The elusive hat-trick

Do we know about all hat-tricks?

As Stuart Broad was about to bowl to Shaminda Eranga at Headingley in 2014, Simon Mann tapped me on the shoulder and asked “Is he on a hat-trick?” He had dismissed Dinesh Chandimal with his first ball of the over, but it took a look at my scorebook to confirm that with the last ball of his previous over he had got Kumar Sangakkara out. So, yes, he was on a hat-trick. By the time that I had picked up my microphone to let Jonathan Agnew, who was commentating for Test Match Special, know, it was too late. Broad was already bowling and duly had Eranga caught by Matt Prior. Aggers described the wicket and suggested “Broad is on a hat-trick” or words to that effect. Somewhat sheepishly, I had to say “Er … that was the hat-trick” and explained that he had got Sangakkara out with the last ball of his previous over. Amazingly, it appeared that no one in the ground, apart from the ever-perceptive Simon Mann, had realised Broad was bowling a hat-trick ball. As Broad roared in for his next ball, the TV graphics proclaimed “Broad on a hat-trick”. It took a while for the message that he had already taken the hat-trick to filter through. At the end of the over the PA announcer said something like “Congratulations to Stuart Broad on completing a hat-trick during that over” to the bewilderment of Broad and his team-mates on the field.

How can we not know that a bowler is on a hat-trick? Especially in a high profile match like that. Here is CricInfo’s commentary (edited for this purpose) on the sequence of events:

62.6 Broad to Sangakkara, OUT. Sanga falls to a blinder of a catch at gully! … Kumar Sangakkara c Bell b Broad 79

63.1 Plunkett to Chandimal, no run Back of a length to begin with, low 80s mph and defended

63.2 Plunkett to Chandimal, 1 run. Length ball, leans out and steers a drive through the covers

63.3 Plunkett to Prasad, no run. Short and wide outside off, Prasad has a wild swing and misses

63.4 Plunkett to Prasad, OUT. Banged in and gloved through, another one gone! Dhammika Prasad c †Prior b Plunkett 0

63.5 Plunkett to Herath, no run. Banged in and Herath jumps out of the way, throwing his arms in the air.

63.6 Plunkett to Herath, no run. Angled in from around the wicket and Herath swings, misses.

64.1 Broad to Chandimal, OUT. Pitched up, draws the stroke and steered to slip! Dinesh Chandimal c Cook b Broad 45

64.2 Broad to Eranga, OUT. He's got it, edged through to Prior! Broad has a hat-trick spread over two overs! Broad hasn't realised, in fact, it looks as if the fact has passed everyone by... but Broad has three in three, Sanga with the last ball of the previous over, then two edges in two balls. Shaminda Eranga c †Prior b Broad 0.

I don’t know whether CricInfo’s commentator knew it was a hat-trick at the time or changed the commentary later, but good on them if they did.

So, Plunkett dismissed Prasad at the other end in the middle of Broad’s hat-trick. (Well, not statistically ‘in the middle’, but you know what I mean). Which seems likely to have put most hat-trick hunters off the scent. Broad’s hat-trick is one of only two in Test cricket that have seen a wicket fall at the other end in the over between hat-trick wickets (excluding the three that were spread over two innings and thus had whole innings take place during the course of the hat-trick). The other one was by Ryan Sidebottom for England against New Zealand at Hamilton in 2008. Like Broad, he took a wicket off the last ball of one over and wickets with the first two balls of his next over. During the over in between Monty Panesar got Brendon McCullum out. I suspect that the fact that the first two wickets, Sangakkara and Chandimal, made 79 and 45 respectively also helped cause the confusion, given that most (but not all) hat-tricks involve at least two players making first-ball ducks, which generally makes hat-trick-spotting easier. So, if Broad’s 2nd wicket had also been a first-baller we are more likely to have seen “Broad on a hat-trick” as he ran up to bowl his hat-trick ball rather than as he ran up to bowl his “4-in-4” ball despite the wicket in between.

Broad’s hat-trick came to mind when I was listening to BBC’s Nottinghamshire commentator, Dave Bracegirdle, during a recent County Championship match. Bracegirdle mentioned some difficulty finding lists of hat-tricks in the aftermath of an evening with three hat-tricks in the T20 Blast. Such lists are readily available, and I thought it was an opportune time to check my records against the lists on established websites. I managed to persuade my laptop to trawl through the ball-by-ball data that I have and alert me of possible cases. All told I added about 25 hat-tricks to my database. Some of those were noted on CricketArchive and/or CricInfo (and thus was my error that I did not have them in my table), but there were quite a few that I couldn’t find noted anywhere. I suspect that even the bowler didn’t know in some instances. At the end of that process my list of T20 hat-tricks (men), for example, has a total of 156 entries. CricInfo has 149 on their list and CricketArchive/Association of Cricket Statisticians and Historians has 135.

All of which leads me to suspect that there remain many hat-tricks that have occurred but are not in the records. Prior to the ball-by-ball era (i.e. the 21st century) hat-tricks were maintained on lists where cases were noted in scorecards or match reports. Given that even with ball-by-ball data we seem to miss some, it is not hard to imagine that many more were missed in the ‘old days’. Hat-tricks spread over two overs, or two innings, or with wickets in the over in between (sandwich hat-tricks?) are the most likely candidates to be missed by cricket’s record-keepers. And there may well be quite a few where a bowler took a wicket with his last ball in a spell and then wickets with the first two balls of a new spell hours or days later, when the previous wicket was long forgotten. Clearly, we do not have a complete list of all hat-tricks that have been taken in major cricket. And, despite our best efforts, we never will. My estimate, and it is pretty much a pure guess, but based on my recent database-trawling I would say we are missing between about 2% to 5% of hat-tricks.

And now for a look at what the Cricket Fact Randomizer has gurgled out today:

Shivnarine Chanderpaul ran out of partners a lot. So, it is not surprising that he holds the Test record for most runs scored in ‘not out’ innings. In total he made 4 094 runs in the 49 innings that he remained unbeaten. More surprising is that this record in first-class cricket is held by an opening batsman, as they tend not to collect as many asterisks as those who spend most of their careers lurking in the lower middle-order. Geoffrey Boycott made 13 489 runs in 162 not out innings in first-class cricket. You can create a fun, and entirely meaningless, statistic by ordering innings in a career from highest to lowest rather than the conventional chronological method, with not outs ordered first. In this way adding Boycott’s not out runs to his highest score in a dismissed innings (243) you can create a ‘highest career average’ of 13 732.00.