Two notable events in the cricketing calendar have taken place in recent weeks. Firstly, the inaugural T20 Global League (South African Franchise League) player draft took place at the end of August, and secondly, the T20 Blast Finals Day in England was played, concluding the English domestic T20 league for another season. As I've mentioned in previous case studies and articles, the T20 Blast in England is the easiest major domestic T20 league to score runs in, both from an average and strike rate perspective. Therefore, overseas players should expect to have their batting data from other domestic T20 leagues boosted if they played in the T20 Blast (Colin Ingram is a great example) while English batsmen should expect their English batting data to decline if they play in other countries. We also witnessed a significant rise in both overall scoring plus boundary hitting in the T20 Blast in 2017, as the table below (showing the mean batting average and strike rate for the last three seasons) illustrates:-
That the Blast is enjoying higher scoring rates isn't entirely a surprise given the increasing T20 scoring rates throughout the world, but these mean figures cannot be replicated in other countries. There are two seasons left of the Blast in its current guise before the English franchise league starts in 2020, and the Global League draft is likely to be a similar format to the future English league, which is stated to be a draft format, as opposed to an auction. However, as expected, recruitment mistakes in the Global League were plentiful, with teams making particular extremely negative value choices early in the draft. Also apparent was how badly prepared some teams were in the draft, with there being many examples of teams needing to use up all their time allowance to select players, or having to make frantic telephone calls at the last minute. Looking at the first two rounds in isolation, some of the selections I'd disagree with are as follows:- Morne Morkel - 1st Round - Pretoria Mavericks - T20 data isn't nearly as strong as other formats Chris Jordan - 2nd Round - Bloemfontein Blazers - Adds value as a bowler, but awful batting strike rate & boundary hitting percentage, and using such an early choice on a non-elite overseas player is not recommended. Wahab Riaz - 2nd Round - Benoni Zalmi - Certainly not the worst bowler in the draft but prone to inconsistency and expensive spells - again, a non-elite overseas early pick. Adil Rashid - 2nd Round - Cape Town Knight Riders - Very solid T20 spinner but batting strike rate & boundary hitting far below average. Again a non-elite overseas early selection. Sam Billings - 2nd Round - Durban Qalanders - Barely above average batsman and below-average wicket-keeper. Detailed analysis of Billings follows later in this case study. Vernon Philander - 2nd Round - Johannesburg Giants - Has barely played T20s in the last few years. A huge gamble on that basis. Aiden Markram - 2nd Round - Nelson Mandela Bay Stars - Another with little T20 experience, and the numbers he's recorded so far don't particularly demonstrate he'd be a big asset in this format. To avoid such recruitment mistakes, detailed player analysis is critical for success and it is certain that preparing with a strong statistical player dossier would enable a team to be assembled with much higher expected value than the average franchise, given equal financial resources. With this in mind, a useful exercise would be to look at T20 Blast data over the last three years and establish which players offer more value than average. Effectively, this would generate a solid list of players which - if the draft was held now, as opposed to 2020 - would well equip a team to succeed in the upcoming English Franchise league. Please note that only England qualified players have been assessed (only three overseas players will be allowed in the new English T20 league), and that player data is weighted in numerous ways based on the Sports Analytics Advantage algorithm. Primarily this values more recent performances more than older ones (so for example, a player's last innings in 2017 will have more weight than their first innings in 2015), and also situational difficulty. All of a player's worldwide T20 performances, both in domestic & international cricket, are used to create this data. Expected Average and Expected Strike Rate refer to expected data for the next season (2018 T20 Blast). Above-Average English T20 Batsmen (sorted by Mean Deviation) (Minimum 400 Balls Faced, Mean Deviation greater than or equal to 1, Mean Deviation is weighted with a split of 40-60 between average and strike rate)
It's probably not a huge shock to see three renowned names at the top of the list - England captain Joe Root, veteran T20 specialist Kevin Pietersen, and renowned white-ball wicket-keeper batsman, Jos Buttler. Root hasn't played much T20 in the last few years, but when he has, he's scored prolifically, albeit at a not particularly elite-level strike rate, while Pietersen follows a similar profile although more strike-rate orientated. Buttler, on the other hand, is much more strike-rate orientated. Some of the surprise names at the top of the list include Adam Lyth, who scored a record 161 for Yorkshire recently against Northants, and even with that innings excluded, Lyth's data is still excellent. His boundary percentage of almost 25% is elite level and he should be considered both for national team selection as well as a franchise T20 cricketer. Also worth mentioning are the Kent duo Daniel Bell-Drummond and Sam Northeast (clubmate Joe Denly isn't far behind after a magnificent 2017) while Dawid Malan's data indicates England are keener on him in the wrong format - he's a much stronger white-ball batsman. Ian Cockbain and Chris Nash are two rather unheralded names in the top ten English batsmen. It's also interesting to witness how low some of England's batsmen are rated using our algorithm - captain Eoin Morgan, with a solid 28.60 expected average but very poor 119.90 expected strike rate (plus woeful 11.74% boundary percentage) rates 39th, and wicket-keeper batsman Sam Billings is 40th, mainly due to an underperforming average. Despite strong data, Alex Hales doesn't even make the top 20, and nor does batting all-rounder Ben Stokes. Certainly, questions should be asked of Morgan and Billings' national team selections in the T20 format. There are a variety of batting options that English franchises could consider, and apart from the few at the top of the list, many have similar levels of ability. Breaking the bank or wasting an early draft pick on the likes of Morgan or Ian Bell is not recommended. Wicket-Keeper Analysis There hasn't been much statistical analysis performed on wicket-keepers anywhere in the public domain, although it's probably fair to suggest that some internal analysis has been performed by some teams. Primarily, analysis of wicket-keepers tends to have previously been surrounding their batting, which is easier to quantify. Analysing wicket-keepers is made even tougher by several factors which are difficult to quantify. 1 - Some keepers will stand up to faster bowlers more, in theory reducing the batsman's propensity to charge the bowler. However, given the increasing nature of aggressive batting in T20, this should be less of a factor than in previous years. 2 - It's very difficult to quantify the difficulty of each catching or stumping chance that the wicket-keeper receives. However, it is extremely unlikely that one wicket keeper will, over the course of several seasons, get abundantly more easier chances than another wicket-keeper. Using an example from soccer, this is much different to strikers who will frequently get much easier or more difficult chances based on their particular style of play. Despite this, in this excellent article by Tim Wigmore, he mentions a conversation with the analyst for Northants, Richard Barker - one of the pioneers of Moneyball recruitment in T20 cricket - and Wigmore writes 'Data could also salvage one of cricket’s most cherished lost species: the specialist wicketkeeper. Lower-order batsmen face so few balls – the average No7 faces seven balls an innings – that Barker believes picking the best keeper could be prudent.' It's an interesting concept to address, and the data below does just this. There are 17 English wicket-keepers still active in T20 who have kept for more than 220 overs in the T20 Blast between 2015 and 2017, and between them when batting, they faced a combined 6,087 balls across 417 innings (whether out or not out), which averaged out to a much bigger 14.60 mean balls faced per innings - it is likely Barker included the occasions where the number seven did not bat at all. Of these, Ben Foakes faced the lowest average balls per innings (9.69) while Jos Buttler, easily good enough as a specialist batsman alone, rather unsurprisingly faced the most (22.27). What was also apparent from the table of these wicket-keepers is how little the combined bye plus leg bye count varied among keepers. A metric - BP20 (bye/leg byes per 20 overs) - was created to grade each wicket-keeper in how many byes they conceded, and the worst of these wicket-keepers for this was Sam Billings (4.59 per 20 overs) while Steven Davies (2.73) was the best, with the mean being 3.65. However, even the best, Davies, managed to save just 1.86 byes/leg byes per innings fewer than Billings, so purely on bye restriction, it would be ambitious to give much credit for picking the best wicket-keeper solely on this basis. Another metric - CP20 (catches/stumpings per 20 overs) was also created, with the best wicket-keeper for this metric being Jonny Bairstow (1.14 catches/stumpings per 20 overs), and the worst, James Foster at 0.59. The mean figure was 0.79. This metric carries much more weight. The average cost of a wicket in the T20 Blast in 2017 was 24.83 runs, so a wicket-keeper who dismisses 1.14 batsmen per innings is immensely more valuable than one who can dismiss just 0.59 batsmen per innings. With the average wicket-keeper in this sample taking 0.79 catches/stumpings per 20 overs, we can quantify the average wicket with wicket-keeping intervention as worth 19.62 runs (0.79 * 24.83), and a wicket-keeper with a CP20 of 1.00, for example, would have a positive CP20 of 5.21 runs ((1.00-0.79) * 24.83). Given this, we can rank the 17 current English wicket-keepers with a big enough sample size purely on their keeping metrics alone, and this is below:-
The top five keepers in this table have considerably better data over the rest, but it's also worth noting that purely on keeping metrics alone, the best T20 English keeper (Bairstow) is only 8.63 runs better than average. Given this, elite specialist wicket-keepers would have some value but with the margin between top and average being relatively small, it wouldn't indicate that a team would be able to play a specialist wicket-keeper with little batting value. In fact, four of the top five wicket-keepers using these metrics also had positive batting data, as can be seen below:-
As mentioned earlier, wicket-keepers, on average, when they walk out to bat, face 14.60 balls. The Batting Value column reflects the number of runs expected for each batsman from these 14.60 balls. Net batting value indicates the difference between a player's batting value and the average wicket-keeper value of 18.87 runs from these 14.60 balls. Obviously specialist batsmen such as Buttler will face more balls on average, but it's a fair and useful measure to compare batting ability among wicket-keepers. Assessing the data, we can see that eight of the 17 batsmen have above average batting data (they outperform the average batsman in the T20 Blast), and four of these, Bairstow, Cooke, Wheater and Buttler) were also highly regarded as wicket-keepers alone. Only the Sussex keeper, Ben Brown, recorded poor batting numbers (primarily due to strike rate and boundary hitting) but high wicket-keeping numbers, and the wicket-keepers with mediocre batting data didn't generally record good keeping data either. Clearly, this gives more relevance again to disagreeing with Barker and picking a batsman who can keep wicket - particularly one who can bat with a high strike rate - gives a batting team an abundance of options. Having a wicket keeper who can justify inclusion in a team solely on batting numbers alone opens up opportunities for picking another batsman, a late-order specialist slogger or perhaps including another bowler to give further bowling options. Picking a specialist wicket-keeper does not allow that luxury. Adding the wicket-keeping and batting values together creates a final metric - Keeper/Bat Value, and the 17 wicket-keepers are ranked according to this below:-
It's probably not a shock to see Buttler at the top of this list, given his obvious value in the format, but it's more interesting to see the likes of Chris Cooke and Adam Wheater - two keepers who haven't even kept in the majority of matches in the last couple of years - as highly rated. Certainly, it seems baffling to think that Wheater is not keeping for Essex, given James Foster's poor performances in these metrics, and these numbers would indicate that picking Wheater over Foster would yield Essex a net of 16.48 runs per match (Wheater's positive value is 7.31 runs, Foster's negative value is 9.17). Tom Moores is an interesting player to look at - his keeping value was solid and his boundary hitting percentage when batting (see further down in this article) is also very strong. Given he is just 21 years of age, it's logical to think that he will improve further and he could be a real asset in the future as a wicket-keeper batsman. He doesn't look a number three, as he's been batting at Nottinghamshire, but he could be a really decent mid-late innings hitter. It's also interesting to note that Barker's wicket-keeper, Adam Rossington, has poor keeping data but excellent batting data (he's another boundary hitter). Northants have also toyed with using Ben Duckett as a keeper occasionally, and his data for byes conceded was one of the worst recorded (although filtered out due to sample size). Summarising, in any draft, Buttler should be a top pick for any franchise, although the likes of Cooke, Wheater and Moores could be very solid back-up options or cheaper choices. Bowler Analysis (Minimum 60 overs bowled 2015-2017, Mean Deviation greater than or equal to 1, Bowling Mean Deviation is weighted with a split of 50-50 between average and economy rate) There were numerous English T20 bowlers with better than average data. In the Blast in 2017, each wicket cost a bowler 27.4096 runs, and the average economy rate was 8.4587 - note this is different to the mean wicket and strike rate figures earlier, which have run outs (not attributed to bowlers) included. All bowlers were graded via their expected average and economy rate if the Blast took place next season, and then related to how much better than the mean they deviated. So for example, if a bowler's expected average was 20, their average deviation would be (27.4096/20) = 1.37. The list of positive expectation bowlers is as follows, ranked in order of mean deviation:-
My previous work at cricketratings.co.uk established that Benny Howell was an elite-level T20 bowler, and he's been able to maintain these numbers, being head and shoulders above any English bowler in T20 based on my algorithm. It's utterly bizarre that he's not got any national team recognition in this format. The veteran, Rikki Clarke, was snapped up by Surrey this year at the age of 35, is still going strong, and most England bowlers - Steven Finn, Liam Dawson, Tymal Mills, Chris Woakes, David Willey, Chris Jordan and Adil Rashid - are in and around the top 10. Another England bowler, Liam Plunkett, is lower rated but his lower-order hitting is a useful asset and Steven Mullaney is another who falls into this bracket. Young prospects in this list include Mason Crane and the Curran brothers, with Tom edging brother Sam slightly using this metric. Another metric we can use to grade bowlers is something I've named 'FOC', which stands for four-over contribution. Essentially, using expected bowling averages and economy rates, we can work out how much a bowler contributes more than the average bowler in their four overs they are allowed to bowl. Using Howell as an example, with an expected economy rate of 6.47, he'd be expected to go for 25.88 runs per four overs (RP4 figure -> 6.47*4), and considering his expected average is 17.28, he'd be expected to take (25.88/17.28) = 1.50 wickets. Earlier on, we established that the average bowler went for 8.4587 runs per over, and took wickets at an average of 27.4096 in the T20 Blast in 2017, so the average four over-figures would be 33.83 runs conceded, taking 1.23 wickets. Similar to wicket-keepers CP20 metric, WP4 for bowlers is the number of dismissals a bowler is expected to yield per their maximum four overs bowled in a match. Again, using Howell, we can then establish that he'd take 0.27 (1.50-1.23) more wickets than the average bowler in his spell, which is worth (0.27 * 27.4096) = 7.40 runs as value to his team. Also, Howell is expected to save 7.89 runs per four overs. Adding this together, we can state that he has a positive FOC of 7.40 (wickets) + 7.89 (economy) = 15.29 runs over the average T20 Blast bowler. The same bowlers were sorted using FOC, with only a few players experiencing major differences to their weighted mean numbers:-
Batting ability for bowlers is also a growing benefit, particularly from a strike-rate perspective, allowing for some quick runs at the end of an innings. Of the bowlers listed above, the following had strong expected strike rates:- David Willey, 152.68 Steven Mullaney, 145.65 Paul Coughlin, 142.46 Liam Plunkett, 141.29 Tim Bresnan, 136.84 Aaron Lilley, 136.13 Tom Curran, 133.71 Moeen Ali, 131.80 In drafts, it is highly likely that marginal decisions will need to be made between bowlers of relatively similar levels of ability. Bowlers who can bat, particularly at a high strike rate, will be increasingly viewed as assets, and even a 10 (6) type innings can have immense value. All-Rounder Analysis (Minimum 35 overs bowled 2015-2017, Expected Batting Average greater than or equal to 12, Minimum 250 balls faced, Batting Mean Deviation is weighted with a split of 1/3-2/3 between average and economy rate) According to Wikipedia, when defining an all-rounder, 'The generally accepted criterion is that a "genuine all-rounder" is someone whose batting or bowling skills, considered alone, would be good enough to win him a place in the team. Another definition of a "genuine all-rounder" is a player who can through both batting and bowling (though not necessarily both in the same match), consistently "win matches for the team" (i.e., propel his/her team to victory by an outstanding individual performance). By either definition, a genuine all-rounder is quite rare and extremely valuable to a team, effectively operating as two players.' Wikipedia certainly is correct - a genuine all-rounder is quite rare. Using our algorithm, only eight English players in the T20 Blast - David Willey, Moeen Ali, Paul Coughlin, Ryan Higgins, Ravi Bopara, Samit Patel, Tim Bresnan and Wayne Madsen had mean deviations at or above average for both batting and bowling, while Dan Lawrence, Benny Howell, Tom Curran and Keaton Jennings also did when comparing NBV and FOC (see below). The same way as we looked at net batting value (NBV) for wicket-keepers - the all-rounders in this sample faced an average of 14.19 balls instead - and four-over contribution (FOC) for bowlers, we could use both these metrics to assess all-rounders, combining for a final ARM (all-rounder measure), and the all-rounders who fit into the filtered criterion above are listed below, sorted by ARM:-
Quite surprisingly, 20 year-old Essex all-rounder Dan Lawrence topped this table, with Benny Howell's superb bowling numbers putting him in second place in the all-rounders table, to go with his first place in the bowling charts. Most of the players at the top of this list were also in top bowlers list, while the players at the bottom of the list had poor bowling data, and should not be considered as viable bowling options generally - Jack Taylor however, certainly has some very strong batting potential. England all-rounder Ben Stokes had one of the best all-rounder batting numbers, but his bowling, largely due to economy rate issues, had a negative contribution. Tom Curran, as mentioned in the bowling analysis, is a young player of huge potential. The 22 year old Surrey all-rounder has solid batting data (much better than brother Sam) to go with above-average numbers as well. Having looked at batsmen, wicket-keepers, bowlers and all-rounders, there is a clear list of players who should be treated as priorities in drafts, and those who are of less worth. It's also useful to look at these players, the 'wild cards' - perhaps players who didn't quite fit into these brackets, or those who have outstanding other numbers, or young players who have small sample sizes. Some of these players are likely to fly under the radar in auctions and drafts and could be picked up relatively cheaply, or in a latter bidding round. Boundary Hitters (Batting Average greater than or equal to 10, Boundary hitting percentage greater than or equal to 18%, at least 150 balls faced):-
Young Warwickshire batsman Ed Pollock is by far the outstanding player in this list, and only missed the top batsman list due to not having faced enough balls. Considering the overall boundary % in the Blast from 2015 to 2017 was 16.33%, scoring boundaries in over 27% of balls faced is truly incredible. There is no doubt that if Pollock can maintain these numbers, he will be a worldwide T20 superstar. The bowlers and all-rounders on the list should be considered as assets, particularly when coupled with high strike rates, with these players able to provide valuable impetus towards the end of an innings. Looking at the data, there are very few out and out lower order specialist sloggers, in the mould of Carlos Brathwaite, and perhaps there is an opportunity for a few of these players to improve their strike rates to try and develop into this sort of player, given their obvious abilities to find the boundary. Certainly, it would be worth questioning those with a below-average boundary percentage (especially those below 13%), and those players with a poor boundary hitting percentage are likely to lose a team value in the latter stages of innings. Ideally these players would be avoided in a draft unless they have over-riding bowling ability. Batsmen with Small Samples (sorted by Mean Deviation) (Mean Deviation greater than or equal to 1, Mean Deviation is weighted with a split of 40-60 between average and strike rate) The batsmen below would have been included in the top batters analysis, if they had faced 400 balls:-
The aforementioned Pollock tops this list by some distance, with Graham Wagg following. Wagg's an interesting player to look at, given his strong average but poor strike rate and boundary hitting percentage. He's been batting in the lower-middle order for Glamorgan, and there looks to be a case for him batting either early in the innings, if an early wicket falls, or to stabilise an innings coming in at a score like 60-4. Again, this illustrates the value in a fluid batting order, as opposed to a fixed one - something that I've mentioned numerous times in previous case studies and articles. Young Prospects Signing domestic young prospects as a franchise represents strong value, with these players more likely than overseas players to show loyalty and availability. Furthermore, there is the potential for longer term contracts to be agreed prior to a player becoming high profile, which would represent a long-term cost saving compared to the standard salary for a high-profile player. The following players showed some ability, but for various reasons (not yet quite good enough, sample size) didn't fit into any of the previous categories:-
Of these, Hampshire wicket-keeper Calvin Dickinson certainly looks one to watch if he can maintain this strike rate and boundary hitting percentage, while Adam Hose is another in this regard. Hose's team mate at Warwickshire, Aaron Thomason, could forge a solid T20 career if he can reduce his very poor bowling economy expectation - his batting data is decent for a generic number 7/8 type batsman. Derbyshire spinner Matt Critchley is another who has very nice all-round potential, while fellow spinner Matt Parkinson (not on this list due to very small sample size) at Lancashire has immense potential based on what numbers he has currently recorded. Dominic Sibley will need to dramatically improve his expected strike rate and boundary hitting percentages if he isn't going to risk playing match-losing innings. An expected average in excess of 36 is great, but if it takes 30 balls (as his strike rate would indicate), that is 1/4 of a team's entire 20 over resource gone while scoring just 36 runs. OVERALL SUMMARY:- Very few English batsmen are elite-level - there are many English batsmen who are better than average, but with little to differentiate between them. Picking a wicket-keeper who adds value as a batsman is much more desirable than a specialist wicket-keeper. Few bowlers have a four-over contribution which has over 10 runs of value - these should be primary draft selections, although not all of these players are household names. Quality all-rounders are a rare commodity, and most of those who have outstanding data tend to be bowlers who bat. Picking batsmen who bowl is frequently a false economy. Selection of boundary hitters is an excellent luxury to have in a squad, and these players have a strong propensity to add value, particularly at the end of an innings. Players with low boundary hitting percentages should not bat towards the latter stages of an innings (ideally they would enter at an early wicket or to stabilise a faltering innings). Drafting these players should not be considered unless their bowling makes them an automatic selection. Hopefully this case study has given you insight into the data that Sports Analytics Advantage can offer cricket franchises around the world in formulating draft or auction plans - please feel free to enquire for bespoke draft and auction strategies via sportsanalyticsadvantage@gmail.com. |