Sachin Tendulkar better than Don Bradman? Surely Not!

When the Wisden 100 was published ten years ago, there was much controversy over the non-inclusion of any innings by the great Sachin Tendulkar. Steve Ferrier's recent updating of the all-time list here also added no Tendulkar innings. Yet despite the fact that it is claimed he has not produced any particularly monumental innings during a career of more than 20 years, his greatness is still heralded the world over. How can this be?

Amongst cricket traditionalists, Don Bradman is unquestioned as the best batsman to ever pad up - they merely point to his Test average of 99.94, far and away the best ever, and say "QED". He and Tendulkar have probably only ever been rivalled in terms of hero-worship by WG Grace, who was probably the most famous cricketer ever, at least the most recognizable outside of the cricket world.

When trying to build up their man, Tendulkar fans point out the generally higher quality of cricket nowadays, together with the much larger number of competing teams and the vast numbers of games in all forms which must be negotiated, among other things which typify today's game. Bradman's protagonists meanwhile highlight the uncovered wickets, lack of protective gear and, interestingly, the lower numbers of competing teams, pointing out that a higher proportion of Bradman's games were against higher quality opposition as a result. In general, however these arguments are largely specious.

Tendulkar is the most capped Test player ever, recently playing in his 181st match, a feat of amazing longevity, and has scored over 2,000 runs more than anyone else has ever managed - and he's not done yet (though neither is his closest rival, countryman Rahul Dravid). Yet that very longevity is used by his critics to deride him, claiming that his records are simply the result of being around for so long. But it's one thing to hang around, and another thing entirely to put together a career with so many highlights.

In terms of sheer numbers per-innings or per-match, there's little doubt that Bradman is streets ahead of every other batsman. He had a higher proportion of double and triple centuries than anyone before or since, and as mentioned earlier his Test average is an order of magnitude above the rest - his 99.94 puts him more than 34 runs ahead of New Zealand's Stewie Dempster, next best of those who played at least ten Tests, though he played in just ten as compared to Bradman's 52. However Tendulkar has fashioned far more centuries than anyone else, a massive 51 as compared to 40 by next-best Jacques Kallis - again, his detractors will point to 298 Test innings (or about one in every six innings), compared to a return of 29 centuries in only 80 innings by Bradman (or about one every three innings). Tendulkar also has a better conversion rate than his contemporaries, but again Bradman's is far superior.

So how could anyone accept that Tendulkar is better than Bradman? The thing about hero-worship is it tends to focus the gaze, but not always in the right direction. Arguments fuelled by stats tend to use them, as Andrew Lang noted a long time ago, like a drunkard uses a lamp post - for support rather than illumination. A few carefully chosen stats are dropped in to "prove" the point which has already been pre-determined, rather than using stats for illumination, i.e. by first posing a question and using the appropriate stats to find the answer.

Method to their Madness?

Regardless of who is subjectively considered as being the best, I believe I may have come up with a way to measure the impact of a long career of sustained greatness in an objective way.

Let?s start by posing the question "Which batsman has had the most impact all-time"? Actually, since there are so many drawn games in Test cricket, let's modify the question slightly, to ?"Which batsman should have had the most impact"? Let's look at batting success in terms of the number of potentially high-impact innings played, and determine impact by looking at whether or not a particular innings of significance should, all other things being equal, have resulted in a win for his team.

It is fair to say that for the most part of his career, Bradman played for better teams - during his career Australia had a won-lost percentage of slighly over 70%. Tendulkar's India teams meanwhile have enjoyed a won-lost percentage of less than 55%. So by using the method described below, looking at individual innings as a whole throughout Test cricket, we can attempt to neutralise the team factor, i.e. try to take away the impact of being surrounded by a successful team.

Throughout Test cricket history, we can determine how often a particular innings of significance would, or should, be expected to contribute to a win in a Test match; I've done this for a previous article, which required me to go through every Test cricket scorecard which produced a result to determine how often each milestone innings contributed to a victory. As an example, let's take Tendulkar. Tendulkar numbers among his achievements 61 fifties and 51 tons, of which 20 were more than 150 and six of those double centuries - historically, each 150, for example, would be expected to prove the difference in a match around 24% of the time, which, for his fourteen 150s would give him around 3.2 win contributions to his team; in other words, if a player makes fourteen scores of between 150 and 200, he should expect around a quarter of them, or three in this case, to be winning innings. We can calculate this figure for each of his major innings, add them to give a total expected win contribution for Tendulkar, and then do the same for all Test batsmen based on their major innings. If we do this, we come up with the following top 25:-

NAME	Tm	EXP
Tendulkar	IND	8.962
Bradman	AUS	7.833
Lara	WI	7.700
Ponting	AUS	6.850
Dravid	IND	6.055
Sangakarra	SL	5.970
Jayawardene	SL	5.826
Gavaskar	IND	5.779
Kallis	SA	5.642
S Waugh	AUS	5.273
Sehwag	IND	5.247
Javed Miandad	PAK	5.214
Hammond	ENG	4.944
Mohammad Yousuf	PAK	4.429
Chappell	AUS	4.265
Sobers	WI	4.264
GC Smith	SA	4.246
Border	AUS	4.176
Richards	WI	4.008
Atapattu	SL	3.995
Hutton	ENG	3.976
Kirsten	SA	3.878
Langer	AUS	3.875
Harvey	AUS	3.702
Hayden	AUS	3.651

(EXP - Expected win contributions based on individual inning).

Tendulkar, Bradman and Lara are the big three and some way ahead of Ponting, who in turn is some way ahead of the rest. In the previous article mentioned above I looked at actual winning performances and, in comparing the number of actual winning innings with the projected figures, it emerges that Bradman, Ponting, Waugh and Sangakkara have significantly more actual wins than their performances would suggest; in the first three cases, this may be partly attributable to the more extreme success of their respective teams, i.e. their teams were able to turn their big innings into wins; Sangakkara is an anomaly here, but this is possibly partly explained by Sri Lanka's higher ratio of games against Bangladesh and Zimbabwe with Sangakkara in the line-up (about 40%), as compared to Australia, with only 7 of 97 wins (less than 8%) against the minnows with Ponting in the line-up. Another reason could be that Sangakkara is just a much better player under pressure.

Basically, the above chart reflects how people tend to think about career longevity, i.e. aggregate numbers with no real account of opportunity, strength of opposition or changing scoring conditions over time. However, as this shows Tendulkar at the top we may be getting some way to nailing down an objective measure of his greatness.

Playing the percentages

Of course, Tendulkar and Lara played in many more Tests than Bradman, so if we look at expected wins as a percentage of Tests played the list is somewhat different:-

NAME	TM	EXP	EXP%
Bradman	AUS	7.833	15.06%
Headley	WI	2.008	9.13%
RG Pollock	SA	1.619	7.04%
Sangakarra	SL	5.970	6.09%
Sehwag	IND	5.247	6.03%
Weekes	WI	2.868	5.97%
Lara	WI	7.700	5.88%
Hammond	ENG	4.944	5.82%
Paynter	ENG	1.119	5.60%
Amiss	ENG	2.741	5.48%
Nurse	WI	1.532	5.28%
Hutton	ENG	3.976	5.03%
SG Barnes	AUS	0.648	4.99%
Tendulkar	IND	8.962	4.95%
Mohammad Yousuf	PAK	4.429	4.92%
Chappell	AUS	4.265	4.90%
Walcott	WI	2.156	4.90%
Jayawardene	SL	5.826	4.86%
Morris	AUS	2.213	4.81%
Ponsford	AUS	1.379	4.76%
HL Collins	AUS	0.893	4.70%
Harvey	AUS	3.702	4.69%
GC Smith	SA	4.246	4.67%
Gavaskar	IND	5.779	4.62%
Pietersen	ENG	3.603	4.62%

(EXP% - expected win contributions as a percentage of total Tests played; minimum five years and 10 Tests)

This brings into the top five such batting luminaries as George Headley, Graeme Pollock and...Virender Sehwag, while Ponting drops out of the top 25. Looked at in this way however, Bradman is now seen to be enjoying his customary position of being about half as "good" again as the chasing pack.

Still we don't have the full story, however - we need to take into account the changing eras of cricket. For example, in the 19th century a score in the 90s would prove to be a difference-maker on far more occasions than in today's high-scoring game.

The best of their time

Below is the revised top 25, again based on expected win percentage taking into account era-specific scoring:-

NAME	TM	EXP	EXP%
Bradman	AUS	8.131	15.64%
Sehwag	IND	7.210	8.29%
Mead	ENG	1.406	8.27%
Sangakarra	SL	7.842	8.00%
Lara	WI	10.428	7.96%
Hobbs	ENG	4.612	7.56%
Headley	WI	1.643	7.47%
Ranji	ENG	1.109	7.39%
Tendulkar	IND	13.022	7.19%
Mohammad Yousuf	PAK	6.257	6.95%
Paynter	ENG	1.357	6.78%
Weekes	WI	3.234	6.74%
Jayawardene	SL	7.940	6.62%
Ponting	AUS	9.692	6.38%
GC Smith	SA	5.716	6.28%
Pietersen	ENG	4.817	6.18%
Kallis	SA	8.795	6.07%
Hammond	ENG	5.080	5.98%
Trumper	AUS	2.861	5.96%
Younis Khan	PAK	3.930	5.87%
McGlew	SA	1.972	5.80%
Atapattu	SL	5.133	5.70%
Cook	ENG	4.093	5.68%
SG Barnes	AUS	0.731	5.62%
Lehmann	AUS	1.476	5.47%

As might be expected, era adjustment has brought in some players from earlier eras, such as Victor Trumper, Ranji and Jack Hobbs, however we've also reclaimed some recent players, e.g. Younis Khan. This is because, nowadays a substantial innings has a higher likelihood in leading to a win than in some previous eras, where they may have more likely led to a draw, e.g. Pollock who played exclusively in the 1960s. But if we look at the top five, Bradman is top by some margin, although Sehwag and Mead now replace Headley and Pollock for second and third. So why has Headley dropped while Bradman's score went up slightly, considering they played around the same time? This is because of the varying impact of particular individual scores - Headley's breakdown of scores is quite different to Bradman's; although their proportion of scores below 200 are quite similar, Bradman had 15% of his scores above 200, whereas Headley had 5%.

The make up of the above list is quite surprising to me. Despite the cut-offs of five years or ten Tests, it is still partially made up of players with relatively short careers, e.g. Cook, whose career is still quite short in terms of years, or Mead, who despite his long career duration did not play in many Tests, averaging only one per year.

Staying power

At the end of the day, though, as I mentioned at the beginning many modern-day fans don't tend to make judgments on percentages - the truly great players are judged on and remembered most fondly for sustained brilliance maintained throughout a significant career. Here, then, is our final ranking, based on total expected win contributions adjusted for era:-

NAME	TM	EXP
Tendulkar	IND	13.022
Lara	WI	10.428
Ponting	AUS	9.692
Kallis	SA	8.795
Dravid	IND	8.485
Bradman	AUS	8.131
Jayawardene	SL	7.940
Sangakarra	SL	7.842
S Waugh	AUS	7.377
Sehwag	IND	7.210
Mohammad Yousuf	PAK	6.257
GC Smith	SA	5.716
Hayden	AUS	5.517
Kirsten	SA	5.367
Langer	AUS	5.343
Border	AUS	5.181
Atapattu	SL	5.133
Hammond	ENG	5.080
Gavaskar	IND	5.075
Pietersen	ENG	4.817
Inzamam-ul-Haq	PAK	4.796
Hobbs	ENG	4.612
Chanderpaul	WI	4.417
Hutton	ENG	4.154
Sobers	WI	4.134
Gooch	ENG	4.125
Cook	ENG	4.093
Azharuddin	IND	4.052
Laxman	IND	4.033
Harvey	AUS	3.999

There are some surprises lower down the order, but keep in mind that this is measuring likely winning performances, so a mercurial batsman like Atapattu measures well despite his low average. The level of likely success of a given innings changes over time, plus this last ranking is based on totals and tends to favour players who batted in more Tests.

Tendulkar has a lead of almost 3 expected wins over the next man, Lara, which considering the small variations below him is a significant gap - the 25% lead over the chasing pack is almost, dare I say, Bradmanesque. That said, there's no doubt that Bradman would have added to that total with additional Tests that would have been played had not World War II shifted the world's focus away from sport.

The previous two tables I think illustrate how the two camps tend to feel about the greats - on the one hand, the pro-Bradman fans would expect to see the likes of Hobbs and Headly ranked highly, whereas Tendulkar's admirers may expect to see Tendulkar, Lara and Ponting at the top. What I've tried to do here is show that there is an objective measure of greatness, other than averages and run aggregates, which illustrates just how special Tendulkar is.

The Last Word

While I'm not here to proclaim that Tendulkar is better than Bradman (for "better" is after all a relative term and means many different things to many different people), by the above measure Sachin Tendulkar is (or rather should have been) the most successful Test batsman ever, and by a significant margin.

I'll leave the last word to Bradman himself, who was so impressed by Tendulkar that he made him the only player of the modern era to be included in his own all-time XI. And he knew a bit about cricket.