Match Factor – Levelling the Field for BatsmenDave Wilson |
I’ve often looked for different ways to try and level the field, as it were, in order to rate cricketers from different eras together. Recently I thought about comparing their individual scoring to the scoring during the era in which they played – I’ve done this before but never really been happy with the results. The difference here is that I decided to couch the player’s performances in terms of what an average team would have scored during his career. In this way, we adjust for changing scoring levels over time. And at the same time find out exactly what each batsman was worth in terms of his peers.
The method works like this:-
– first, we take the player’s career runs scored (Tests)
– next, we express that by how many team innings of the period that represents
– finally, we take the ratio of his equivalent team innings to his actual innings
– this gives us his match factor
Actually the study ended up as an innings factor rather than a match factor – I initially tried it on a per-match basis but it turned out to be inaccurate in some cases, depending partly on where in the line-up the player batted and subsequently on how many actual innings the player batted in.
As ever, an example is key to understanding. Let’s look at a bygone great who happened to be WG Grace’s favourite batsman, Arthur Shrewsbury; when Shrewsbury played, just before the onset of the Golden Age and that era’s subsequent impact in terms of increased scoring, the average team score during his Test career was a very low 186 (actually 186.2). Shrewsbury meanwhile totalled 1277 Test runs, so his equivalent team innings is 1277 divided by 182.6, or 6.86. As his actual Test innings played was 40, his Innings Factor is therefore 6.86 divided by 40, or 0.1714.
Here is how Shrewsbury (and everyone else) fares using this new measure. What these tables show are the player’s Innings factor as described above, the team average during his career, the resulting equivalent team innings, his total number of Test innings, and the player’s highest ICC Test Player ranking for comparison; now, I’m not trying to say that my rating is in any way as significant as the ICC ratings, it is simply the most appropriate statistically-derived Test player career ranking to use as a barmoeter for this new measure. For the innings factor I imposed a threshold of 40 Test innings (later I’ll discuss those notables who didn’t qualify but would otherwise have been highly ranked):-
So what we have here is the top ten ranked batsmen as rated based on a) how many runs they scored, b) how many innings it took them to do it and c) the general level of scoring for that time. As can be seen, this is a significantly different top ten to that generated by the ICC rankings, with only four players appearing in both lists. Indeed, four of those in my list don’t even make the top 20 as ranked by the ICC.
But I’ll say this – I’ve been producing largely inconsequential cricket rankings for a number of years now, and the above top ten batting list made me significantly happier than any of those previously derived. I would certainly rate Weekes, Pollock and Lara in my own top ten, and I’m sure many CW forum members would also include Headley (actually, see this current forum thread for just such a discussion). Perhaps unsurprisingly, we can see the massive gap between Bradman and the second placed man.
Now here’s a mindblowing fact – the gap between Bradman in first place and Weekes in second is the same as the gap between second place and the man in 229th position!
All but two of those listed survived to the latest round of the aforementioned CW forum vote (Walcott and Barrington). Here is a direct comparison of the ICC top ten with the above list:-
Hobbs and Ponting are equal 3rd in the ICC List, while Walcott, Sobers, Sangakkara and Richards are equal 6th. The most disappointing omission to me in the ICC list is Lara – in my opinion any discussion of the greatest ever (or second greatest ever, if you insist) has to include Lara – however, as both ratings are based on numerical measures there is no subjectivity in either case. Hutton, May and Sangakkara are a little lower in my list but still in the top 20. Ponting is the biggest difference (see later). Three of the ICC top ten didn’t make the fourth round of the CW forum vote – Walcott, May and Smith.
Moving on to the next ten:-
|15||0.1661||309.3||12.45||75||Steve Smith *||10|
Only three players are more lowly ranked by the ICC than we have them here, and in truth just two, as Gavaskar is not that far away from his ICC ranking (Harvey’s ranking is also about the same). Interesting that Gavaskar is the highest rated Indian batsman in my list – I know a number of my co-forum members rate Gavaskar very highly, though in truth the difference between his rating and Tendulkar’s in 20th position is very small (I haven’t calculated the uncertainty of measurement). The latter of course has the highest number of equivalent team innings of all batsmen, which considering that he holds all of the major aggregate records is no surprise. Hutton drops quite a bit as compared to his ICC ranking, while Hammond surges in the opposite direction, but the biggest difference is the aforementioned Shrewsbury, who ranks at number 13 here but doesn’t appear in the ICC top 100. In fact, with a highest career ICC rating of 733 as compared to Trumper at 100th spot who peaked at 801, it’s possible Shrewsbury doesn’t even make the top 200 by ICC’s measure. It could be that my method of era-leveling makes a fairer job of rating pre-WW1 players. Certainly Shrewsbury’s presence in the top 20 demonstrates that the field appears to have been levelled as intended.
However, as if in confirmation of the ICC ranking, the general lack of love for Shrewsbury was highlighted in his removal at the first culling in the recent forum poll for second greatest batsman. Steve Smith is the highest rated active player at 15 – where he will end up when all is said and done is anyone’s guess.
As a small digression, here are the top ten players with the highest numbers of equivalent team innings:-
Tendulkar, in his Test career, produced the same number of runs as more than 53 average teams would have done, and is significantly ahead of Ponting – the gap between them in equivalent team innings is the same as between second and 11th. Not quite Bradmanesque but certainly way out on his own – Tendulkar is about 20% ahead of the next man, while Bradman was about 42% ahead of Weeks in the innings factor top ten list shown earlier.
Back to the innings factor rankings, here are numbers 21-30:-
Three West Indian batsmen are significantly higher rated using this measure than the ICC gives them credit for – Kanhai jumps from 52 to 27 while Worrell jumps from 77 all the way to 28, however the most significant change sees Seymour Nurse at 24, whereas he is not ranked in the top 100 by the ICC. Mohammad Yousuf and Younis Khan are 30 places apart on the ICC top 100, but sit side by side here.
|33||0.1550||309.3||14.57||94||David Warner *||44|
Ricky Ponting. How come Ponting is so much lower by this measure? It’s because it’s a career measure – if you compare Ponting’s career ICC profile with Lara, for example, who ranks above him in my system, Ponting has a much shorter peak period. Indeed, looking at Lara’s ICC rating profile makes me wonder why he doesn’t feature more highly in their system, however the inner workings of their ratings are not available to the general public. Saeed Anwar, Arthur Morris and Bob Cowper rank much higher here, though Cowper only just makes the 40-innings threshold, while Hayden drops quite a bit.
The remainder of the top 50 looks like this:-
|41||0.1517||311.0||23.66||156||Hashim Amla *||28|
|44||0.1509||309.3||11.01||73||Joe Root *||21|
With Amla and Root joining the ranks, there are four current players featured in the top 50, though whether they will remain there is up for debate – as noted with Ponting, a career rating is a different proposition to a rating which rewards peak performance, and one or more of these may not eventually feature. Two Bradman-era Australian batsmen are featured in this group, Hassett and McCabe, as well as a personal favourite of mine, the grossly underrated (though more as an all-round player) Bobby Simpson, as is that other underappreciated all-rounder, Englishman Ted Dexter (see this feature for my feelings on Lord Ted, plus this far more eloquent piece from CW’s Martin Chandler). The classy Victor Trumper ranks much higher in my list than in the ICC list, which may be more fuel for the charge of the ICC system overlooking early-era players. Eddie Barlow, stalwart of that great South Africa side of the ’60s, does not feature in the ICC top 100.
The next 25 are:-
|55||0.1473||311.4||25.93||176||AB de Villiers *||11|
|56||0.1467||309.3||13.05||89||Kane Williamson *||35|
There are some all-time greats here, including Graeme Smith, Allan Border and Steve Waugh, though these are all fairly close to their ICC rankings. Less of a surprise now that we have players featured who don’t make the ICC top 100, though Bradmanesque run-machine Ponsford, the graceful Graveney, the dogmatic Boycott and the recently departed Martin Crowe are surprising absentees. Charlie Macartney and Reggie Duff are exemplars of the early era players we discussed earlier – Duff’s life tells one of the more tragic stories in cricket. With two more current players featured in this group that brings the total to six so far.
And then we come to Navjot Sidhu. Running these little statistical exercises will sometimes throw up a player who has you checking all your figures and data entries over and over (Greg Matthews in my impact study is a good example), and Sidhu certainly falls into that category. His batting ranged from flaying spinners to all parts of the ground, to his 11-hour 201 vs West Indies and five consecutive fifties in the 1987 World Cup, though he is probably now better known for his politicking and for the culpable homicide verdict against him in 2006. Notably there was a slight dip in overall scoring in the period 1995-1999, during which time Sidhu’s own scoring rate increased by more than 28% per match, which may explain his high rating by this measure.
|0.1417||311.2||32.02||226||Alastair Cook *||53|
|99||0.1397||309.3||7.82||56||Cheteshwar Pujara *||68|
Chanderpaul has dropped significantly, from 29th in the ICC list to 91st here, as has Michael Clarke. There are two more current players to bring the total to eight, as compared to 12 in the ICC top 100.
Below is a comparison of players included in my list against those they replaced in the ICC list:-
|My Rank||Player||ICC Rank||Player|
|13||Arthur Shrewsbury||39||Gautam Gambhir|
|24||Seymour Nurse||47||Angelo Mathews|
|50||Eddie Barlow||50||Michael Vaughan|
|53||Bill Ponsford||53||Adam Gilchrist|
|54||Mohammad Azaruddin||57||Gundappa Viswanath|
|63||Tom Graveney||Ross Taylor|
|66||Geoff Boycott||62||Alvin Kallicharran|
|Darren Lehmann||64||Jonathan Trott|
|68||Charlie Macartney||70||Damien Martyn|
|72||Martin Crowe||71||Glenn Turner|
|Navjot Sidhu||72||Herbie Taylor|
|75||Reggie Duff||76||Dilip Vengsarkar|
|78||Lawrence Rowe||77||Kim Hughes|
|79||Mark Slater||79||Herschelle Gibbs *|
|80||Justin Langer||Dennis Amiss|
|81||Mark Richardson||84||Ian Bell *|
|82||Polly Umrigar||88||Marcus Trescothick|
|83||Tom Hayward||91||Jimmy Adams|
|86||Maurice Leyland||92||Ian Chappell|
|87||Bruce Mitchell||Ian Botham|
|92||Patsy Hendren||94||Robin Smith|
|94||Basil Butcher||95||Jeff Stollmeyer|
|97||Norm O’Neill||96||Simon Katich|
|98||Dean Jones||97||Geoff Howarth|
|100||Daryll Cullinan||99||Paul Richardson|
Most are below the top 50, and lower than that you wouldn’t necessarily be surprised to see top 100 lists differ. The main difference between those who don’t figure in the ICC List and those who do is that those included in the ICC list are basically from higher run-scoring eras – the average team score during the ICC players’ above is over 296 runs, the average for those in mys list missing from the ICC list is only 281.
Comparing these lists, based though they are on differing objective criteria, I can’t resist weighing in with a subjective opinion – there are a number of players in the ICC top 100 whom I don’t think should be anywhere near a top 100 batting list, such as Botham, Stollmeyer and Howarth, but quite frankly the same criticism could be levelled at my list (notably the aforementioned Sidhu). There are also players in each list who should be included in both, e.g. Boycott from my list, and I would be tempted to offer up for consideration Butcher and O’Neill. A quick look at the ICC ratings profiles shows those who don’t make my list as typically having a fairly short-lived peak, which as noted may penalise them as regards a rating based on a full career, as mine is.
Here are the totals for each nationality represented in both lists:-
My list has more Australians and fewer England players – not sure how that could have happened.
Finally, here are the highest-ranked notables who fell short of the 40 innnings threshold:-
All of the above had innings factors which, if maintained for 40 innings would have put them into my top 100.
I had earlier suggested that the ICC list is not favourable to players of the Golden Age, and of course we can readily assess that. Here are the era breakdowns of both lists:-
As can be seen, whatever means are used to derive the ICC ratings these clearly tend to favour modern players – there are seven more players who are current or whose careers took place mainly in this century, while four fewer represent the pre-WWI period than do so in my list. It could be though that this is because higher scoring eras are favoured – as noted earlier, the average team scores of players missing from my list but featured in the ICC list is significantly higher when the reverse is taken into account.
At the top of this article I mentioned that the gap between Bradman and Weekes in second was the same as the gap between Weekes and the man in 229th position. Well, when you think about it there may be a similarly huge gap if we look solely at historical batting averages. I’m not sure if anyone’s done this before, but with the same 40-innings threshold applied it turns out that, in terms of career batting average, Bradman’s average is to Graeme Pollock’s as Pollock’s is to….. Graeme Swann, way down in 447th position; just let that rattle around in your brain for a while!
You have previously written some amazing statistical features.
How ever I found this article possibly the most interesting because of the comparison between your method and the ICC Rankings.
The direct comparisons were easy to understand and in some cases quite eye opening.
See you soon Love Dad
Comment by Dad | 11:35pm BST 21 September 2016