record? In terms of defeats, yes. But what was quietly gets ignored is that McClaren has a better WIN record than Venables: 53% to 48%.

In fact, in the win percentage stakes McClaren is a mid-table England manager.

	P	W	D	L	win%	loss%
Alf Ramsey	113	69	27	17	61.1	15.0
Glenn Hoddle	28	17	6	5	60.7	17.9
Rob Greenwood	55	33	12	10	60.0	18.2
Sven-Goran Eriksson	67	40	17	10	59.7	14.9
Walter Winterbottom	139	78	33	28	56.1	20.1
McClaren	17	9	4	4	52.9	23.5
Bobby Robson	95	47	30	18	49.5	18.9
Don Revie	29	14	8	7	48.3	24.1
Terry Venables	23	11	11	1	47.8	4.3
Graham Taylor	38	18	13	7	47.4	18.4
Kevin Keegan	18	7	7	4	38.9	22.2

And in terms of points per game (3 for a win, 1 for a draw), McClaren is again mid-table, just 0.02 behind.... Terry Venables.

	P	W	D	L	win%	loss%	Points per game
Alf Ramsey	113	69	27	17	61.1	15.0	2.07
Sven-Goran Eriksson	67	40	17	10	59.7	14.9	2.04
Glenn Hoddle	28	17	6	5	60.7	17.9	2.04
Rob Greenwood	55	33	12	10	60.0	18.2	2.02
Walter Winterbottom	139	78	33	28	56.1	20.1	1.92
Terry Venables	23	11	11	1	47.8	4.3	1.91
McClaren	17	9	4	4	52.9	23.5	1.82
Bobby Robson	95	47	30	18	49.5	18.9	1.80
Graham Taylor	38	18	13	7	47.4	18.4	1.76
Don Revie	29	14	8	7	48.3	24.1	1.72
Kevin Keegan	18	7	7	4	38.9	22.2	1.56

I happen to think Venables was a better manager than McClaren, but the Observer's table is not the way to prove it.

I'm fed up of the argument that English football is suffering due to all the foreign players. I've looked at some of the arguments on my other blog, but the figures used by the Sunday Times suggest that, post-Bosman, the England team is basically screwed. Or is it?

The figures they give are not explained fully - are they representative of squads, average teams or just taken from the first match of the season? (This first match is often quoted in articles on this subject. I wouldn't be surprised if it's the most convenient figure.)

And here they are. I have added the increase and percentages:

Foreign players	1994	2006	increase	% incr
England	141	316	175	124.1%
Germany	163	222	59	36.2%
France	144	178	34	23.6%
Italy	71	170	99	139.4%
Spain	86	191	105	122.1%

What does this show? Spain has had a similar increase to England in terms of percentage, but the national team is still a) full of world-class players b) without a trophy since 1962. Italy has had an increase of 99 players, roughly 4 per club. And they just won the World Cup, so no worries there then. Germany have seen a milder increase, but still have the second highest number (222 - equivalent to 12 foreigners per club which could be the entire first team), and have won Euro 96 as well as being runners up in the 2002 World Cup. It just doesn't add up.

But.... should English players try traveling abroad? Currently there are 10 English players playing in other leagues aside from Scotland. Now, let's take Brazil. The only figures I could find for Brazil suggested over 850 players abroad. France according to this article in Time has over 100 - and that was in 2002. Italy I couldn't find figures for, but it's certainly a lot more than England. When I find the numbers, I'll add them here.

But the point is the same - travel is linked with success. Overseas players are exposed to a different culture and style. And it's clearly an indicator of talent that exists in the first place, talent that is nurtured in academies (France) or on the streets (Brazil).

Plus, my esteemed colleague Tim Harford, the Undercover Economist, has a few thoughts on the subject. (The FT doesn't do sport that often, so I'd pay attention). Oh, and here's a another economist, Stephen King's analysis (not the horror writer).

The Fink Tank sounds awful, but actually it's an interesting idea. You can see the results in the Times, which proclaim: "Daniel Finkelstein has ranked every Premiership player using a mathematical and statistical model to end all the debate"

Well, I don't think it ends the debate, but we'll come to that later.

The model works by identifying "the relationship between goals scored and every kick of the ball made by every player for every club". I presume this takes into account defending as well as attacking. But it mention tackles, pressure that indirectly relates to goals, or other parts of football. A more detailed methodology is needed.

It penalises players for lack of time on the pitch - which is strange, as some players are used deliberately as a "super-sub" or impact player, and their lack of pitch-time is a virtue.

Here is the methodology:

The Fink Tank Predictor provides forecasts and ranking systems for English and European club football, based on a statistical model of matches based on more than five years of football scores.

In looking at player rankings for this season, the phrase “time-adjusted points” means the number of points the player would have added to an average team in the full season, compared with an average replacement. The points are then adjusted to reflect the amount of time spent on the pitch – minimum 400 minutes.

OK, so what do we get? Christiano Ronaldo is by far and away the best player, with 19.12 points, ahead of Frank Lampard on 16.24. Ronaldo was acclaimed for his performances this season, Lampard was not. Strange, but there you go.

There are 3 goalkeepers in the top 10. This is strange, as the methodology suggests that the Fink Tank is the relationship between goals and every kick of the ball. Were their long-ball goalkicks often converted into goals? Or do saves count? They certainly haven't been scoring, or putting in crosses like a winger.

To see if the Fink Tank points reveal anything strange, I have analysed each team's points, and averaged it per player - some teams had only 18 players, others had 24, so points per player seems fairest. I then related this points per player average to the Premiership table, to see how it related to final standings, points, goals scored and goal difference. Given that FinkTank points are given for influence in goals, you would expect some correlation with the goals scored column.

And what did we get? Well, the results are in this spreadsheet. To break it out:

Club	position	Points	Goals	Goal diff	total Fink points	Players	points per player
Manchester United	1	89	83	56	99.94	20	4.997
Arsenal	4	68	63	28	98.89	20	4.945
Chelsea	2	83	64	40	82.4	21	3.924
Liverpool	3	68	57	30	50.36	21	2.398
Tottenham Hotspur	5	60	57	3	35.02	21	1.668
Portsmouth	9	54	45	3	3.36	20	0.168
Blackburn Rovers	10	52	52	-2	-0.75	20	-0.038
Everton	6	58	52	16	-5.52	19	-0.291
Middlesbrough	12	46	44	-5	-5.31	18	-0.295
Newcastle United	13	43	38	-9	-12.36	21	-0.589
Aston Villa	11	50	43	2	-16.03	22	-0.729
Reading	8	55	52	5	-24.04	20	-1.202
Charlton Athletic	19	34	34	-26	-27.19	22	-1.236
West Ham United	15	41	35	-24	-29.87	24	-1.245
Manchester City	14	42	29	-15	-32.7	24	-1.363
Sheffield United	18	38	32	-23	-42.58	22	-1.935
Fulham	16	39	38	-22	-41.83	20	-2.092
Wigan Athletic	17	38	37	-22	-61.48	21	-2.928
Bolton Wanderers	7	56	47	-5	-72.26	18	-4.014
Watford	20	28	29	-30	-95.9	20	-4.795

Source: Times newspapers, Premier League

What looks weird here? Well, what the hell are Bolton doing near the bottom? Their goals scored is good, goal difference ok, points and position put them at 7th - yet their FinkTank points put them at next to last by some distance.

Bolton are the biggest anomaly, but there are others. Arsenal have a lot of FinkTank points, but seem to be underperforming in the league. the Fink Tank puts them second to ManU and a long way ahead of Chelsea.

I don't want to comment too much on individual players, but it seems very odd that Paul Robinson is ranked 402 - and that Hleb of Arsenal is in the top 10. I have also looked at the top goal scorers and their Fink Tank points, and the correlation is poor - only 0.41.

Teams however, seem to come out roughly correct. The correlation between FinkTank points per player per club and the 4 rankings in the Premiership table are good:

Correlation of points scored in Premiership to Finktank points per player per club 0.852

Correlation of Premiership table position to Finktank points per player per club -0.785

Correlation of goal difference to Finktank points per player per club 0.870

Correlation of goals scored to Finktank points per player per club 0.838

Correlation of Premiership points to position (for reference) -0.954

These are all strong correlations. We should expect to see table position to be a weaker correlation as the gap between teams is not shown in position on a list, and this is true. The Premiership points scored is a better guide.

Also, goal difference is closer to FinkTank points than goals scored - perhaps the methodology looks at negative play as well as positive.

Overall, the Fink Tank is a good predictor of how a season pans out. What it can't do is explain Bolton. Or, more pertinently, it seems to ignore the effectiveness that Bolton have as a team. It also inflates the position of Arsenal. Is this because Arsenal's build-up play is more intricate than other clubs, and Bolton play more route-one football? If the methodology of the Fink Tank was more open and detailed, perhaps these perceived truths could be proved one way or other once and for all.

Here are the charts to show the relationships in the table above.

Sven just won't go away. He has not been the manager of the England football team for the best part of a year now, yet the media circus goes on. There was his recent appearance on Inside Sport (May 21), where he claimed (as usual) that he was value for money. There is the astonishing fact that the FA has paid him £25m and he is still on a salary for doing zilch.

And then there was that Sunday Times magazine cover story from April 29. In which they dish little dirt, reveal a few "facts" to make him seem like a philanthropist (He apparently turned down over 40 appearances worth over £2m in fees), and set his record straight.

According to Sunday Times and their England manager's league table, "Even counting two draws that went to penalty shoot-outs as defeats, Eriksson was England's most successful coach of the modern era. The table only counts competitive matches, not friendlies"

Well, this is an interesting point of view. (By modern era, I presume they count all the managers in the table.) What exactly is success here? I have amended the table to include points and points per game, which work on the basis of 3 pts for a win and 1 for a draw, and I have moved two of Eriksson's draws into the lost column as the STimes suggests.

		Played	Won	Won%	Drew	Drew%	Lost	Lost%	Points	Points per game	Trophies

Eriksson	38	26	68.4	7	18.4	5	13.2	85	2.2		0

Greenwood	26	17	65.4	5	19.2	4	15.4	56	2.2		0

Robson		44	22	50.0	15	34.1	7	15.9	81	1.8		0

Taylor		19	8	42.1	8	42.1	3	15.8	32	1.7		0

McClaren	6	3	50.0	2	33.3	1	16.7	11	1.8		0

Hoddle		15	9	60.0	3	20.0	3	20.0	30	2.0		0

Revie		10	6	60.0	2	20.0	2	20.0	20	2.0		0

Winterbottom	29	15	51.7	8	27.6	6	20.7	53	1.8		0

Ramsey		33	20	60.6	6	18.2	7	21.2	66	2.0		1

Keegan		11	4	36.4	3	27.3	4	36.4	15	1.4		0

Venables	5	2	40.0	3	60.0	0	0.0	9	1.8		0

Wilkinson	1	0	0.0	1	100.0	0	0.0	1	1.0		0

(C) Times Newspapers Ltd, 2007

Eriksson's record is very similar to that of a premiership winning team. The exact same number of games - 38 - as in a season, and a similar number of wins and points as required to win the league. To compare, the recent winners of the premiership:

year	Team  		Pld  	W  	D  	L  	F  	A  	GD  	Pts

2007	Man United  	38  	28  	5  	5  	83  	27  	56 	89

2006	Chelsea  	38  	29  	4  	5  	72  	22  	50 	91

2005	Chelsea 	38	29	8	1	72	15	57	95

2004	Arsenal		38	26	12	0	73	26	47	90

2003	Man United  	38  	25  	8  	5  	74  	43  	40  	83

2002	Arsenal  	38  	26  	9  	3  	79  	36  	43 	87

2001	Man United  	38  	24  	8  	6  	79  	31  	48  	80

Eriksson's England fits right in there.

So that's all good. Except it isn't. Look at the table again. Eriksson has the same points per game as Greenwood, with Hoddle, Revie and Ramsey not far behind on 2.0, so this points idea is a bit misleading. It was my way of trying to help out the Times with a bit of proper data analysis. But this is all a giant red herring.

International football is not about points, or tables. It's about competitions and trophies. And in this regard, Eriksson has 3 quarter-finals and that's it. The key column is Trophies - and only one manager has anything in there: Ramsay and the '66 World Cup.

If semis "count" towards success, Robson is next with Italia 90, along with the European championship teams of '96 (Venables) and '68 (Ramsey again). And then there are a bunch of quarter-finals, of which Eriksson has three.

Success can be measured in many ways. Eriksson had an excellent record in competitive matches, right up until the point that his team lost. And in that sense, they were failures. The players were good enough to win big cups, and he didn't deliver. Arguing that his win-loss record puts him at the top is an interesting diversion, but that's all. Reputations and records are measured in cups, and Sven, like everyone except Ramsey, has a zero in that column.

Before I totally forgot, I just wanted to look at two of the most interesting sporting statistical stories of the last few months. Both were different but excellent examples of what stats should be used for - exploding sporting myths.

The first was in Prospect magazine of February, and looked back at the career of Shane Warne. Aside from the widely publicised stat about him being the highest run scorer in test cricket without a century, it made an interesting case for both Glenn McGrath and Muttiah Muralitharan being considered better bowlers. True, Warne got more test wickets (with 708, still the record), but the other two both have over 500 and are near the top in terms of strike rate and runs per wicket, which Warne is not.

Does Warne have a defence? Hard to say. Against Murali, a Warne advocate could point to the fact that he travels much better, with more wickets (43), a better average, economy and strike rate, and more 5-fors in only 4 more tests away from home. Murali, in contrast, has a much better record at home in all areas of his bowling, benefiting from pitches that are prepared for him.

More damningly, Murali has 137 wickets against Bangladesh and Zimbabwe, the two weakest teams in test cricket, and an average below 16. Warne has played these two only 3 times, with 17 wickets at around 25. This in effect inflates Murali's record by about 20%, and shows that his record is not all it seems.

Glenn McGrath is a harder comparison to Warne. He enjoyed opening for Australia pretty much through out his career, using the new ball for a side that generally trampled on the opposition, and had an outstanding career. Having been in the same side for most of their tests, it's hard to see who has had the bigger impact. For what it is worth, both bowlers have suffered when the other has not been in the team, suggesting they bowled best in tandem. In the 104 tests they played together, their stats are:

		Warne	McGrath

Tests		104	104

Wickets		513	488

Average		24.9	21.3

Win%		68%	68%

In the tests played without the other, the stats are less good for both:

		Warne	McGrath

Tests		41	20

Wickets		195	75

Average		26.8	23.3

Win%		51%	65%

Overall, Warne is a higher wicket taker but with a poorer average. But without McGrath and with Warne, Australia's win percentage goes from just under 70 to just over 50. Perhaps McGrath is more important after all.

Rugby myths

The second was a Sunday Times piece from January 28 on Rugby that analysed data from the 2005 and 2006 Six Nations, and found some facts that completely go against the current wisdom of how to play. The most interesting were:

Possession is everything No, it isn't. In the 2006 Six Nations, England dominated the battle for possession - indeed, in two of their five games they won 80% more than their opponents - but they ended a lowly fourth

Ball retention is crucial No, wrong too. In 2005 and 2006, 83% of tries took three phases of play or fewer. The longer you kept the ball, the less likely you were to score

Goalkicking wins matches It can, but no Six Nations match in 2004 or 2005 was lost by the team scoring most tries. Tries are still vital.

Yellow cards spell disaster Nope. Most teams reduced to 14 hold up well. Scores made during sin-bin periods rarely turn a match from the course it was already on

Some Six Nations teams are outstanding at forcing turnovers Not really. Last season every team retained between 90% and 94% of the ball they took into contact

The drop goal is a means of easy points No. In recent years, fewer than one attempt in three has succeeded

Source: IRB Game Analysis Centre

What this should mean is that coaches should teach a fundamentally different approach to playing the game. Firstly, it's not about winning the ball, just doing the best with it when it comes your way. Secondly, if you hit the fourth phase, kick for territory - you have little chance of scoring. Thirdly, infringe all you like - the sin bin doesn't hurt you that much. Fourthly, only drop for goal when you are in range and it's the fourth or higher phase.

Unlikely, but there you go.

Football wages are an eternal source of fascination and disgust in equal measure. I have no idea why they are always quoted as a weekly wage rather than per year, as is the case with almost every other profession. But aside from that anomaly, last Sunday it was interesting to compare the lead story of the sports sections of the Sunday Times and the Observer. Depending which you read, you would come away with very different impressions of football salaries. Here's the Sunday Times:

Terry demands £60m deal

JOHN TERRY’S contract negotiations with Chelsea have broken down over the England captain’s demand to be the best-paid player at the club for the next nine years. At current rates, the deal would be worth a minimum of £58.2m, making it the richest in British sporting history, but with new signings at the world’s biggest-spending club it would inevitably rise.

The "limitless parity" clause would guarantee that Terry was the highest-paid player at the club until beyond his 35th birthday. While Chelsea were prepared to increase his wages to the club’s current ceiling of £121,000 a week, which is paid to Michael Ballack and Andriy Shevchenko, they could not accept the liability of promising the defender equivalence with the club’s best-remunerated player for the best part of the next decade.

Any way you cut it, there's a big deal going down. If Chelsea's offer of £30m+ over 5 years is too small, Terry may well walk, but this is the reality of the sums involved. Lest we forget, Beckham is getting up to £25m a year with bonus, and about £5m of that is salary.

And then I picked up another Sunday paper. In the Observer, it's a different story:

Premiership's top clubs set £100,000 limit on wages

Frank Lampard and Cristiano Ronaldo have 'no chance' of receiving the massive pay increases they are demanding from Chelsea and Manchester United. That is what they will be told this week by the chief executives of their clubs... the decisions signal a sea change in attitudes to players' wages, which, in some cases, have spiralled to more than £100,000 a week.

What is really going on? Football wages have exceeded inflation for years, and it's not due to the usual factors that you would expect in an industry. There is no lack of people able to do the jobs. There is an increase in labour liquidity, given the Bosman ruling and the EU transfer laws. Plus, players are more willing than ever to work abroad. These factors should if anything, depress wages or keep them in line with inflation.

But, as the Premiership wage study in the Independent shows, this is not the case.

Wages are up 65% from 2000. There is also an interesting disparity between what forwards earn and other positions, with defenders on £653k to the forwards on £806k. John Terry's demands seem even more outrageous given his defender peers get on average nearly 20% less than their striker counterparts.

But these sportsmen are not being incentivised correctly. What other group of high earners gets similar coverage? Finance professionals in the City. And how do they get paid? With low salaries and huge bonuses. Bonuses are measured in performance - great in a bull market like the one of the last few years, but measurable all the same.

Why football should follow suit. Players should have their salaries cut to a basic level, and then big bonuses for performance and results. There is no shortage of stats from Opta on how players perform, which could allow outstanding players in poor teams benefit. And win bonuses could easily be used, with levels going up for league position and rounds-progressed in tournaments. Win bonuses exist now, it's just that they are paltry compared to salary.

The current structure encourages players to play well only when their contract is up for renewal. Once signed, they can cruise, knowing that getting dropped is the only big damage that they can suffer to their reputation - and many club's rotation policy negates that effect anyway.

So are they going to be salary capped? In reality, no. The problem is that it's too late. Any club that tries to impose a new wage structure could easily see their players walk out or get bought out. It would only work if all clubs were to impose it uniformly - unlikely given the current situation. Changing wage structures in any industry is hard enough, but in football, where the structures to impose industry-wide changes are even weaker than usual, it would be nigh-on impossible.

Wimbledon have announced that this year, the men and women will get equal prize money. The predictable responses covered the usual arguments about equality, entertainment and fairness. As Pat Cash noted in the Sunday Times:

"Men need to be far fitter, work harder in their preparation, compete with much more intensity once they get out on the match court, show greater powers of endurance and play many more tough matches... Here’s the deal in everyday terms. What would a man think if he worked in an office next to a woman and she did 40% less work than he did and left a good hour earlier every day, but went home at the end of each week with the same amount of money in her pay packet?"

An interesting viewpoint. But then again, the 2005 Wimbledon finals don't bear this out. The women's match lasted longer, had more games and points, and was far more entertaining.

Roger Federer beat Andy Roddick 6-2 7-6 6-4 - 1hr, 41min
Venus Williams beat Lindsay Davenport 4-6 7-6 9-7 - 2hr, 45min

Even Pat Cash would be hard pushed to deny that Venus Williams deserved an equal paycheck that day.

Jon Wertheim notes in his excellent column:

While on balance, I applaud this decision, I find the WTA's gloating a bit disingenuous. Check out the prize money from last year and you'll see that the year-end ATP Masters Cup pays nearly 50 percent more than the analogous WTA Championship. The typical Masters Series event pays nearly double the WTA Tier 1. If I'm a WTA exec, I'm not sure how I respond to the question, "How can you clamour for equal prize money when the market consistently suggests your product is worth less?"

Well, the market may not be an even playing field. The WTA may have a far less energetic marketing department than the ATP. The men's tour may be in bigger stadiums and attract better sponsors due to other factors other than the quality of the tennis - poor management, or sexism, for instance. This is hard to measure, although I suspect Wertheim's analysis is correct.

Given that the men's and women's tours are separate, we need to find tournaments where they play together for a better comparison - the Grand Slams. The marketing, stadiums and other factors are effectively the same for men and women. We can then look at the attendance. If Women's tennis is so poor, the crowds will stay away.

Some commentators have pointed to TV viewing figures to show that men get up to 20% more viewers. Again, although this is an interesting indicator, it doesn't allow for external factors - e.g. what else was on TV at the time, such as football, and doesn't allow for average higher viewing figures on different days of the week. Why should Saturday and Sunday TV figures be level?

In theory, the ground attendance figures should tell us nothing. Demand for tennis tickets exceeds supply, so the stadiums should be full for both finals. However, fans will go to the ground on the chance of a return ticket or to watch on the big screen, and to watch the doubles and juniors.

The French, US and Australian events stagger the draw, so that men and women play on the same day except for the finals. This makes a comparison difficult, as there are no figures for specific matches on one day, just overall ground entrance. You can't tell if the stadium empties for the women OR the men.

The data for the finals of the Australian Open over the last 18 years is available, and the men have a much bigger attendance: on average, 1,278 more bums on seats. The highest was 2005, with 3,973 more spectators. This doesn't look good for the women.

But Wimbledon does something different, which allows for an interesting comparison. It alternates the draw from the Tuesday of the second week onwards. The Quarters, Semis and Finals for the women are on Tuesday, Thursday and Saturday. The men play their equivalent rounds on Wednesday, Friday, Sunday. If we look at the attendance figures we should in theory see a surge in numbers for the days the men play. And helpfully, Wimbledon's daily attendance figures for the last 6 years are online, including days affected by rain.

The only other factor is that over time, there are fewer matches such as doubles and the juniors on the outside courts, which should show a gradual tail-off in crowds during the 2 weeks.

If women's tennis is such a poor product, the data for the quarters and semis doesn't add up. There are on average 758 MORE people for the women's quarters, and 1,370 MORE for the women's semis. But the finals show a 1,000+ advantage to the men. (Luckily, the number of rain affected days was evenly spread over the days.)

                       2001    2002    2003    2004   2005     2006   Total average

Wednesday - Tuesday  -1,406  -2,081   1,215    -338    643   -2,583     -758

Friday - Thursday    -1,307  -5,544    -365  -1,150    956     -858   -1,378

Sunday - Saturday     1,545   1,905   2,276   1,172  1,669   -1,594    1,162

What does this mean? Crowds are just as keen to watch a combination of women's matches on a given day, but if there is just one - the final - they are less enthused. Why might this be?

Perhaps one women's match is too high a risk of being over in quick time; or with two matches there is more chance of seeing an entertaining match, whereas the men are a safer bet? If so, then women's tennis is indeed a poorer product, but only in isolation - more matches means more chance of good play.

Interestingly, the attendance figures and prize money gender differential for the final have been incredibly close for the last few years - both at around 5%. The women should, if anything, get more than the men for the earlier rounds but less for the final. This contradicts the perceived wisdom that the men's tour has greater depth than the women.

In which case, perhaps the women should play a five-set final. Then the attendance on the last 2 days would be a better comparison, and we could leave the equal pay dispute alone for ever.

The big cricket selectorial dilemma that faced England before the Ashes was whether to pick Ashley Giles or Monty Panesar.

The merits of the two spinners was based on a couple of premises. One was that England needed Giles to bat at number 8. The other was that Panesar might take you more wickets, but he can't bat.

This is not to criticise Giles, who has flown home from the tour to be with his wife who is seriously ill. But the stats show that this should never have been a debate.

Giles is a run-of-the-mill performer in test cricket. His 143 wickets in 54 matches is a slow rate. His bowling average is poor, at 40.6. And his batting? An average of just over 20 runs is not exactly earth shattering. If he was an all-rounder, it would have to be in the 30s to justify a place.

And Panesar? While his batting average of 11 is low, he will improve. Giles isn't going anywhere with his batting average. But Panesar's bowling is much better: 40 wickets in 11 matches at an average of 32.

Their strike rates are revealing too: Panesar takes a wicket every 69 balls, Giles every 85.

So what do England do? They pick Giles, which on average sacrifices taking wicket per match for adding 9 runs per innings. If England selectors think that each opposition wicket is worth just 18 runs, then they have another thing coming. This year, an opposition wicket has cost England on average 36 runs (7551 runs conceded, 209 wickets taken).

But is Giles a "big match player" as the England selectors have suggested? No more than Panesar. His batting has been solid when required, but he has taken 5 wickets in an innings just five times, to Panesar's three - in 43 fewer tests. He has four 50s to his name - but his highest score is 59, so these have not been converted into big scores. Scoring "important" runs or taking "crucial" wickets is hard to measure, but Giles isn't doing it consistently, otherwise his averages would be better.

One stat where Giles is the equal of Panesar is in who they dismiss. Only 25% of both men's total wickets are bowlers - three quarters are batsmen, keepers or all-rounders.

Their overall statistics are:

        Mat  Runs  HS   BatAv  100  50  W    BB    BowlAv 5w Ct St

Giles   54   1421  59   20.89   0   4   143  5/57  40.60  5  33  0

Panesar 11   68    26   11.33   0   0   40   5/72  31.85  3   2  0

An interesting discussion appeared on the John Wertheim tennis mailbag on the Sports Illustrated site. How long do people take to win a second slam? The theory goes that people win a slam, then have a bit of a dip before they get to win another. After that, it should get easier, at least for a bit, until time and injuries and competition catch up.

The SI mailbag suggested that it is about 10 tournaments between first and second slam.

Does the data back this up? Looking back to the start of the open era (1968 onwards), and counting back-to-back tournament wins as 1, the average time it takes a player to win their second slam is 5.47 tournaments - just over one year. This is the equivalent of winning Wimbledon, and netting your second slam at the US Open the following year - a lot less than the 10 events mentioned on SI.

The stats for men and women are very similar - the "event gap" between first and second is 5.37 for men, 5.60 for women. And the trends are also mirrored, with the gap decreasing across the first few wins. After win 6 or so, the data becomes a bit too random to be meaningful.

The blips in all this are Arthur Ashe, Virginia Wade and Hana Mandlikova, who all won their second slam within the normal timeframe, but took a lot longer to win their third. Their stats are:

Arthur Ashe: slam #1 to slam #2: 5 events. Slam 2-3: 22
Virginia Wade: 1-2: 13; 2-3: 22
Hana Mandlikova: 1-2: 5; 2-3: 18

Taking these three out makes a marked difference to the figures, with the combined gap between wins 2 and 3 dropping to 3.5 from 5.

The men and women also have a very similar chance of being a one-slam wonder: 46% for men, just under 43% for women.

Here is the overall table for open era "slam gaps" for men and women up to 10 grand slam wins:

Slam victory interval	1-2	2-3	3-4	4-5	5-6	6-7	7-8	8-9	9-10
Number of players	47	34	28	23	18	16	12	9	6
Average event gap	5.47	5.03	3.79	3.96	3.67	3.31	3.08	3.67	2.50

And you can download and view the entire spreadsheet here: robminto_slamgaps.xls

	Country	Club	Type	City	Stadium	Capacity	Built	Seats
1	ESP	FC Barcelona	Soco	Barcelona	L'Estadi (Camp Nou)	98 934	1957	all-seater
2	USA	UCLA Bruins	AF	Pasadena	Rose Bowl Stadium	92 542	1922	all-seater
3	USA	LSU Tigers	AF	Baton Rouge	Tiger Stadium	92 400	1924	all-seater
4	USA	Alabama Crimson Tide	AF	Tuscaloosa	Bryant Denny Stadium	92 158	1929	all-seater
5	USA	Georgia Bulldogs	AF	Athens	Sanford Stadium	92 058	1929	all-seater
6	USA	USC Trojans	AF	Los Angeles	L.A. Memorial Coliseum	92 000	1923	all-seater
7	USA	Washington Redskins	AF	Landover	FedEx Field	91 665	1997	all-seater
8	IND	Bengal Cricket	Cric	Calcutta	Ranji Stadium	90 000	-	all-seater
9	ZAF	Orlando Pirates	Soc	Johannesburg	FNB Stadium	90 000	1989	all-seater
10	USA	Florida Gators	AF	Gainesville	Ben Hill Griffin Stadium	88 548	1929	all-seater
11	USA	Auburn Tigers	AF	Auburn	Jordan Hare Stadium	87 451	1939	all-seater
12	ITA	AC Milan/Inter	Soc	Milano	Stadio Giuseppe Meazza	85 700	1927	all-seater
13	ECU	Barcelona SC	Soc	Guayaquil	Isidro Romero Carbo	85 000	1986	all-seater
14	RUS	Spartak Moskva	Soc	Moskva	Luzhniki Stadion	84 745	1956	all-seater
15	AUS	National Stadium	Mix	Sydney	Telstra Stadium	83 500	1999	all-seater
16	UKR	FK Dynamo Kyiv	Soc	Kiev	Stadion NSK Olimpiyskiy	83 160	1948	all-seater
17	USA	Texas A&M Aggies	AF	College Station	Kyle Field	82 600	1927	all-seater
18	IRL	GAA-use	GAA	Dublin	Croke Park	82 500	2002	69 500
19	ITA	AS Roma/SS Lazio	Soc	Roma	Stadio Olimpico	82 307	1952	all-seater
20	USA	Florida State Seminoles	AF	Tallahassee	Doak Campbell Stadium	82 300	1950	all-seater
21	USA	Oklahoma Sooners	AF	Norman	Gaylord Family Stadium	82 112	1923	all-seater
22	BRA	Clube Atlético Mineiro	Soc	Belo Horizonte	Estádio Magalhães Pinto	81 987	1965	59 042
23	TUR	National Stadium	Mix	Istanbul	Atatürk Olimpiyat Stadi	81 653	2001	all-seater
24	USA	Clemson Tigers	AF	Clemson	Memorial Stadium	81 473	1942	all-seater
25	USA	Wisconsin Badgers	AF	Madison	Camp Randall Stadium	81 321	1917	all-seater
26	DEU	Borussia Dortmund	Soc	Dortmund	Signal Iduna Park	81 264	1974	53 675
27	USA	Nebraska Cornhuskers	AF	Lincoln	Memorial Stadium	81 067	1923	all-seater
28	USA	Notre Dame Fighting Irish	AF	South Bend	Notre Dame Stadium	80 795	1930	all-seater
29	USA	UAB Blazers	AF	Birmingham	Legion Field Stadium	80 601	1926	all-seater
30	ESP	Real Madrid CF	Soc	Madrid	Santiago Bernabeu	80 354	1947	all-seater
31	USA	S. Carolina Gamecocks	AF	Columbia	Williams Brice Stadium	80 250	1934	all-seater
32	USA	New York Giants/Jets	AF	E. Rutherford	Giants Stadium	80 242	1976	all-seater
33	PER	Universitario	Soc	Lima	Teodoro Fernández	80 093	2000	60 000
34	USA	Texas Longhorns	AF	Austin	Texas Memorial Stadium	80 082	1924	all-seater
35	CHN	Multi-use	Mix	Guangzhou	Guangdong Stadium	80 012	2001	all-seater
36	COD	National Stadium	Mix	Kinshasa	Stade des Martyrs	80 000	-	-
37	LBY	Al-Ittihad Tripoli	Soc	Tripoli	11 June Stadium	80 000	-	-
38	CHN	Shanghai United	Soc	Shanghai	Shanghai Stadium	80 000	1997	all-seater
39	BRA	São Paulo FC	Soc	São Paulo	Cícero Pompeu de Toledo	80 000	1960	-

Source: worldstadiums.com

Much has been made of the superior effort of the European team in Ryder Cups over their higher-ranked and more illustrious American counterparts. The theory goes that Europeans are more suited to the team game, and the solipsistic yanks are just not team players - the stats back this up, as Gavyn Davies shows. Tiger Woods in particular is often cited as having a very poor Ryder Cup attitude and record when stacked next to his 12 majors and umpteen titles.

So far, so true. But a couple of things look out of place here.

Tiger Woods may have performed badly, but his record in Ryder Cup play is as follows:

Cups played: 4
matches: 20
Overall W-L-H: 7-11-2

This is a scoring record of 40% - not good for the outstanding golfer of his generation.

However, the US team in the four Cups in which Woods has taken part have scored 13.5, 14.5, 12.5 and 9.5 points, out of a possible 28x4=112 - a return of 44%. So Wood's contribution is not that much worse than the rest of the team. It's only a 4% difference.

If Woods had played 4 percent better, he would have won only one more match overall in Ryder Cup play - enough to have swung the 1997 cup to the US away from Europe if, for example, he had beaten Constantino Rocca in the singles, but not enough to make a difference in any other cup.

So - Woods may be poor, but the rest of the team is not doing much better. Given that since 1997 only TWO majors have been won by Europeans (the 1999 Masters by José María Olazábal and Open by Paul Lawrie) and another 13 Americans have won Majors aside from Woods, Tiger should not be the scapegoat.

Stats are one of the beautiful things of sport. Win loss records, goals per game, averages, totals. Then every so often, a statistical measure gets taken up that is misleading to such an extent that I find it amazing it isn't scrapped. In particular - break point conversions in tennis.

The stats from the Federer-Roddick US Open final show why Roger is so good. He hit 69 winners to only 19 unforced errors - that's a superb reflection of his quality of play. But why oh why do tennis statisticians include break point conversions? Let's think about this...

Tennis is all about winning games in the set. It doesn't matter if you win a game to love, or after 10 deuces. You get the game on the board. That's why some players lose despite winning more overall points (for example, Marat Safin vs Tommy Haas)

So, if you have a 0-40 on your opponent's serve, and break on the third go, the missed break points are irrelevant. Why do we count them? Roddick was 2 of 8 in the US Open final, and Roger 6 of 16. But the interesting statistic is when a player has break points and loses the game. If they win the game on their 5th break point, who cares?

If we look into the point-by-point stats of the match, Federer's 10 missed break points (BPs) were distributed across 6 games, with 4 of those games still won by Federer and 2 that "got away", (one with 5 missed BPs). So, in reality, although he "missed" 10 BPs, only 6 were opportunities that mattered in only 2 games.

Roddick had 6 missed BPs, which were in 3 games that he eventually lost. So Federer missed fewer game opportunities (2) than Roddick (3). That's where the match was won and lost. Tennis analysts should scrap the break points, and look at the games that got away.

The current cricket farce between England and Pakistan about ball tampering has highlighted some rather peculiar rules in the game.

The one that really interested me was that the Pakistan team were "docked" 5 runs at the initial stage when the umpires decided the ball had been tampered with. 5 runs is a hopeless punishment as a deterrent, and the numbers back this up. In 2005, there were 48,637 runs scored in test cricket - and an average innings of just over 291. This means that a 5 run penalty is effectively a 1.7% fine per innings, and therefore only 0.43% of the average match - which is nothing. In the case of the game at the Oval, Pakistan had scored over 500 runs, so it was less than 1% in the context of just their innings.

This seems paltry compared to other sports. In tennis, a warning is followed by a point penalty. A point penalty might sound small, but considering it is 25% of a game, it carries more weight. The recent final between Roddick and Ferrero in Cincinnati had 109 points - so a point penalty would be just under 1% of a typical match, but may be more important given that due to the scoring system, fewer points are required to swing a game to one player.

In football, the penalty is the most typical sanction of foul play, although it can only be awarded in the area around the goal. But it is still a controversial decision to award one, and is not done so lightly (despite the reputation of some referees).

The penalty usually results in a goal. I don't have the penalty conversion rates, but even if it was as low as 50% (which it won't be), it is still a much harsher sanction than the penalties in tennis or cricket. In the 2005/6 Premiership season, there were 2.48 goals per game. Assuming a 50% conversion rate (i.e. a penalty is worth 0.5 goals), awarding a penalty is about a 20% match swing to one team. (I won't look at rugby, as the penalty is given for more technical infringements and accidental mistakes than in cricket, football or tennis, and is an acceptable source of points rather than a punishment.)

So - why does cricket give the five run fine? It is a paltry sum and has zero impact compared to the shame and disgrace that is heaped upon the team or player involved. If the ICC really wanted to make it effective, it would be a 100 run fine. This would actually change a match, like a penalty in football, and give teams cause to think long and hard before tampering with the ball.

Quick aside: in all this fuss, no-one has noticed that Mohammed Yousef has scored a remarkable 631 runs in a 4-match series, which is the fifth-best 4-match total ever.

Runs	Player	Series	Season
712	JH Kallis	South Africa v West Indies	2003/04
706	RT Ponting	Australia v India	2003/04
703	GA Headley	West Indies v England	1929/30
693	EH Hendren	England v West Indies	1929/30
631	Mohammad Yousuf	England v Pakistan	2006

As the US Open approaches, I thought I would try to look at Federer's place in the all-time list, and came across this article from July where John Newcombe, the Australian tennis "legend", works out why Federer CAN'T be considered a great.

His theory stems from the fact that Federer "has not been able to win the French Open and Nadal has not only been beating him on clay, he's beaten him and matched him on hard court. So if you're going to put Federer there, you've got to put Nadal there."

Ok, I get the idea that Nadal has a good record against Federer, but a quick look at the grand slam list shows that Roger is on the up. He's won 8 (so far), which pits him with Agassi, Connors and Lendl. Not bad - he's only 24.

Rank	Player	Australia	France	Wimbledon	US	Total
1	Pete Sampras (USA)	2	-	7	5	14
2	Roy Emerson (AUS)	6	2	2	2	12
3	Björn Borg (SWE)	-	6	5	-	11
	Rod Laver (AUS)	3	2	4	2	11
5	William Tilden (USA)	-	-	3	7	10
6	Andre Agassi (USA)	4	1	1	2	8
	Jimmy Connors (USA)	1	-	2	5	8
	Roger Federer (SUI)	2	-	4	2	8
	Ivan Lendl (CZE/USA)	2	3	-	3	8
	Fred Perry (GBR)	1	1	3	3	8
	Ken Rosewall (AUS)	4	2	-	2	8

So who does Newcombe really rate? Not Sampras it seems, despite his unsurpassed record of 14 slams and 6 years in a row as year-end number one.

"Pete Sampras cannot go down in that category because the best he ever did at the French Open in 11 tries was one semi-final," said Newcombe.

"So there's no way he can be categorised as one of the best three or four of all time."

Well, who does that leave? The five who have won the "career slam" are:
Don Budge, Fred Perry, Rod Laver, Roy Emerson and Andre Agassi.

Now, if we apply the Newcombe theory, Agassi must be in the top five. But... he had a losing record against Sampras: 14-20. This isn't good at all. Accoring to Newcombe, If you can't beat your big rival you're not the best. Agassi is out.

What about Emerson and Laver? Well, their slam record is very close with Emerson on 12 and Laver on 11, but we all know that Emerson's tally is inflated by the fact that all his rivals turned pro. The head to head says it all. Laver wins 14-1.

What about Budge and Perry? Tennis records get a bit hazy pre-1968, so it's hard to assess their rivalry. If we take the matches that really count, the slam finals, can we learn anything there? Budge's slam finals record was played 7, won 6. Perry was played 10, won 8. Hard to separate them on this measure. However, in their only slam final meeting, the 1936 U.S. Championships, Perry won 2-6 6-2 8-6 1-6 10-8. Perry edges it - just.

So Newcombe's theory leaves Laver and Perry as the top players of all time. And being John Newcombe, we can discount Perry as a pom, leaving a good old Aussie as the best ever. A one-eyed, statistically inept view, but there you go.

Should England have dropped Geraint Jones for Chris Read? It's caused a lot of discussion, but the numbers back up the selectors.

If we take the last 10 tests* of the current wicketkeepers of the top eight cricket nations, what do we see? In the runs department, Jones is indeed bottom of the list by quite some way, with 100 less runs than the next man, Adam Gilchrist, and a paltry average of 17.9. Sadly, he leads the way with catches: 43. England are true to their word - the keeper needs runs.

Top of the tree in terms of runs would be Sangakkara after his mammoth 287 against South Africa in the all-time record stand with Jayawardene, but he was selected as a batsman for that match, not keeper.

Chris Read should be aiming for the middle of this table - an average of 31.5

Player	Country	Mat	Runs	HS	BatAv	100	50	Ct	St
Kamran Akmal	Pak	10	708	154	50.57	3	2	32	4
KC Sangakkara	SL	10	758	185	42.11	1	4	22	2
BB McCullum	NZ	10	431	111	33.15	1	2	24	3
M Dhoni	Ind	10	453	148	30.2	1	2	33	8
M Boucher	SA	10	461	85	27.11	0	4	37	0
D Ramdin	WI	10	370	71	26.42	0	2	22	0
A Gilchrist	Aus	10	352	144	25.14	1	1	31	5
G Jones	Eng	10	251	55	17.92	0	2	43	2

(source: cricinfo.com)
*Last 10 tests is the basic weighting used by the world rankings as well.