Win probability after the first innings

Yesterday afternoon I went to North Sydney Oval to watch the Renegades play the Strikers in the second WBBL game of the season. Yesterday’s post had pointed out that no team had won a domestic Australian women’s T20 match since 2009/10 after having been bowled out. When the Strikers were bowled out for 116, I pointed out this stat, but I was also aware that 116 was a relatively good score for a bowled-out team and likely more than most other cases of teams who lost all of their wickets.

The Strikers went on to win with relative ease after bowling out the Renegades, making it the first match in women’s domestic T20 in Australia since 2009/2010 where the team batting first lost all of their wickets and went on to win, and also the first match since at least 2009/10 where every single wicket was taken.

To help understand these chances of success, I have calculated the probability that a team batting first will win based on their score at the end of their innings. The following charts show these statistics for both men and women (although it’s worth noting the volume of data is much greater for the men).

halftimewinchanceF.png

To take the three matches played yesterday as examples:

  • The Perth Scorchers scored 119 in their first innings in the morning game. Teams scoring 116-120 won 49.4% of the time, and lost 47% of the time. Based on this, you would expect a very close contest, with Perth having roughly 50% chance of victory. We got a close result, with the Hurricanes defeating the Scorchers with an over and 5 wickets to spare.
  • The Adelaide Strikers recovered from an early collapse to score 116 after losing all of their wickets. This fell into the same category as the Scorchers innings, but the Strikers went on to win with ease after another batting collapse by the Renegades.
  • The Melbourne Stars performed strongly in the evening game, scoring 147. Women’s teams scoring 146-150 went on to win in 75% of cases, and so did the Stars – although the Thunder did well to come within six runs.

halftimewinchanceM.png

I might update these charts from time to time – at the moment they use data up to 22 November.

You can look at the data yourself here if you want to check the exact proportions, and see how many match results are included in each category.

There’s a lot more to be said about win probabilities – things get more complicated when you look at points midway through the first or second innings, at which point you also need to factor in wickets lost. I will definitely be returning to this point, but for now you can get a good sense of the likely result at the halfway point.

Advertisements

Is the WBBL lifting women’s T20 scores?

At the start of the first WBBL season, I wrote a blog post about women’s cricket which amongst other things analysed ‘normal’ par scores in women’s T20 cricket. I returned later to do a post specifically focusing on par scores in both men’s and women’s T20 cricket.

After writing these posts, I kept an eye on WBBL scores and thought that they were beating the par scores more often than expected, and it made me wonder whether the higher level of professionalism and the inclusion of international players was meaning that the WBBL was producing higher-scoring games. This brings us to today’s post, where I’ll run through some statistics showing how the number of runs scored per innings has changed in the WBBL, and some explanations as to why. I’ll also give some evidence that this trend is not taking place more generally.

I started with this first chart. I separated out scores based on whether the team won the match, and whether they were playing in the first innings and second innings. I had issues last season with working out how to treat winners in the second innings, as they do not necessarily have the same incentive to score as fast as possible, and we don’t know how many runs they would have scored if they had batted for the full twenty overs. The chart shows the average run rate, not the average score. Multiply by 20 to get the average score.

wt20closeness2.png

(Where a team is bowled out, I calculate their run rate as if they had batted without scoring for the remainder of their overs. Where a team ceases to play because they have won the match, I don’t do this. The data dates back to the 2009/2010 season, which is the first for which I have data for a domestic Australian women’s T20 competition.)

Interestingly, you don’t see an increase in run rates for winning sides in either innings. You see a mild increase for losing teams batting first, but the big change is for losing teams batting second.

These are the teams which are chasing down a total, and in most cases will be behind in the chase. In 2009/10, the average run rate for these teams was around 4.9 runs per over, which would total 98 runs. It almost reached 6 runs per over (120 runs) in the first WBBL season in 2015/16.

So is this is a general trend, or just one in Australia? Unfortunately I don’t have a complete dataset of women’s domestic cricket in countries like England, India or South Africa (only New Zealand has as much data as Australia), but I can compare the Australian domestic women’s game with T20 internationals and the equivalent men’s events:

wt20closeness1.png

No evidence of a similar trendline in these other datasets.

My next theory: losing teams are not being bowled out so quickly, so they last the whole innings.

wt20closeness4.png

Not true! There was a significant jump in the proportion of matches ending with the second team being bowled out in 2015/16. After slowly dropping from 2009/10 to 2014/15, last season this measure jumped from 22% to 41%. It’s also interesting to discover that not a single team that lost all ten of their wickets has gone on to win their match in the last six seasons of the WBBL and its predecessors.

I also checked out the average number of wickets lost in each innings of play:

wt20closeness3.png

The number of wickets lost per innings ticked up in 2015/16. It’s also worth noting that losing performances in the second innings have consistently lost a wicket more than losing performances in the first innings, over the last seven seasons. There’s a similar trend seen in the Big Bash League.

So losing teams, particularly those batting in the second innings, are scoring more runs relative to the teams that beat them. This is happening alongside those teams also losing more wickets (particularly in the second innings), and being much more likely to be bowled out if chasing down a target.

So if losing scores are going up, and winning scores are remaining steady, what does this mean for the closeness of games? Are they getting closer? Well, yes, they are.

wt20closeness5.png

Back in 2009-2012, the gap between the winning team and losing team was usually around 1.75-2 runs per over (35-40 runs, if the winning team bats first or wins in the final over).

This slowly declined, before crashing in 2015/16, when the gap in WBBL01 was barely one run per over.

The same isn’t true in other T20 tournaments.

wt20closeness7.png

There was an improvement in close matches in the first season of the franchise-based Big Bash League in 2011/12 but there hasn’t been a clear trend since then. There are no clear trends in close matches in T20 internationals for either men or women since 2009.

(I’ll save you yet another chart, but there is also no such trend in New Zealand women’s cricket).

So what can we learn from these stats? I have two main theories which are backed up by this data.

Firstly, the move to franchises in the WBBL last season improved competitiveness in the tournament. The NSW Breakers have been traditionally dominant in state-based women’s cricket. They’ve won all but three seasons of the National Women’s Cricket League (50-over cricket) in its 21-season history. NSW had also made the final in five out of six seasons of the state-based T20 competition which preceded the WBBL. This would explain why we see a particular increase in competitiveness in 2015/16. The old state sides had much more of an uneven pool of potential players, and the ability for players from the bigger states to move to smaller states for more opportunities would have been much less for women due to the minimal pay.

But I don’t think that explains everything – since competitiveness has been improving prior to the last season. In future posts I might try to explore this question further.

 

Usman Khawaja’s unique BBL season

I’m a Sydney Thunder fan, and I was pleased to see how well they did in 2015/16. They managed to scrape into the finals, before winning both the semi-final and the final away from home to take home the trophy. This came after the team took the wooden spoon for the first three seasons before managing second-last in 2014/15.

But something bugged me about their performance, and that was the unique role of Usman Khawaja. During the last BBL season, Khawaja popped in and out of the Sydney Thunder side depending on injury and his test duties, and when he was present the team did consistently well – winning three matches and losing a fourth narrowly.

I thought there was something unique about Khawaja’s performance, and wanted to check it out by delving into the data. Namely, that Khawaja’s total run-scoring in his innings was much higher than any other BBL player, and that the way he scored his runs was different to most other high run-scorers in the tournament.

In later posts I’m going to delve into some theories about what matters in T20 cricket when you’re a batsman. In short, while scoring at a fast pace is important, I’m going to argue that wickets are still valuable in T20 and we should also judge batsmen on their ability to stay in without losing their wicket, at least in some circumstances.

I’m judging batting performance here as the number of runs the batsman scored as a proportion of the total runs scored by his team in matches which he batted in. This metric gives you a sense of how much a player pulled their weight in the team: they may get there by consistently scoring and staying in for most of the innings, or by scoring at a breakneck speed. In either case, staying in without scoring or scoring fast and getting out within the first over won’t serve you well in this metric.

khawajachart1

Khawaja is way ahead of the pack on this measure. Usman Khawaja scored 345 runs in his four innings, out of a total of 692 runs scored by the Thunder in those innings, a score of 49.86%. The next highest score on this metric is 29.6% by Aaron Finch of the Renegades, followed by 29.1% for Shaun Marsh of the Scorchers. Continue reading “Usman Khawaja’s unique BBL season”

Par scores in T20 cricket

Par scores in T20 cricket

I haven’t blogged for the last two weeks – partly because life has been busy, but also because I’ve struggled to come up with anything to say that provides particular insights about individual BBL or WBBL matches that are being played. I will return to this, and will continue to post key stats about various matches on the Strike Rate twitter account.

In this post, I’m posting my analysis of ‘par scores’ for T20, and how they vary between the men’s and women’s game, and in different parts of the world. This is useful for understanding what sort of score can be expected in particular conditions.

Par scores are calculated as run rates, which can be converted into total scores by multiplying by 20. This is more useful than raw total scores, since not all innings last for the full 20 overs. When a team wins in the second innings, they often win with overs remaining, and thus their total score is less than their potential score if the match had been fully played out. Run rates are adjusted to assume that a team which loses all its wickets played the full 20 overs. So if a team is bowled out in 17 overs, the run rate is calculated over 20 overs, not 17.

To get a sense of a normal range, for each category I have included the first quartile, median and third quartile figures. 50% of scores lie between the first quartile and the third quartile.

Gender Result 1st quartile Median 3rd quartile
Men loss 5.90 6.90 7.80
Men tie 6.95 7.60 8.20
Men win 7.25 8.18 9.10
Women loss 4.21 5.10 5.95
Women tie 5.20 5.48 5.95
Women win 5.70 6.45 7.38

Remember – to translate these run rates into a full innings, multiply the figure by 20.

Here’s the same figures as a chart.

runrate_par1

For men’s T20, a median winning score is around 164 runs, while a median losing score is more like 138. For 82 matches which ended in a tie, 50% of scores were between 139 and 164. It’s worth noting that 25% of losing scores are 156 or over, while 25% of winning scores are under 145.

For women’s T20, the whole range of scores is lower. The median winning score is around 129, while the median losing score is 102, and there is a big overlap between winning scores and losing scores.

Below the fold I’ll break this data down by year and host country. Continue reading “Par scores in T20 cricket”

W/BBL live blog – Sunday 20 December

2:05pm – The Thunder would’ve been favoured strongly to beat the Strikers after scoring 148. In 18 women’s T20 matches played in Australia where the first team scored 146-150, that team went on to win 13 times. But the Strikers have won today.

The Sixers look likely to win their match, needing only 39 runs off 45 balls with plenty of wickets to spare.

1:52pm – The Renegades have beaten the Heat, recording their first win. It also looks like the Strikers are about to record their first win over the previously-unbeaten Thunder, and the Sixers look likely to beat the Scorchers – it’s a day of underdog wins.

1:15pm – We have five WBBL matches and two BBL matches today. One of these matches has already finished and three other WBBL matches are underway.

The Hurricanes have already won the first match against the Stars, which is the Stars’ first loss in the tournament. The Hurricanes scored 6/144, and in response the Stars were bowled out for 114 off 19.2 overs.

The Hurricanes now have six wins and one loss, and the Stars have four wins and one loss. The Hurricanes’ innings was built on the performance by opener Hayley Matthews. Matthews scored 77 off 51 balls. Her previous best score was 32, and she had only scored more than ten once in this tournament – and has already had two golden ducks. Meg Lanning scored 20, which was her lowest score in this tournament.

At the SCG, the Scorchers just finished their innings on 8/107. Previous results in women’s matches in Australia suggest they have an approximately 30% chance of winning with those figures. The Scorchers innings started very poorly and they did well to reach 107, but this is the Sixers best shot at winning a match, after losing their last six matches.

The Brisbane Heat are looking wobbly after scoring 110 at the Gabba. The Renegades are on 2/69 off 11.2 overs and look likely to have their first win.

There was some unexplained delay in Brisbane in the Adelaide Strikers innings, as they aim to chase down the Sydney Thunder’s score of 6/148. At the moment the Strikers are on 1/60 off 9 overs – it will still be hard for them to win, but they have quite a few wickets to spend.

W/BBL live blog – Saturday 19 December

10:43pm – That’s it for tonight. I’ll return tomorrow with sporadic coverage of the three televised Twenty20 matches (along with some coverage of the other four non-televised WBBL games). In the meantime, you can also follow me on Twitter.

10:40pm – Globally, there’s four cases of a team scoring 180 or above with a lower top score. The Dolphins in South African cricket in 2006 scored 191 with a top score of 32, in 2008 Central Districts in NZ scored 180 with a top score of 33. In 2007, Otago scored 185 with a top score of 33, and in 2008 the Deccan Chargers scored 182 with a top score of 35 in the IPL.

In 2006, the Hyderabad Hawks scored 180 with a top score of 35 – the same stats as the Heat tonight.

10:35pm – One stat I’ve been tweeting about. The Heat scored 180, which is quite a high score (22/26 first-innings scores in the range have won) but wasn’t enough. One fascinating thing about their scoring was that they used seven batsmen, all of whom made it into double figures, and none of whom scored above 35. No team has ever scored anything like 180 in Australian conditions while having that amount of balance in their side.

In 2008, NSW scored 7/159 with Steve Smith top-scoring on 33, and they lost to WA. In 2010 and 2011, Sri Lanka A and Queensland each scored 149 with a lower proportion of runs scored by their top-scoring batsman.

6:42pm – So tonight’s BBL match will start in half an hour between the Brisbane Heat and the Melbourne Renegades. The Brisbane Heat had a terrible season in 2014/15 and will be hoping to recover from that position. I’m going to be out for a few hours so will likely return in the second innings with some more analysis.

6:39pm – We’ve now seen twenty out of 56 regular-season WBBL games, and some clear trends are emerging, but these are clouded by major differences in the number of matches played by each team.

The Melbourne Stars and the Sydney Thunder have both won all of their matches, although they have only played four and three games each, and they haven’t played each other. The Stars and Thunder will play each other twice in Sydney on the final weekend of the regular season, with matches at Olympic Park on Friday 15 January and at the University of Sydney on Sunday 17 January, on either side of the SCG Sydney Derby.

The Hurricanes and the Heat have both impressed. The Hurricanes chalked up five wins before losing their sixth match today, while the Heat started out with a tough competition, losing their first two matches to the Stars before narrowly losing to the Scorchers. They have since won five in a row and could win most of their remaining matches.

The Scorchers are ranked fifth, and that seems about right. They’ve won three out of five matches, which has included defeats of the Hurricanes and Heat. They’ve only played one match against a lower-ranked team, so they’re likely to improve as their schedule gets easier.

The Strikers, Renegades and Sixers have all fallen short, with no wins after four, four and six matches respectively. The Strikers and Renegades have each had some close calls, but the Sixers have been disappointing throughout the tournament.

6:28pm – The Hobart Hurricanes were aiming for a sixth successive win, but scored a below-par 108, which was then chased down by the Perth Scorchers. The Hurricanes’ bowling started well, with Scorchers opener Ellyse Villani going for a golden duck on the first ball, but the score wasn’t enough to hold them off. The Scorchers now have won three out of their five matches.

6:24pm – The Stars easily knocked over the Sixers – chasing down the Sixers’ total of 80 in only 15.5 overs and with the loss of only three wickets. The Stars have now won all four of their matches, while the Sixers have lost all six of theirs.

6:20pm – The Heat beat the Strikers this afternoon at the Gabba, with one over to spare. Star batter Grace Harris was lost to a golden duck, but her usual partner Beth Mooney paired with Jess Jonassen, scoring 48 and 67 not out respectively, most of those runs in an 89-run partnership before Jonassen was joined by Lauren Winfield, finishing off the match. The Heat have now won five matches in a row after losing their first three. The Strikers have lost all four of their matches, despite performing strongly today.

4:11pm – The Adelaide Strikers scored 9/125 in their innings against Brisbane at the Gabba (televised on One). This was a good recovery after collapsing in the middle of their innings, losing six wickets between 11.2 overs and 16.2 overs. Teams scoring 120-130 runs in the first innings have won 55% of matches.

4:07pm – The Hobart Hurricanes scored 8/109 in their innings. The team batting first only won in 10/35 matches when scoring 105-115 runs. But they then took a Scorchers wicket on the first ball, which will improve their position.

4:06pm – The Sydney Sixers women have continued their poor form, scoring 9/80 in their innings. The Melbourne Stars are on 0/6 off 1.3 overs. In women’s T20 matches in Australia where the first team scores between 70 and 90, that team lost 18 matches, tied one match and only won one match.

4:03pm – In the two WBBL matches this morning, the Thunder defeated the Renegades with three balls to spare, and the Hobart Hurricanes held on against the Perth Scorchers, who fell short by one run. The winning scores in the two matches were 130 and 133.

11:08am – Hobart is on 5/118 off 18.4 overs – which is headed for a score which would give them about a 50% chance of winning.

10:05am – We have six Twenty20 matches being played in Sydney and Brisbane today, with the first starting right now in Blacktown.

In the WBBL, the Hurricanes and Scorchers are playing at Blacktown at 10am, then the Thunder and Renegades will be playing in Brisbane at 11am. In the afternoon, the Sixers play the Stars at Drummoyne at 2:30pm and the Scorchers play the Hurricanes at Blacktown at the same time.

The first televised WBBL game starts at 2:40pm AEDT at the Gabba, with the Brisbane Heat hosting the Adelaide Strikers. Following that match, at 7:10pm AEDT the Brisbane Heat men’s team will be hosting the Renegades at the same ground.

I’m not going to be present for the whole day but I’ll drop in every now and then with little stats updates – and if you want to follow along please comment below.

Right now the Hobart Hurricanes women are on 0/5 off 1.3 overs at Blacktown. Both teams have performed well, with Hobart winning all four of their matches and Perth winning three out of their four.

Thunder vs Sixers – story of the numbers

Thunder vs Sixers – story of the numbers

Last night I was in the crowd watching the Sydney derby, the opening match in the 2015-16 Big Bash League, which meant I wasn’t in a position to do a deep dive into the stats of the night.

My plan is to use the Twitter account in conjunction with a liveblog for nights when I’m able to watch the match with my computer in front of me – which won’t be every match by any means – and when I can’t I’ll try to come back with a summary of some key stats which help you understand the match.

Last night the Sydney Thunder beat the Sydney Sixers for the first time in eight outings, after the women achieved the same two weeks ago in Penrith. It was the first Sydney Derby played at the Showground Stadium at Olympic Park after the Thunder downgraded from the cavernous Stadium Australia to the smaller stadium across the road – and it was a big success with three-quarters of the seat’s filled and a good vibe.

Taking wickets was the key in this match – the Thunder’s score at the end of their innings was predictable based on the small number of wickets they had lost by their halfway point (allowing them to score faster), but only gave them even odds of winning – it was the steady fall of key Sixers wickets which pushed the Thunder total out of reach.

Continue reading “Thunder vs Sixers – story of the numbers”

WBBL – after two weeks

We’ve now had twelve matches in the Women’s Big Bash League, with the three matches in Melbourne and Sydney on the first weekend followed up by another nine matches in Perth and Launceston last weekend.

While the Sydney Thunder have only played one match, the Brisbane Heat have played six matches, so the results between the teams are still quite lopsided. So far three teams have won all their matches, three have lost all their matches, and two have mixed results.

The winning Thunder and Stars sat out the weekend, while the losing Heat and Sixers returned and were joined by the other four teams.

In summary, this is where each team stands:

  • Stars and Thunder have impressed, but we haven’t seen them this weekend.
  • The Heat lost three close matches, and have bounced back with impressive performances which suggest they will be a team to beat. They’ll still be weighed down by those three losses.
  • The Hurricanes have won all four of their matches, putting them at the top of the table, but two of those were by the slimmest of margins.
  • The Strikers and Renegades have each lost two matches – in each case, they’ve lost one narrowly and one by a large margin.
  • The Scorchers had one easy win, one narrow win and one loss.
  • The Sixers have been disappointing, losing all four of their matches, usually by large margins.

Below the fold I’ll run through the most impressive individual performances.

Continue reading “WBBL – after two weeks”

BBL – the best batsmen and bowlers

In this post, I will run through the top batsmen and bowlers using a set of seven metrics I have previously used. Yesterday I ran through what these metrics look like for the entirety of each squad, but it’s worth seeing who the stars are.

Like I did for my equivalent WBBL post, I will split these lists between players who have played at least five T20 internationals, and those who haven’t.

It’s also worth noting that the BBL will be played concurrently with international cricket in Australia. Most of the Australian test squad are signed to one team or another, but that doesn’t mean they will be available. Cricket Australia planned to free up a large number of test players to be available for the early pre-Christmas matches, but that is now looking to be limited.

Where a player is likely to not be present I will note that, and for that reason I’ll list ten players each in the international category to include people who are likely to be present. In particular Steve Smith and Usman Khawaja have withdrawn from playing for the Sixers and Thunder respectively in the opening match to give them more recovery time.

Continue reading “BBL – the best batsmen and bowlers”

BBL – each squad’s batting/bowling stats

I had a bit of writer’s block thinking about how to preview the BBL season, which starts on Thursday. There’s so much potential data about the players, and it’s hard to decide what is likely to provide new insights, rather than just demonstrating what is already conventional wisdom.

I decided to follow roughly the same model as I used for the WBBL. So I’m going to post here today the key batting and bowling stats for each squad selected for the BBL. I’m also going to do some analysis of what the equivalent numbers looked like prior to last season to see how useful it is in predicting outcomes.

Tomorrow I will post about the key individual players to watch, according to these same statistics.

Continue reading “BBL – each squad’s batting/bowling stats”