County bowlers may start to show signs of fatigue this third week of the season. I’ve previously heard the response that it wasn’t a problem for bowlers in the good old days.
And if that’s true, then fair enough – either 1) my hypothesis is wrong (that elite pace bowlers can’t play four-days-on-three-days-off without a decline in performance) or 2) the old days were easier, meaning players could trundle all season long with no drop in performance.*
So I had a look at Fred Trueman’s record. Before we get into the results, it’s worth reminding ourselves of how the County Championship used to work. To modern eyes it looks relentless. Most games started the day after the last one. That meant August 1962 had nine three-day county fixtures, the only respite being Sundays (mid-game) where there was no play.** That is phenomenal fixture congestion. Did it impact performance? Yes.
Fred Trueman bowled 802 overs in the 1962 County Championship. Was he aware he averaged 15% more when one game started the day after the previous one? Rising to 23% if you just look at the County games.
The outcome was confounded by Trueman playing four Tests over the summer, so I looked further back – to the summer before he became an international cricketer, 1951. There the gap really was stark – averaging 51% more if there was no gap between games.
In the 15 matches where Trueman was unrested, he went wicketless in five of them. That didn’t happen in any of the ten games where he was coming in off at least a one day break.
In next week’s county games look at the pace bowlers with the heaviest workloads already this season: they’ll collectively be off their best. That’s Durham, Kent, Gloucestershire and Middlesex.
PS. I think there are three things bookmakers miss when compiling county odds: Fatigue, Home Advantage, and the benefit from fielding first. So far, I’ve be unable to exploit this. At least there are 13 game weeks to go…
*There’s a third option – that players used to be tougher. And a fourth – that the warm-ups and drills make modern games more tiring than they need to be. I choose to discount these, but make your own mind up.
**I’m just about old enough to remember when some Tests had a break on a Sunday. Feels like another world.
England just whitewashed Pakistan away from home. Two innings of spin played a major part.
First Test: debutant Will Jacks took six wickets, outperforming xW by 2.5*. Third Test: debutant Rehan Ahmed took five second innings wickets for 48, outperforming xW by 3.5 (ie. on average it was really a 1-point-five for).
Jacks’ performance was on a surface where at one point 882 runs had been scored for the loss of ten wickets. Jacks has only bowled 411 first class overs, and 46 of them were at Rawalpindi.
I have a theory: unfamiliar spinners get a boost, relative to bowlers whose variations have been publicly scrutinised in HD. The boost is measurable by the outperformance vs xW (ie getting more wickets than you’d expect given the quality of balls bowled). We can’t test this directly (I don’t have the data, plus it’s subjective to know who is “unfamiliar”). But we can test a proxy: performances on debut.
The first two columns of the below chart show how it used to be. Lots of promising bowlers get a chance, very few succeed. Most wickets are take by the best bowlers (Murali & Warne) at a low average. These days that trend has reversed.
What’s driving that? I’d suggest it’s the element of mystery – spinners come in many forms, and each has a subtly different style and (possibly) variations. In the 1990s a top bowler may have been able to surprise batsmen with old tricks. Little chance of that these days. Already we know Rehan Ahmed bowls with a high arm, his leg break doesn’t turn as much as the googly, both deliveries have a scrabled seam with lots of topspin. And that’s just from one game.
There are probably some secondary factors, such as better selection, more judicious use of a second spinner, and fewer stars around to bring down the average of non-debutants.
If unfamiliar spinners have an advantage, there is an incentive to give the old ball to any vaguely capable new Test batsman. Keep an eye out for it in future, especially with England.
Take from this that one swallow does not a summer make. Give a spinner a few games to see what he’s made of**. Then we’ll decide if you’ve found the next Swann.
*I think. Didn’t write it down and Cricviz take stats off the app sharpish these days.
** Note the deliberate use of “he” – these stats aren’t based on the women’s game; so it would be inaccurate for me to assume the same principles apply.
Leg spinners get better as their careers go on (courtesy of @sanderson_club). Rule of thumb: takes 700 overs to reach their peak.
2. Rehan Ahmed xW – scraped from CricViz’s app.
3. ODI Spin Debutants – Note how wide the gap used to be between debutants and all spinners in ODIs. Just like in Tests.
England’s spinners just took 8-405 within a total of 657. Pace bowlers only sent down a third of the overs: 50 across three days. A guy with 21 first class wickets at 53 bowled as much as the opening bowlers combined. So what happened, any why?
Anderson got 3 overs with each new ball, then three or four overs with a 30 over old ball. Just 16 of the first 120 overs. No doubt one eye on the second innings, and one on the Test starting on the 9th. I presume he would have bowled more had he had more success (22-4-52-1). Seems that England used short bursts to explore if Anderson would have favourable conditions (with new or old ball). He didn’t.
Stokes similarly got 3 overs with a 40 over old ball, then with the second new ball had a go slightly earlier with a 24 over old ball. That wasn’t what I’d expected – but Stokes averages over 40 with the new ball, so credit to Stokes’ England for giving new(ish) ball duties to the spinners.
Robinson had eight overs with new balls, then only 13 overs at other stages in the innings. He bowled so little that I didn’t see much of it, and can’t really comment.
Expect Mark Wood to return to the side for the second Test, in place of Liam Livingstone (knee injury). I see two reasons why Robinson didn’t bowl more: either rest bacause he’ll play the second Test ahead of Jamie Overton, or he looked so ineffective his overs were given to the part time spinner.
Will Jacks wasn’t someone I thought we’d see bowling much. 21 wickets at 53 in First Class cricket, and just as significantly he has bowled fewer than nine overs per game.
He made a good start – exerting control over the top four (1-76 at 3.3 rpo).
Babar Azam counter attacked, scoring at 7.4rpo before Jacks got him caught at point – a win for Will Jacks
Then 4-37 against numbers seven to eleven. Admittedly only two wickets were against the top order, but a good return (6-161) from a sixth bowler.
Imam ul Haq received more than half his balls from Jack Leach, making 54 before being dismissed. Not clear that England got their matchup right there if that was pre-planned.
So I can sort of understand England’s approach, and they may yet win. But when the opposition make 579 and you’re conserving bowlers because there are 15 days of Test cricket in 21, then maybe the scheduling is having unintended consequences that devalue the sport.
England 24%, New Zealand 46%, Draw 30% for Lord’s says my model. Here I’ll explore why I disagree with the bookmakers (who say 40%/25%/35%*), then give crusty crumbs of comfort to England supporters, and finally share what I think players would average if we played the series forever.
Advantage New Zealand
The visitors will have seven batters who have performed at Test level (Devon Conway came along just when needed, with 767 runs in his first seven Tests). Meanwhile England’s top three average in the twenties (Lees/Crawley/Pope). Yes, two of them are better than that, but you get the point.
New Zealand should have four front-line pace bowlers, well suited to early-summer in England. As a collective, their five possible quick bowlers have 105 wickets at 26 in blighty**. The hosts chose Jack Leach (which I see as a brave move at Lord’s). Pace tends to do well at Lord’s – 26 rpw vs 39 spin. I expect pace to dominate in early June.
Have an actual allrounder (Colin de Grandhomme). Contrast this with England who will probably gamble on Root to be fifth bowler. England may need to discount Stokes’ bowling (four wickets at 57 this summer, 11 at 43 in the last year of Tests). With Leach/Root as two of the five bowlers, England would have to field last if they won the toss, even if that’s not optimal. New Zealand can be more flexible.
Reasons to be cheerful (if you’re English)
New Zealand (World Test Champions, last two years W8 D2 L3) have an excellent record but may have peaked as a team. Ross Taylor and BJ Watling have retired. Trent Boult (32) is likely rested for the first Test, having only played four Tests since Jan 2021. Like all modern tourists, they’re undercooked. Being 70-8 last week against an attack featuring Ben Gibbon hints NZ aren’t in mid-season form. Don’t mention: England have W7 D7 L12 in the last two years.
James Anderson. Last three years averages 23. Yes he’s two months shy of his 40th birthday, but still has 12 wickets at 23 in the county championship this summer, which is strong in the context of the runfest we’ve seen. Don’t mention: Anderson’s strike rate is 69 this championship summer, and in the last 12 months averages 29 in Tests. If NZ can see him off, plenty of first innings overs against spin await.
Pope – averages 32 against pace, 22 against spin. Has a great chance to prove himself at number three. Similarly, Stokes (40/33) prefers pace. And Leach (37/6, tiny sample size) Don’t mention: Root (44/69), Bairstow (29/43) are the opposite. Or that Crawley averages 18 against Right-arm Medium in Tests.
Expect an even contest between bat and ball, with average scores just over 300.
And, in case my model is running in Garbage In Garbage Out mode, here’s the expected averages I fed it:
Am all ears as to how England are favourites – find me @edmundbayliss on twitter, or comment here.
*I’m so far off-market here that this can’t be right. What could I be missing? Leach’s spin? Overrating Williamson? Is Potts the new Archer?
** Synonym for England / Great Britain. Note: must check how proper writers avoid saying “England” as every tenth word.
Pace bowlers need rest. The County Championship structure begins with a rhythm of four days on, three days off. I think that’s too much and impacts performance.
Last week Ryan Higgins bowled 25 wicketless overs. So did Darren Stevens, going at over four an over. There’s more: Chris Rushworth, Jackson Bird, Ajeet Dale, Michael Hogan, Jamie Atkins. All had played three games in 18 days. In the third game their collective figures were 0-506.
Zoom out. I’ve looked at the pace bowlers who have played all three games this spring, and how they compare to the fresher bowlers that haven’t played all three:
Flipping heck. A 38% difference in average. This is much bigger than when I’ve looked at this before. But then those were just back-to-back Tests (7%), or for short recovery between List A games (5%). This is the harsher concept of back-to-back-to-back. It is, admittedly, just one week I’m looking at – adding the error bars we’re comparing averages of 25 (+/- 5) with 34 (+/- 7). I’d be surprised if the real variance is more like a (still whopping) 15-20%. 38% just feels too high*.
Let’s get into the implications of this:
Selection – if the above table is right, then Ryan Higgins (expected average 24) becomes a 33 averaging lump playing his third game on the trot. This means rotation is required. Puts the sides relying on one or two strike bowlers at a disadvantage (like Glamorgan, Derbyshire, Gloucestershire). Here’s the pace bowlers with the heaviest workloads so far – keep an eye on them next week***:
Season Structure – if four-days-on-three-days-off doesn’t work, what about four-days-on-four-days-off? Could start the first game on a Wednesday, the next the following Thursday etc, ensuring games still include weekends.
I’m a traditionalist, but if 14 games per season is damaging the competition, then maybe 12 (in the same window) is better. Standards must remain high. I’ll track and we’ll know more by the end of the year.
Scheduling – Looking ahead, there are four teams that play all the first six games. Then two (Durham & Leicestershire) are involved in all the seven April/May matches. Quite a disadvantage.
County Stats – is the failure of many recent batsmen to make the step from county to Test because batsmen are having things too easy against fatigued bowlers? Maybe it sounds far fetched, but worse suggestions have been provided.
Next steps – The hypothesis is that the knackered bowlers will underperform next week. We’ll see what an extra week’s data says. If I’m right, Essex and Hampshire may struggle.
* Yes, there were lots of top quality fresh bowlers deployed for the third round of games, but this only explains 3% of the gap.
** Lovely weather we’re having. It might be that there’s normally rain and cloud around, giving bowlers more helpful conditions and more rest. Thus (as The Leading Edge Cricket Podcast point out), the Top 6 batter average stands at 40 this year, up from last year’s 31.
***Not all games are equally tiring. Winning by an innings in two days, you get two rest days. A rain affected game is probably as good as a week off. When Hampshire scored 652/6 Abbott and Abbas had their feet up. I still need to think about who has had the best chances to recover.
I’ve put together previews for each county. Since I’m better at statistics than writing, I’ll let the charts speak for themselves. The only thing to add is that I did these before the first round of games (honest!)
If you have any questions, feel free to comment below or find me on Twitter @edmundbayliss.
England’s red ball batting doesn’t have much going for it. But it has Joe Root.
Given that, it kinda makes sense for England to Put Root First when picking a team – then work out who should average 29 in the other six slots.
But word on the street is that Root wants to bat at three. At first glance that’s brave – Root’s data at face value says he’s best at four or five.
First, let’s bang some error bars* on those averages. Root at three averages 39 (+/- 11). At four that’s 51 (+/- 10). He’s probably better at four, but it’s possible the difference is just chance. Even after 53 innings at three, raw averages can’t prove to us that he’s a better three than four.
What about other batsmen? Is there a theme: is it easier to bat at second drop? There are fourteen batsmen who’ve batted more than 20 innings at both positions this century. Not that many people, so it’s open to interpretation. My reckoning: if you’re good at four you’ll be fine at three.
Going back to Root, is there something specific that means he’d struggle at three? Firstly, it’s worth noting that 60% of his dismissals at three were against India or Australia (at four, that figure is 48%), so there’s a bit of mitigation for his lower average one place up the order. Now, here’s Root’s record by over:
Joe Root is very good at batting. Averaging 40 in the first ten overs? He could be an opener. What makes that average even better is that if he’s batting early in the innings, chances are conditions are tough – because at least one wicket has already fallen.
So we’ve seen that Root should be fine at three. He’s good early in the innings. Others who’ve batted well at four have done fine at three. But is three his optimum position? No. He averages 44 against pace, 69 against spin. At four, he’ll get more spin**.
Does that mean that Root at three in the West Indies is a mistake? Well, he might be missing out on his favourite bowling: when West Indies play at home, spin makes up 23% of overs (England in England: 20%). My conclusion: if West Indies play a spinner, Root should bat at four. If West Indies are pace-heavy, then Root at three to try to meet the West Indies head-on is a gamble, but not an unreasonable one***.
*Error bars based on number of completed innings in each position.
**There’s a case for batting at five, but then you lose some runs due to being left not out. Plus England have plenty of lower-middle-order batsmen.
***Though, of course, nobody needs to know England’s batting order until the first wicket falls. Knowledge is power.
Welcome. On average I think Australia are 79 runs per innings better than England, so a result like 3-0 is on the cards. Here I’ll rank the players and draw out some of the themes to expect.
It’s too easy to overanalyse. Let’s start simply. Two very good bowling attacks. Lots of OK batting. So without weather, expect results. In terms of the stronger batting, Warner/Smith/Labuschagne for Australia all average >70 at home in the last four years. As good as Root/Stokes are, England are outgunned. Australia are the better team, they are at home, and so they are favourites.
Probabilities My model says the odds should be 64% Aus, 22% Eng, 14% Draw (at Brisbane, excluding weather). If anything, I think that’s a bit generous to England. Oddly the bookmakers give England a 19% chance in Brisbane, when that game is likely to be rain affected. I wouldn’t be backing England…
Home Advantage Australia have a great home record. If you’re thinking 5-0, that’s not outrageous. They’ve whitewashed their opponents in five of the last thirteen series of three or more Tests. And Australia have won all eight day/night Tests they’ve hosted, so even the possibility of two D/N matches might not help England (who have lost three of their four pink ball games).
Spin is generally ineffective in Australia. Lyon is freakishly good there though: (avg Lyon 32, others 60). Another edge to the hosts. Note Lyon averages 40 vs RHB, 24 vs LHB in the last four years. England are heavy on RHB, which makes sense. If I were picking the squad, I’d focus on picking RHB that are best against pace, and rely on their right-handedness to blunt Lyon a bit. Look for Lyon to be into the attack early if Burns (LHB, avg 27 vs OS) is still there after 20 overs.
PS. Good to see England went for Pope over Bairstow. Bairstow averages 29 vs pace, 43 vs spin (Pope 36 / 25). I’d want the better batting against pace. A spin specialist won’t help you after you’re bowled out for 210 on the first day.
Ground characteristics. Gabba / Adelaide are Australia’s best recent grounds, with 80% home wins. Others are more like 60% home win. Might be something to do with spinners struggling:
England’s attack vs LHB – might be overrated on the strength of Broad’s reputation. Anderson averages 10 more against LHB recently, Leach averages 56 against LHB. Lots of pressure on Broad as the specialist LHB muncher. With Warner, Head, Harris and Carey batting backwards, it will be interesting to see England’s plans.
Rotation – 25 days’ play out of 42. Rule of thumb: add 7% to a pace bowler’s average in back-to-back Tests. There are two ways rotation can impact a team through a series: bringing in weaker bowlers, and failing to rest tired ones. I think the latter is the bigger risk in this series.
Adelaide is the obvious game for a rest (there are reasonable gaps between the other Tests, so a pacer could play four out of five). Australia have four excellent pace bowlers for three slots, so can merrily rotate (though Cummins being favourite for leading wicket taker indicates that he’s expected to play himself into the ground). Would England dare rest Broad or Anderson while the series is alive? Maybe. Wood and Woakes are adequate replacements.
If Stokes looks peaky, England may have to play Bess ahead of Leach to rebalance the side. Then they’ll really be in trouble.
The Toss. Teams tend to bat. Note the increased chance of draw if Aus bat first. Just as Leamon/Jones suggested in Hitting Against the Spin – it’s harder to force a win batting first.
PS. Hope the above wasn’t too disjointed – the series starts a day earlier than I’d thought. As a reward for making it to the end, here’s the details on some bets I’ve made:
James Bracey (FC avg 37) is now in the England squad. That surprises me, so I’m going to take a closer look.
37 is a very good average. But players with better averages are a long way from Test honours (his captain, Chris Dent, averages 38). Their stats are based on facing second division bowling. Thus I was expecting Bracey to average in the low 30s this year in the more challenging conference system.
Actually, Bracey has started 2021 brightly, with 478 runs in five games. That’s enough to boost his expected* Test average to 29 (+/-7)**. Here’s his record:
That +/-7 is important. England rate Bracey highly – with Lions tours already under his belt. Thus you might presume he’s doing better in the nets than in the middle – and put him at the upper end of the range, with an expected Test average of 33/34.
Context 1: batting in the top three. When cricket was a summer sport, it would be odd to pick a batsman averaging under 40. But top order batting in England in the spring and autumn is hard. Comparing Bracey to a selection of his peers (five number three batsmen in D2) from 2017-19, they averaged 30 and he outperformed them by 21%. Looking at this season, the top three clearly get the worst of the conditions. We should adjust expectations accordingly.
Context 2: Other bright young things. It looks like England are picking players on potential. If we look at the best batsmen born after 1997, most of them are in or around the squad.
Expected 2021 FC Avg
Best U24 batsmen in County Cricket. Note that if England are looking at players with upside as they move up the age curve, Josh Bohannon (averaging 46 this year) and Ricardo Vasconcelos (70) might be worth watching.
Context 3: First Class vs Test. I thought Zak Crawley (FC avg 32) made his Test debut far too soon. Yet he’s averaging more in Tests after 20 innings (34) than FC. It could be that performances against medium pace and junk spin should be excluded when working out who has international potential. I’ve started a ball by ball county database to get more detail – here’s the limited data so far, and yes – strip our the medium pacers and Bracey looks great. Watch out for his performances against off-spin though.
James Bracey is young, and bats in the top three – so his ability won’t be reflected in his average. He’s not (yet) a star, but it’s not inconceivable that in future he could average 35 in Tests and earn England’s troublesome number three spot.
*Expected Test average is calculated on the basis of runs scored, adjusted for difficulty, divided by number of times dismissed. Normally I’d use the last four years of data. It’s also adjusted for where they play: batting at Bristol is quite easy, I make it the fourth easiest ground for batting, inflating averages by 4%.
** Uncertainty: calculated with the formula Uncertainty = 2 * Average * (Times dismissed)^-0.5. This gives a 90% confidence estimate of someone’s ability, which gets more precise as the number of innings increases.
With six days to go until the County Championship begins, here’s my view of how each county will fare, based on player by player ratings. Analysis of conditions is in part I which is here.
Expect Division 1 to comprise Essex, Warwickshire, Surrey, Hampshire, Lancashire and Yorkshire.
Note that Group 2 is the toughest: Somerset or Middlesex would probably qualify if they were in Group 3.
Warwickshire’s depth means I rank them second in Group 1 – even though Durham and Nottinghamshire have a better first XI.
Other than last place in each group, and Essex winning Group 1, the tournament is pretty open.
Essex: Obviously the best team in Group 1. May be beatable in the spring, before Harmer is effective. Khushi may be able to displace Walter or ten Doeschate. Good pace bowling reserves
Warwickshire have the best chance of joining Essex in Division 1, with a tasty pace attack and the Brookeses, Lamb & Miles waiting in the wings. The batting may rely on Sibley, Rhodes and Malan. Hopefully Mousley (age 19) gets a good run in the side.
Nottinghamshire have a strong starting XI, with three all-rounders providing balance. They’re better than the 2019 and 2020 tables say. I like the number of above-average players who’ll be playing 2nd XI. Impressive team that Peter Trego might not get into the best XI.
Durham: I like their top five. They also bat deep. My analysis says their batting’s as good as Essex’s(?) At 25-1 I like those odds. Hopefully Borthwick can bounce back on his return from Surrey. Weak in the spin department, but Durham in April doesn’t really call for it. Expect they’ll finish fourth but they are underrated.
Worcestershire are asking a lot of Mitchell’s batting. RHB vulnerable to LS/SLA. Good enough bowling; a bit more in reserve/spin options would be nice. 2nd bottom in Division Two in 2019 (last full season); not expecting too much from Worcestershire this year
Derbyshire are a young team. Did well last year. Historic data may not do them justice as the core of the team reach their peak. Abbott and Reece mean just three specialist bowlers needed. Lots of unknowns: Just need a couple of them to come good.
Surrey: should win Group 2. Huge & talented squad; availability better than last year. Just the Currans & Roy missing from the Group stage. Burns and Pope should find the lack of spin to their liking after a challenging winter. Can Surrey keep the momentum up in Division 1 without Burns, Pope, Foakes, Roach?
Hampshire: Division 1 beckons. Batting depth covers slight shortage of quality (with apologies to Vince and Northeast). Openers have previously exposed the middle order too often. Good signings Abbas/Abbott. Opponents will be hoping one of those two is resting when they face Hampshire.
Somerset‘s bowling ensures results, but batting not at the same level. Lammonby’s three 2nd inns Bob Willis Trophy hundreds tell us he can bat. Only one batsman aged between 24 and 33; Hildreth & Davies may find age catching up with them. de Lange a brave choice of Overseas: I’m not certain he makes their strongest red ball XI.
Middlesex are by no means a bad team, but much to do to reach Division 1 from a tough group. No big weaknesses for April/May conditions. An opportunity for Walallawita (22) to become the side’s premier spinner (or for Middlesex to be bold and play without a spinner in conditions that don’t necessarily need one). Harris ensures a short tail (if selected!)
Gloucestershire – Higgins, Brathwaite and Dent are class. The rest of the relatively young top order will have to find a way to make runs. Gloucestershire are normally competitive, but a top two finish is likely beyond these bowlers. Would be good to see Howell play this year.
Leicestershire are a bowler light. Second best openers in the group. Best chance is if Azad/Harris can wear down the opponent’s pace attack. Look out for Rishi Patel, don’t let the First Class average of 17 fool you. Average age 25: this is a squad that could grow together.
Lancashire: Best batting in the group. Two very good bowlers will miss out when Anderson plays. Will be interesting to see how Lancashire balance their XI around Parkinson: will they go with five specialist bowlers if they want to include a spinner?
Yorkshire are good enough for Division 1. Bess and captain Patterson are by no means guaranteed a place. Jordan Thompson is worth looking out for. Heck of a lot of youth players in the squad.
Kent have a nice attack. Podmore will be a handful early in the season. Stevens has still got it, even though he’ll be bowling to a keeper half his age. The Crawley-Denly axis may decide whether Kent can pip Lancs/Yorks for a D1 spot. Can Crawley improve FC avg of 32? Denly’s LS mean Kent can go with four pace bowlers. Should they be worried about the size of their squad?
Northamptonshire can trouble any batting order, I think they are under-rated. Ben Sanderson the star player. Wayne Parnell and Tom Taylor add batting depth. An overseas all-rounder was definitely the right choice.
Sussex: I’ll disagree with the bookies here – I think Sussex will struggle. Wiese, Archer and Jordan are hard to replace. Overseas players and Ben Brown will hold the batting together. Lots of young reserves – which I may have under-rated (see the Notes section)
Glamorgan: Labuschagne could win them some games in May. But will it be too late by then? Neser, Hogan, van der Gugten will concern openers, but lower-middle-order batting will get an opportunity as they tire. Will need to see more from the batting of Selman, Lloyd, Root, Carlson.
Bat rating = expected batting average
Bowl rating = expected bowling average
The lack of 2nd XI games and abridged Bob Willis Trophy mean 2020 has less weighting than most years. This is likely to adversely impact the rating of young players where their data is mostly from when they were less experienced. Don’t get angry if the numbers under-rate your favourite 23-year-old.