Call Steve Waugh's bluff
Surjit S. Bhalla - 26 February 2001
The Aussies are on a roll, and the stakes for them are high. Steve
Waugh recently stated that the record breaking 15 wins in a row would
(almost) be for naught if they don't win in India.
The Waugh of words started when Steve mischievously suggested that the
Indian side was preparing to face the Aussies by preparing the dull
tracks of yesteryears. Remember those cricket killing fields of the
India-Pakistan 1960 series, or the India-England 1980 series, or the
India-Pakistan series of 1987 (before the final Test)? Remember also,
how Test cricket was reborn in Bangalore when India narrowly lost a
match on a grossly underprepared pitch? But the crowds came in droves,
to watch a cricket contest, and not batting for the record books. And
Gavaskar scored 96, a score which I judge to be an adjusted score of
179 runs.
So if Mr. Waugh has raised the dead pitch bogie, he may be arguing a
just cause. But according to me, Steve is trying to be clever -
perhaps too clever by half, which will be the case if his bluff is
called. It is no secret that Australia would like to extend their
unbroken record, and beat India in India. And their chances of doing
so are higher on a dead track than on an underprepared pitch! In other
words, if the Indian cricketing authorities confirm Steve's ostensible
nightmare, they will be fulfilling his dream.
The CricketX team has developed models of both Test cricket and one-
day cricket. Complete records of Test match data (since the first Test
at Melbourne in 1877) and one-day cricket (since the first match, also
at Melbourne, in January 1971) are incorporated into an integrated
model of cricket. This model has an important UHH (Untouched by Human
Hands) property i.e. the model generates all the results that are
necessary for a cricket fan. For example, what are the odds of a team
winning a match, or a series? Who is likely to score the most runs,
take the most wickets? Which player is the best, both on a career and
form basis.
The model incorporates various "known" factors that affect a player,
and team's, performance - own ability, opposition strength
(compartmentalized as bowling and batting strength), home team
advantage, nature of pitch, "tension" conditions in a match etc. None
of these parameters are subjective - they are generated from within
the crimetric (cricket econometrics) model. The inner workings of
this model have been discussed at various forums - London School of
Economics, Princeton University, Delhi School of Economics, Indian
Statistical Institute to name a few. This model formed the basis of my
book on Test cricket (Between the Wickets: The Who and Why of the Best
in Cricket) published in 1987, a book described by John Arlott as "a
salutary and illuminating book which will humble all of us who love
and dispute about the game. Let us check our prejudices by these
computerized judgements".
Arlott was right - the model judgements have to be at least
approximately right. Indeed, that is the bedrock of the philosophy
behind my work - it ain't good unless it forecasts well - and it
cannot forecast well unless it meets the simple smell test of common
sense and intuition.
The key forecast of interest to most fans is who will win the
forthcoming India-Australia series. And why. If the India-Australia
series were to be played in a third country on neutral grounds with
zero home-team advantage then India would have a 50 percent chance of
losing, 14 percent chance of winning and a 36 percent chance of a
draw. (These are forecasts for the first match of the series; after
every match, the model is updated on the basis of data from the new
match, and new forecasts made). But this series is being played in
India, on Indian pitches and other natural "home-team" advantages that
Indians have e.g. familiarity with the grounds, heat conditions, food,
lack of homesickness etc.
The model has separate indices for batting and bowling; in batting,
India is about 35 percent behind the Aussies, and in bowling,
Australia is about 19 percent ahead. But my analysis of past trends
show bowling is more important than batting (has a higher weight) in
explaining performance in Tests (the opposite is true in one-day
cricket). So India starts off at a considerable disadvantage.
The batting and bowling indices of the two teams are used to generate
the above probabilities e.g. India's probability of losing a match is
50 percent. The model uses these indices to generate a three-way
forecast - the probability of a team winning, losing and drawing a
particular match. Thus, it is not 50 percent of winning since the
probability of a draw (36 percent) is also present! The most
"difficult" prediction, in terms of accuracy, are drawn matches. This
is expected since apart from the pitch, rained out days are important
factors affecting a draw - and the probability of rain is not part of
our model, yet! If drawn matches are excluded, then the correct
forecasts for matches involving Australia are close to 67 percent, and
those involving India, 73 percent. For Zimbabwe, the correct
prediction record is a high 82 percent.
The probabilities of an Indian loss get reduced to 44 percent with the
knowledge that the pitch will be neutral (neither favouring bowlers or
batsmen). The model generates a pitch index which has varied from 0.40
(in the 1964 England-Australia Test at Manchester when Australia
scored 656 and England replied with 611) to 4.4 (Australia 112 all
out) vs. England (61 all out) in January 1902. Incidentally, the pitch
index for the India-Pakistan match in Bangalore in 1987 was 1.9. The
distribution of the pitch index over all the matches played is as
follows.
Pattern of pitches in Test cricket - 1877 - 2001
Pitch Index range according to our model No. of Tests % of Matches
Less than 0.75 278 18.2
0.75 to 0.9 320 20.9
0.90 to 1.1 431 28.2
1.1 to 1.4 314 20.6
Greater than 1.4 185 12.1
Average Index = 1.04
Total 1529 100
Getting back to the basics. On a neutral pitch (pitch index between
0.90 and 1.1), and playing in India, the probability of an Indian loss
is 44 percent. But if the pitch favours batsmen (pitch index between
0.75 and 0.9) then the chances of an Indian loss get reduced to 41
percent. This would seem that Steve Waugh is right in not wanting a
dead pitch. But note the steep decline in probability of an Indian
loss - from 44 to 35 percent - if India prepares pitches that favour
spin bowlers - and this is what Mr. Waugh is really not wanting,
rather than his feigned protestations about a dead pitch. I think Mr.
Waugh does protest too much.
Pitching for the Winning Pitch
Probability of(in %)
Nature of Pitch Indian Loss Draw Indian Win
Flat Dead Pitch - Index 0.75 - 0.90 41 40 19
Sporting Neutral Pitch - Index 0.90 to 1.1 44 39 17
Turning (underprepared) pitch - Index 1.1 to 1.4 35 42 23
One way to assess the accuracy of the model is to compare it with
market forecasts. As reported in the Indian press, legal betting odds
in Australia (provided by CentreBet) before the first Test were as
follows: a drawn series 2/1, an Australian win 4/5 and an Indian win
9/2. These odds are market odds and hence have immense credibility -
ours are based on a model which uses Test match data since 1877! Using
an average set of probabilities reported above, the odds we come up
with for a drawn series are 3/2, for an Australian win 3/2 and for an
Indian win, 4/1. The market odds and the odds arrived from our model
are not that far apart, which is reassuring.
There is an additional dimension - the psychology of an Aussie loss.
Think about it - here is a team that has won 15 matches in a row,
against a team that has won only 4 of their last fifteen matches -
actually only 2 since victories over Zimbabwe and Bangladesh are not
that meaningful. What if the Australian team loses to India! The
humiliation and the psychological setback for Australia would be
large, and well worth the effort. If on a spinning track, India loses,
well, that was predictable, and India can go on and hope to win the
next match.
Now think of a dead pitch. The probability of Australia losing on an
Indian dead pitch is 19 percent (in Australia, on a dead pitch, the
probability of an Australian loss is only 10 percent). Not
insignificant, but small. But think of the humiliation India would
receive if it lost to Australia on a dead track. And since on such
wickets, the margins of victory are high, the ignominy of a loss
should not be underestimated.
Bottom line: By preparing a flat track, the Indian cricket
administrators would be falling into a carefully laid out Aussie trap.
The last thing Indians should do is to prepare dead tracks. Also,
"neutral" wickets should not be prepared. The best way to neutralize a
heavy disadvantage in batting and bowling is to underprepare the
wicket - not too much to make the result random (India-Pakistan, 1987)
but enough to make the Australians run for the money - and give India
a two-thirds chance of not losing, and a one in four chance of
actually winning.
(The author is one of India's most noted economists. Dr. Bhalla
received his Ph.D from Princeton University in 1976, following which
he joined the World Bank, after brief stints at the Rand Corporation
and Brookings Institution. In 1990 he was appointed the Bank's Chief
Investment Officer. Currently Dr. Bhalla is the Managing Director of
Oxus Fund Management (OFM), an asset management and emerging markets
advisory firm. He also heads the website CricketX.com which is the
only website which posts forecasts of cricket matches, both static
i.,e pre-match) and dynamic i.e over-by-over).
How good is CricketX, after all?
© CricInfo & CricketX