5th January, 2018. In addition, I decided to omit some other valuable data, either general or league-specific, such as whether teams should focus their bowling attack on pace or spin, batsman/bowler match-ups, age curve analysis - let's not forget, players over 30 questionably command higher average salaries than younger players - and specifics on overseas players, particularly whether these overseas players have the skill-set required to thrive in likely conditions against likely opposition. I have this data privately, but decided not to publish it, with it being a little too sensitive to be publicly available currently. Any interested parties should contact sportsanalyticsadvantage@gmail.com for further information on this, or bespoke recruitment strategies. The next thing I'd like to mention is that while the title of this article is 'How to win a T20 league', the effective title should be considered as 'how to qualify for the knockout stages of a T20 competition while giving yourself an excellent chance of winning it' - a title which is not quite as catchy! The reasoning behind this is very similar to the comments in Michael Lewis' excellent book, 'Moneyball', where the stated objective was to reach the play-offs (while obviously having a decent chance of winning the title), as opposed to actually winning the league. Primarily this is due to the fact that it is extremely rare to have a greater than 60% chance of winning a one-off T20 match against another team which has also qualified out of the group stages (by definition, almost certainly an above-average T20 team themselves in that particular competition). Therefore, mathematically speaking, it is extremely unreasonable to give a coach an objective of being champions when variance in the knockout stages of T20 events is so prevalent, although many franchise leagues give the group winner two chances to make the final, which assuming two 60-40 match-ups, gives a team top of the group a chance in excess of 80% of making the final. To establish the requirements to win a T20 league, I worked out the mean figures for a variety of metrics across the last two editions of the major T20 domestic leagues (Big Bash, Pakistan Super League, Indian Premier League, T20 Blast, Caribbean Premier League, Bangladesh Premier League), listed below:- Players Used Bowlers Used % Squad Used As Bowlers Team Batting Average Team Batting Strike Rate Team Boundary % Team Bowling Average Team Bowling Economy Rate Team Domestic Player Batting Average Team Domestic Player Batting Strike Rate Team Domestic Player Bowling Average Team Domestic Player Bowling Economy Rate I then further established mean figures for each metric split by a teams that came in the top two of their league, teams which qualified and teams which failed to qualify, which enabled me to assess the statistical differences between the three categories, and in addition, crucially identify which metrics drive success and failure more than others. These findings are below:-
From the data above, there are a number of strategies that can be developed by T20 teams looking to gain an edge over their competition. Strategy 1 - Use a small, high quality squad. Teams that came top or second in their league used an average of 17.42 players in their season, a figure which rose to 17.79 for all qualifying teams, and higher again to 18.42 players for teams that failed to qualify. Therefore on average, teams who failed to qualify used exactly 1 player more on average than teams who came in the top two of their league. It's probably fair to suggest that views on whether using numerous players are quite polarised. There will be some who believe - like myself - that having a big squad can be useful, although there is one large caveat. This caveat is simple - the players in a big squad should have wide-ranging skill-sets, so that teams can make use of them when conditions see fit. For example, a Big Bash T20 team in the pace-friendly Adelaide might look to go with their standard pace bowlers, but against Brisbane Heat (a team weak against spin) at the Gabba (a ground where spin has been much more economical in recent years), they may wish to pack their team with spinners, or at least spin options among batsmen that wouldn't usually get much, or any of a bowl, so that expectation in bowler/batsman match-ups can be maximised. The problem is that many T20 teams who use more players than average in the season don't tend to do so for such strong strategic decisions, and I'd suggest that it is more likely that they do so because they have little idea what their best team is, or their recruitment was so bad to start with. In addition, a lack of understanding regarding whether player success or failure is down to form, ability, variance, or poor batsman/bowler match-ups hampers teams in this regard. Of the 15 teams which used 15 or fewer players in their season, three won their leagues, three came runners up and nine (60%) qualified from the group stages. Contrast this with the lack of success of the ten teams who used 22 players or more. Just four - RCB (IPL 2016) Middlesex (T20 Blast 2016), Comilla Victorians (BPL 2017 - used extra players in later group stage matches having already qualified) and Rangpur Riders (BPL 2017) managed to qualify. Five of the ten teams came 6th or worse in their league. Strategy 2 - Use your best bowlers more Teams who qualified used 64.19% of their squad as bowlers, while teams who did not qualify used 66.94%. Similar to the point I made above about team composition, using large numbers of bowlers is fine, as long as it's for solid, thought out reasons, such as a usual batsman who bowls mediocre spin being matched up as a bowler against a batsman much weaker against spin than pace. The problem is, most captains don't tend to do this, and instead make bowling changes for variety or out of desperation. If every captain was to think 'is this bowler the best option for the particular batsmen at the crease, assuming the game state of the match' when making bowling changes, they'd be much more successful. Teams which used ten or fewer bowlers throughout their season qualified from their leagues 17 out of 27 times (63.0%). Strategy 3 - Solid top order batsmen are undervalued It's important not to confuse solid for slow here - a player with a poor strike rate will never be of benefit to a T20 team - but there was a marked difference between the batting average of teams who were successful and those who were not. Teams who came in the top two of their leagues had a batting average of 26.01, falling to 25.27 for teams who qualified, and then a marked drop to 23.33 for those teams who failed to qualify. Teams who scored more than 28 runs per wicket across the last two season are listed below:-
As you can see only one team of the 16 - Rising Pune Supergiants in the 2016 IPL - came below fifth in their league while 11 qualified (68.8%). While there are some cases where wickets are overvalued by teams and players, it's evident that getting off to a strong start in innings is mandatory and generally losing few wickets will be a positive strategy. Compare this to the 14 teams which averaged fewer than 20 runs per wicket:-
Of these, only four teams qualified (25%) and only one team came above 4th in their league. Evidently, Rajshahi Kings in the BPL, having achieved this ignominy in consecutive years, may wish to revise their recruitment strategy. Strategy 4 - Batting Strike Rates isn't a critical driver of success Teams which came first or second struck at 130.23 runs per 100 balls, and this just fell to 128.14 for teams who failed to qualify. As the table below illustrates, teams which struck at over 135 runs per 100 balls didn't fare magnificently:-
Of the 19 teams, just 10 qualified (52.6%), and of the teams with poor batting strike rates (below 120) had similar results, with 10 of the 18 teams qualifying despite poor strike rates, as can be seen below:-
Strategy 5 - Elite Boundary Hitters are worth recruiting Of the 25 teams who recorded mean boundary percentages in excess of 17%, 15 qualified (60%). Four won their group, with a further six being runners up - therefore 40% came in the top two of their competition, shown below:-
The landscape for teams with below 14% boundary scoring was not nearly as positive:-
Of these 16 teams, just five qualified, and only two came in the top two. Nine teams came 6th or below in their leagues. Strategy 6 - Quality bowlers are the biggest driver of success, and teams need to ensure that they recruit better than their rivals in this area Generally speaking, teams should prioritise their resources towards finding the best bowlers, as opposed to the best batsmen. Many more teams who had much better bowlers than batsmen succeeded, than vice versa. Teams who qualified from their leagues managed bowling averages of 25.56, while the numbers of those who failed to qualify dropped markedly to 29.67. Significantly, teams whose bowling averaged below 25 had a superb chance of qualifying, as the table below illustrates:-
Of these 34 teams, 27 qualified (79.4%) and there were eight group winners (23.5%). In short, if a T20 team's bowling averages is sub 25, they have almost double the chance of winning their group than a random team. In addition, a further eight teams with this bowling success were runner up, so 16/34 (47.1%) of these teams came in the top two of their particular competition - again almost double the likelihood of a random team. Dhaka Dynamites in the BPL look to have this strategy absolutely nailed, twice coming in the top 3, with only Glamorgan in the T20 Blast in 2016 being superior. Adelaide Strikers, Peshawar Zalmi, Rangpur Riders, Perth Scorchers and Islamabad United also featured in this list twice. Strategy 7 - Bowling Economy is less important than wicket-taking, unless it is at either extremes I've made this point several times previously, and the numbers bear this out again here. Teams that qualified from their group had a mean deviation of 1.02 here, with teams failing to qualify recording 0.98 - not much of a difference at all. However, this does come with a huge caveat. As the numbers below show, a team bowling economy rate below 7.50 is a huge driver of success, and the five teams who were sub 7.20 all qualified.
Of these 20 teams, just Rangpur Riders and Comilla Victorians, both in the 2016 BPL, failed to qualify. So - 18 teams with excellent bowling economy rates qualified (90%), and this was as good a guarantee of qualification than any metric. Conversely, the worst teams hugely struggled:-
Of these 21 teams, only five qualified (23.8%) and 12 came 6th or below in their league. It would appear that Brisbane Heat (Big Bash), Somerset, Kent and Worcestershire (all T20 Blast) need to make vast improvements to their bowling attacks, with this quartet featuring twice in the worst bowling economy list. Strategy 8 - Make sure your domestic batting isn't a liability Largely a BPL & PSL issue, teams which had poor domestic batting averages were at a disadvantage to their rivals, as witnessed below:-
This also went for the domestic player batting strike rate:-
Of the 17 teams which recorded a domestic player strike rate of 115 or below, only three qualified in the top two of their group. Strategy 9 - Teams should use a significant part of their resources to sign good domestic bowlers As the stats below show, teams who had a bowling average for domestic players of below 25 had a great record for qualifying:-
Of these 32 teams, 24 qualified (75%) and 20 (63%) came in the top three of their particular leagues. Teams with domestic bowlers much worse than average did not record nearly as much success, with just two qualifying:-
These findings about domestic bowlers were also backed up by analysis of their economy rates:-
Of the 14 teams who had domestic bowling economy below 7.50, 13 qualified, and 8 came in the top two of their group. Conversely, the teams who had domestic bowling economy in excess of 8.65 found it very difficult to qualify, with just two doing so:-
At this stage, it's worth making some further observations. In team drafts, it is very commonplace to see teams picking up barely above average overseas players in early rounds, a strategy which all this data disputes. Teams should look to adopt the reverse strategy - pick up as many good domestic players as possible early in the drafts, before switching their attention to overseas players towards the end. There are so many good overseas players available, and left unsold, that waiting for overseas players does not yield a big disadvantage. All the data above clearly demonstrates that it is domestic players (which comprise between 6 and 9 players in each team, depending on the league) are the big drivers to success. Overseas players are merely the icing on the cake. League Specific Observations As well as being able to identify the above strategies, a further area worth looking at is where teams in each league can improve. Big Bash Teams used fewer players than average (16.94) and fewer bowlers (10.75), while teams managed marginally better batting data and worse bowling data than average, particularly from a strike rate perspective. Domestic players batting was very strong (24.46 average, 129.53 strike rate) indicating that generally, Big Bash teams have little need to sign marquee overseas batsmen (although many do!). Instead, they should concentrate on improving their bowling - domestic Big Bash bowlers averaged in excess of 30 and went for 8.30 runs per over in this sample. Very few batsmen played match losing innings. On average, each team had just 0.75 batsmen with a strike rate of 115 or below when facing in excess of 80 balls. With the exception of overseas recruiting, Big Bash teams look to be better than average regarding strategy. Pakistan Super League Teams used marginally fewer players (17.80) than average and considerably less bowlers (10.80) which led to a final bowler percentage of 61.45%, around 4% less than the worldwide mean. Batsmen struggled in this competition, averaging just over 22, striking at barely 120 runs per 100 balls, while bowlers thrived, recording much better numbers than average. As mentioned earlier, domestic batsmen had an awful record, averaging 20.04 at a sub 120 strike rate, but domestic bowlers did well, averaging just over 26 at economy around 7.50. Each team had an average of 1.90 players in their squad with a batting strike rate below 115, facing at least 80 balls. Match losing innings, therefore, were common. With all the above considered, teams should focus their recruitment on domestic bowlers and overseas batsmen. Indian Premier League Teams used more players than average (20) which is understandable given the longer format of this particular competition. 13.38 bowlers were used by each squad (66.7%). Batsmen thrived in India, averaging almost 27 at a strike rate above 132, hitting almost 17% of balls bowled for boundaries. Bowlers fared less well, averaging over 30 at an economy rate of 8.25. Domestic batsmen did fine, averaging 27 themselves, but at a lower strike rate of 128, showing that perhaps the benefit of overseas batsmen here is an ability to score quicker than domestic players. As with most bowlers in this competition, domestic bowlers performed worse than average. Match losing innings were uncommon. Just 0.56 players per squad had a sub 115 strike rate when facing 80 or more balls in a season. As arguably the best T20 franchise league, it's unsurprising that this poor standard is not tolerated by teams. Given that domestic bowlers struggled in general, I'd be keen to look to recruit good overseas bowlers, as opposed to batsmen, but with overseas options plentiful in this competition, it isn't a strong lean. T20 Blast A competition running throughout most of the English summer, it is arguable that quality is diluted by both the number of competing teams (18) and the needs of the English national team. This tournament features high average strike rates in excess of 130, as well as the related high boundary percentages. Team bowling economy isn't pretty, at an average of just over 8.30. Domestic batsmen did fine, averaging almost 24 at a strike rate over 130, but domestic bowlers didn't fare superbly. Despite this competition featuring comparable strike rates to the IPL, there were many more (1.03 per squad) players capable of match losing innings - perhaps a nod to the diluted quality of the event. Given all the above, I'd generally look at recruiting either overseas bowlers or bowling all-rounders for T20 Blast teams. Caribbean Premier League Teams used few players in this event, making use of just 16.92 per squad on average. Batsmen did not perform well, averaging below 24, at a sub 125 strike rate, while bowlers found conditions to their liking, with a superb average economy rate of around 7.90. However, domestic bowlers did not perform as well, with an economy of over 8.20, so with this in mind, it would appear that teams should focus their overseas recruitment on overseas bowlers. Having said this, an average of 2.33 batsmen per squad had a strike rate below 115 when facing 80 balls - a woefully high figure - so coaches should impress on their players that they need to hit out or get out much more frequently. Bangladesh Premier League Teams used many more players than average (20.21), and also many more bowlers (14.29), leading to teams using 70.9% of their squad as a bowling option. Having just watched this tournament, I am convinced that this was through lack of ideas than tactical nous. The other traits of this event are basically a more extreme version of the PSL - batsmen performed poorly, averaging 22.4 with a strike rate around 120, while rather predictably based on this, bowlers generally did superbly. If an overseas T20 bowler (particularly a spinner) wants to look better than they actually are, they should go and play in Bangladesh. A feature of this competition is how poor the domestic batsmen are from a strike rate perspective, striking at barely 110 runs per 100 balls, but domestic bowlers are adept in their home conditions, averaging sub 27 at an economy around 7.7 runs per over. With such poor general batting data, it is also unsurprising that 2.43 players per squad had a strike rate below 115 when facing over 80 balls in the competition. Evidently, with domestic batting so poor and five overseas players allowed in the competition, franchises should pack their teams full of overseas batsmen and domestic all-rounders and bowlers. |