Margin of Victory
Points allocated for a Match
Projected Scores while First Innings in progress
Wicket Adjusted Run Rate (WARR)
Pulse – Tracking team points
Top performances in a match
Most Valuable Players
Analysing Performances within a Team
I discovered Ball Difference as the uniform margin of victory while watching the opening game of Cricket World Cup 2011. A lot has happened since. I tried to publicise the theory, explaining the simplicity and benefits to anyone willing to lend me an ear. Later I built a custom database to implement this theory while adding complexity to build a normalised player performance model. Now I am not interested in explaining or promoting Ball Difference. Hence no further posts in this blog or comments elsewhere. Ball Difference is the fundamental building block of the Relative Value Player Performance Model for limited overs cricket. It will get a mention with regards to explaining other measures but not otherwise.
This post is a curtain raiser for the forthcoming Cricket World Cup 2015. The format for Cricket World Cup 2011 was identical and thus reviewing the team and individual performances will also explain the proprietary charts that will be used during the live tournament coverage on this site (and Twitter).
I maintained a hand-made table in March 2011 about the largest margins of victory in that edition which can be found here. That table looked like this:
|2||A||Ken||69||23.5||NZ||72/0||8||252||New Zealand won by 10 wickets (with 252 balls remaining)|
|19||B||Ban||58||18.5||WI||59/1||12.2||226||West Indies won by 9 wickets (with 226 balls remaining)|
A simple bar chart can capture this information.
The margin of victory appears in blue on the left hand side of the yellow bar. Matches are sorted by the size of victory with narrowest margin appearing at the bottom. Match summary is also shown for every match above that yellow bar. For example the highest margin of victory was recorded in the second match between Kenya and New Zealand. Its match summary Ken 69(23.5); NZL 72/0(8) describes the runs scored, wickets lost and overs needed for both teams. New Zealand won the match hence its three letter short country code is capitalised. At the other end, we have the tied match between India and England – IND 338(49.5); ENG 338/8(50) where short country codes are capitalised for both teams.
This time I will not maintain the alternate points table comparing Net Run Rate and Ball Difference. Please see Wikipedia entry for NRR which mentions three other alternatives. As stated during the introduction I will henceforth use Ball Difference strictly in my own models only. It will not be mentioned otherwise.
The chart can be used to understand various types of matches. One team can totally outplay another either by bowling them out cheaply followed by a quick chase – Bng 58(18.5); WIN 59/1(12.2) OR post a big total and bundle out the opposition – SAF 284/8(50); Bng 78(28). At the other end we can have very close matches where both teams find it easy to bat – IND 338(49.5); ENG 338/8(50) OR easy to bowl – ENG 171(45.4); Saf 165(47.4). Points are allocated to each player based on the ease/complexity in batting/bowling along with the margin of victory.
The first chart shows points allocated to each player for batting, bowling and fielding. An earlier post describes how to read this chart.
How good is the first innings total of 230? At the group stage England defended 243 against West Indies – ENG 243(48.4); Win 225(44.4) and lost a close match to Bangladesh after scoring 225 – Eng 225(49.4); BNG 227/8(49). It means that generally a first innings total below 200 is below par and greater than 300 is above par but whether it will lead to victory depends purely on how the second innings pans out. In this match Sri Lanka posted a 10 wicket win. It will be a massive win in the longer format but in the limited overs edition chasing 230 in 30 overs, even if wickets are lost, should be treated as a bigger win. Dilshan and Tharanga scored at less than 6 runs per over yet saved more than 10 overs for a very comfortable win. The batting unit gets more points (511 against 333) and the 10 wicket win means all the batting points get shared only between openers. Dilshan scored more runs consuming fewer deliveries than Tharanga to earn more batting points. Trott and Morgan earn the most batting points for England followed by the bowling points earned by Mendis for his economical spell.
Let us look at another comfortable win by Sri Lanka inspired by bowlers.
265 is merely a decent first innings effort but bowling the opposition out for 153 is splendid bowling effort. It means more points for Murali’s 4/25 than Sangakkara’s hundred. Next we have another 10 wicket win at quarter finals stage which can be attributed to bowlers.
Mohammad Hafiz bowled economically in a supporting capacity and then led the opening act scoring nearly at run-a-ball. Shahid Afridi was the top bowler with 4 scalps.
Team batting second has an advantage in pacing their chase. Team batting first sometimes loses too many wickets in pursuit of extra runs. In other words, bowling team can restrict first innings total in last 20 overs even after conceding too many runs in the first 30. During group stage South Africa restricted Indian batting after a frenetic start.
FIPS stands for First Innings Projected Score. Based on the scoring pattern, a par value is determined which is plotted as the thin straight-ish line. Worm is a familiar curve illustrating ball-by-ball scoring pattern. The worm is coloured Red if Team batting first is ahead of par value or Blue if the bowling side is deemed ahead. After each delivery, based on runs scored and wickets lost till that point, two projections are made. The red curve is the projected final score if Team 1 bats above average that point onwards till the end of innings. The blue curve shows how Team 2 can restrict scoring through an above average performance.
Before the first ball is bowled, Team 1 will aim to score above 300 and Team 2 will try to restrict them below 200. Indian openers opened up after the first two overs making the most of power play overs. At the end of 15th over, the projected score was in excess of 350 which assumes that Indian batsmen maintain an above average tempo for 35 more overs, At that stage South Africa could expect to keep India under 265 by doing a much better job in remaining 35. After 30th, India had lost only 1 wicket to reach 197. Could they double this score? Not according to this model which projected only 350 runs (not 400) if the batting continued to be above average in the last 15 overs too. The projection for South Africa was 277 at this stage and it could work with quick wickets only. 9 more overs passed before the second wicket partnership was broken. After 39th India, after scoring 258, was projected to reach 359 while South Africa faced an above 300 score even if it could put a very good 11 over spell.
India lost 3 wickets in the next two overs. But losing 4 wickets with only 9 more overs to go does not make much difference to projections. India was still expected to cross 350 and South Africa wasn’t expected to keep them below 300. 2 more wickets were lost in next 4 overs. At 288/6 with 30 deliveries to bat, South Africa still faced an above 300 score. In the next 3 overs, 7 runs were scored for the loss of 2 more wickets. At 295/8, does one expect a team to score 5 more runs in 12 more deliveries? India added one more run and lost the last 2 wickets in the next 4 deliveries to finish at 296. South Africa bowled remarkably well in the last 11 overs or Indians failed to bat sensibly or both.
The projected scores are just that: projections. Broadcasters typically make a projection based on current run rate, 6 an over, 8 an over or 10 an over which are all in favour of batting side. FIPS model churns out three projections after each delivery: an average projection (not shown in the chart) similar to current run rate, a higher projection in favour of batting side (shown in red) and lower projection in favour of bowling side (shown in blue). During the World Cup 2015, I will publish these projections typically after 15th, 30th and 40th over whenever possible.
Albeit less dramatic, a similar pattern can be found in the tied match played earlier between India and England. India scored 292/3 after 45 overs which generally does not end at 338/10. It shows that bowling side can, and sometimes does, make amends. Hence a need for a projection in favour of bowling side.
Another custom chart is coming up to review England’s chase.
This chart was discussed at length in an earlier post.
The thin straight-ish line called ‘Target (ball-by-ball)’ specifies an expected score after each delivery for successfully chasing the target. Blue flat line indicates the final target while red line is the target score to reach after minimum overs. The runs scored worm is blue when chasing team is ahead and red otherwise. At the 40 over mark, England did very well to lose only 2 wickets while staying close or ahead of required score. England were favourites to win the match at that stage. Yet staying close to a mammoth required score and keeping up with it right up to finish line are two different things. Indian bowlers started pulling back the match with some quick wickets. After 49th, England needed 14 to win with 2 wickets in hand. India still marginally ahead. But 9 runs were scored in the first three deliveries. One bad delivery could result in a win for England. India did not conceded any boundary but 4 more runs in last 3 balls meant that the match ended in a tie.
Dilshan & Tharanga played watchfully for the first 7 overs but kept up the scoring rate after that to remain comfortably ahead of the target curve progressively increasing the margin between where they should be and where they were. This match ended with only 10 overs remaining. In the other quarterfinal, Pakistan finished the job much earlier by staying way way ahead of a modest target.
Now let us look at a failed chase.
Early wickets are the key to defend a low score. But South Africa started confidently by staying above the par curve in the opening partnership. England took 3 wickets by 20th over to ensure South Africa did not run away with the match. At 30 over mark, South Africa were still favourites but 4 wickets for not many runs in the next 10 overs swung the pendulum in England’s favour. Partnership for 8th wicket seemed to do the job but England struck at the right time and ensured that the tail did not wag to win by 6 runs.
Which brings us to the final chart for a single match.
This topic was covered in the most recent post. England were well below Par in the first innings handing over the points advantage to South Africa.
It was unlikely to win the match by bowling economically for 50 overs. Taking 10 South African wickets was the more likely way of overhauling first innings points deficit. Even after the fall of 9th wicket, South Africa had plenty of overs to reach the target. England’s margin of victory could be a few runs at best but South Africa would win the match, for the loss of 9 wickets, with overs to spare registering a comfortable win by balls remaining. Hence the difference between red and blue curves is bridged only towards the very end when England snuffed out the tail quickly.
In the high scoring match, England fell behind significantly after the first innings by conceding too many runs. This time the chase was near perfect and points deficit was steadily eroded. Around 40th over, we see a sharp spike narrowing the gap in points which means that England became favourites to win for the first time in that match. India increased the gap by breaking that partnership and taking a few more wickets to retain ascendency. But bulk of the mammoth target was already chipped. England stayed in the hunt with timely hits over the rope to get really close. The red & blue curve are too close in the 50th over and the match deservedly ended in a tie with both teams ending with 50 points apiece.
Australia thrashed Canada. Australia was ahead at the end of first innings. Canada did reasonably well for the first 20 overs. They could not break the opening partnership but the runs were checked which meant that the points gap narrowed slightly. Of course upset was never a possibility. Watson picked up scoring between 20th and 30th. Points deficit was too large when Watson got out and last rites were performed in the next 5 overs.
I would like to watch my team win with plenty to spare. A top performance is one that lacks drama. Bowling first, bowlers should keep opposition very quiet preferably bowling them out. Then top order should ensure that target is reached without fuss, Otherwise bat for 50 overs to score over 300 and win by 100 runs or more. That is why the top performances in the next 3 charts will invariably be from one-sided matches.
The first and third top performance is from the same match – a 10 wicket win inspired by openers. Watson’s 94 is sandwiched which was part of the chase against Canada discussed in earlier section. Dilshan makes 4 appearances and Tharanga is in the fray thrice making them ideal openers. Sehwag’s 175 against Bangladesh is 6th in the chart because India’s margin of victory was not significant. According to this Relative Value model, runs were cheaper in the opening game reducing the impact of the rapidly scored big hundred.
All the top bowling performances are from the matches where one side was bowled out for a very small total and the other side failed to do the same with ball in hand. Bennett’s 4/16 was part of Ken 69(23.5); NZL 72/0(8). Benn’s 4/18 resulted in Bng 58(18.5); WIN 59/1(12.2). Peterson’s 4/12 came in SAF 284/8(50); Bng 78(28). A top bowling performance is one where a player not only does better than teammates but also the opposition while the team’s objective of bowling out the opposition is collectively achieved.
Hafeez was the star performer in the one-sided quarterfinal win against West Indies. Peterson’s lusty blows to score 22* off 9 deliveries came in the match against Bangladesh mentioned earlier. Dilshan contributed with bat and ball in the big win against Zimbabwe – SLK 327/6(50); Zim 188(39). Bennett and Benn made it to the top without contributing with bat. If Sehwag’s 175(140) is of interest, or the hundred by Strauss in the tied match, then check the margin of victory. In this Relative Value model, a big hundred is valued more when neither your team mates nor the opposition lineup scores comparable runs alone or collectively.
Next we look at overall performances during the entire tournament. I like the league format where all teams play the same number of matches. In such cases we can simply add the points earned during the entire tournament. If a player misses any match it will be ignored since an equal number of opportunities were available to all, some just failed to grab them. In CWC 2011, two teams played 9 matches, two other played 8, another four played 7 matches and remaining 6 played six league games. We will disregard the number of matches and simply add all points. Those who feature at the top may have played more matches. May be they deserved to because some of those top performances ensured additional opportunities.
No surprises with Dilshan & Tharanga at the top. They ensured that Lankans got a solid start and matches were won with plenty to spare. These two are followed by the top order batsmen from leading test nations. Relative Value model expects a team to play its best batsmen for maximum overs. If they do their job then lower order batsmen will not get enough chances to score. Since limited overs matches are different from the long format, we do not have to worry about equal opportunities. A team is likely to win more if the top order does its job more often.
Unlike top batsmen, top bowlers must make way for lesser bowlers in the team. So this list is not dominated by a certain type of bowler. Shahid Afridi had an outstanding series especially with the ball in the subcontinent. India needed Zaheer Khan to perform which he did admirably.
All-rounders are expected to lead this chart as they get to contribute with both bat and ball. Afridi scored more bowling points than the batting points earned by Dilshan. But Dilshan compensated better with his bowling. Hence we find Dilshan ahead of Afridi. Yuvraj was the man of the tournament but India’s overall record was patchy. Sri Lanka eased into the knock out rounds while India lost to South Africa and tied with England during group stage. Both Pakistan and Sri Lanka recorded emphatic 10 wicket wins at the quarter final stage. India won the world cup by winning the three crucial knock out matches but Sri Lanka was the form team. The figures across all the world cup editions based on this Relative Value model confirm that CWC 2011 was the only instance where the team that won the title failed to earn maximum team points. Runners up Sri Lanka were ahead of India. Yuvraj is not far behind Afridi but he played an extra match. Yuvraj along with Zaheer were the chief architects for India; not the celebrated top order.
Next we look at the same three charts by superimposing their averages. An earlier post about Quartiles, Geometric Mean and Arithmetic Mean can be found here. The chart used is called CandleVolume (or BoxWhisker) that generally depicts the traded volume of a stock along with OHLC (Open, High, Low & Close) prices for the day. Open and Close values form the body of candlestick (or box) while High and Low appear as upper and lower shadows (or whiskers). Q3 (the third quartile) is the high whisker indicating that 75% performances were below this value. Q1 (the first quartile) is the low whisker and 25% performances (roughly 2 matches if we assume that top nations played 8 matches each) were below that value. Top of the dark black box indicates the Arithmetic Mean which is the generally understood Average Value. One strong performances masks other failures when we look at this average in isolation. Geometric Mean is a different type of average which will aid us in recognising consistency and its value will be the one along the bottom of the dark black box.
Tharanga misses out in this revised table. Most of his points were scored in a few matches. He scored two hundreds but also got dismissed for single digit scores thrice and once under 20. The Geometric Mean for his batting scores was not high enough to warrant a place in this selection. Yuvraj, Guptill, McCullum, Hafeez, Dhoni, etc miss out for the same reason. Instead we have Ashish Bagai from Canada, representing a weak side, who managed to put consistent performances without any scintillating effort in scoring over 300 batting points in 6 league matches. Sehwag and Mahela both made the cut but notice how wide the black box is. The higher the height of the black box, greater is the propensity of few good scores masking ordinary performances. Compare this with Sangakkara. The whiskers are neither above 120 nor below 60. It means roughly twice he scored more than 120 points. In addition he registered less than 60 points very few times. Such a narrow band of values certainly helps in getting the Arithmetic and Geometric Mean closer.
Is such a consistency a very good thing if a player does not cash in once getting set? The bottom of Dilshan’s black box (or G) is above the top of Kumar’s box (or µ) which means he may not be consistent enough within his zone but his zone is well above others. It is a bit like complaining about the wide disparity between Bradman’s G (=50) and µ(=90). That gap of nearly 40 is higher than the career average of most players. It is best to evaluate all four averages, Q1, Q3, G and µ together to get a feel for combined performances across the tournament.
A similar pattern can be observed between top 2 players in this chart too. Afridi did very very well in a few matches and failed in some others. Zaheer was more consistent throughout the tournament without any stand-out effort. Zaheer’s box isn’t tall at all but Afridi’s taller box is placed above his. Kemar Roach was the least consistent bowler in this selection.
Very few players passed the standard I had set to qualify as an all-rounder. It was partly done to discard those who already made a mark with bat or ball alone. This selection looks at 10-over bowlers who can bat a bit better than others.
Now we will look at the same information but compare players within a team. We will expect top order batsmen to do well and the middle order to make amends whenever they get a chance. It means we expect a good team’s top order batsmen to score more batting points than its middle order. If that does not happen then the concerned team needs to find a better top order. Relative Value model should be used to compare a specific type of player with another. It is designed to measure each player’s contribution not adjust/inflate contributions from certain positions to compensate lack of opportunities.
This chart is presented by typical batting position while mentioning the number of matches played by each member. Sehwag, Tendulkar and Gambhir were alright but none of them fared as well as Zaheer the bowler. Yuvraj was the most valuable player which is stating the obvious. A Ball Difference of 32 means that India won a match with about 5 overs to spare.
Dilshan stands tall as the batsman consistently supported by Sangakkara and occasionally in match winning starts by Tharanga. Murali and Malinga take the honours in bowling. The average winning margin was double that of India. Sri Lanka won by more than 10 overs on an average. The only problem is that they lost the final even though they have more tournament points (>7000) compared to India (<7000).
Now we look at the same information but with Quartiles and Averages superimposed. We find that Ashwin did well in the two matches he played. Kohli and Dhoni were not good enough overall though it is common knowledge that Dhoni played one good knock against Sri Lanka when the stakes were highest.
During CWC 2015, I will upload only the quartile charts for a team as it combines all the relevant information. Most of us despise formulae, resent averages, detest charts and dislike multiple axes. If a picture is worth a 1000 words then a chart should be good enough for few tens. These will appear on my Twitter handle as part of my whispers in the cul-de-sac while others trumpet their services in the marketplace.
I did a similar exercise earlier so I will spare the details. My selection criterion was simple: no more than one player from any team. I chose Ashish Bagai over Shakib or Kevin O’Brien for the eleventh spot. Other 10 selected themselves.