# Before And After. All Blacks Offensive Stats

A more quantitative approach to rugby union

tackles-sacks-estimations

A more quantitative approach to rugby union

This last Saturday the first half looked totally different. The Wallabies controlled possession and territory (about 70% of each). Even so, the New Zealand scored two tries. And then the red card for Scott Barrett, arguably the most dramatic moment of the match.

Right after Will Genia finally kicked the ball into touch, comments have started popping-up on my Twitter timeline. There’s been a good deal of tweets contesting Jérôme Garcès’ decision. Some of them suggested a new rule should be introduced – a red-carded player would be sin-binned and then, after spending 10 minutes on the naughty boy’s chair, replaced by a reserve player.

But I still remember the 2017 encounter. Australia came back strong in the second half but were only able to reduce the margin. I won’t argue the send-off was the most important moment this year. I’m perplexed, however, as to why the All Blacks offence became more aggressive and much more efficient after losing Barrett. So, here’s some before-and-after numbers to deepen the perplexity. I collected the stats on carries, passes, and attacking rucks additionally broken down by vertical and lateral zones. The vertical (lateral) zones help to understand how the ball travelled East-West (North-South). But we’ll start with the aggregate numbers.

The total time of controlled ball possession, 13 min 8 sec, didn’t differ much from the values observed in the Pacific Nations Cup (sic!). But the All Blacks controlled the ball in the first half for only 3 min 31 sec. That’s a 27%-73% split, which was also observed for the phases. Talking about the phases, in the second possession of the second half, the All Blacks recorded 23 phases. That was 4 phases shy to the total of the first half. It was also the possession resulting in a Beauden Barrett’s try.

Total | Per Possession | ||
---|---|---|---|

Time | |||

first half: | 3:31 | first half: | 0:12 |

second half: | 9:37 | second half: | 0:44 |

Phases | |||

first half: | 27 | first half: | 1.6 |

second half: | 72 | second half: | 10.8 |

Carries | |||

first half: | 42 | first half: | 2.5 |

second half: | 112 | second half: | 8.6 |

Passes | |||

first half: | 51 | first half: | 3.0 |

second half: | 141 | second half: | 10.8 |

Attacking Rucks | |||

first half: | 13 | first half: | 0.8 |

second half: | 61 | second half: | 4.7 |

Quite frankly, I thought I would have to perform statistical testing in order to better describe the difference in performance between the first half and second half. But there’s no need for any statistical procedure. All metrics show a vast improvement in the second half. Equally informative are the entries in the vertical movement matrices. (Plays resulting in tries have been excluded from the analysis). For instance, if a play started in one’s own 22-10 zone, 79.8% of all plays finished in the same zone. If we’re interested in the ball travelling forward from this zone, the chance that this actually happening was 10.7%. Slightly smaller was the chance that a play would end up in the own22 (7%).

Total | Own22 | Own22-10 | Own10-50 | Opp50-10 | Opp10-22 | Opp22 |
---|---|---|---|---|---|---|

Own22 | 87.5% | 10.7% | - | 1.8% | - | - |

Own22-10 | 7.1% | 79.8% | 10.7% | 2.4% | - | - |

Own10-50 | - | 15.5% | 69.0% | 12.1% | 3.4% | - |

Opp50-10 | - | - | 10.5% | 68.4% | 19.3% | 1.8% |

Opp10-22 | 0.9% | - | - | 7.1% | 85.8% | 6.2% |

Opp22 | 1.8% | - | - | - | 7.0% | 91.2% |

Not surprisingly, the frequencies appearing in the main diagonal element of the matrix take the greatest values. Most of the plays finish in the same zone they started. Since the zones are of different lengths, the movement between shorter zones (own10-50 to opp50-10) is perhaps easier. Indeed, the largest off-diagonal frequency was observed for the transition from opp50-10 to opp10-22 (19.3%). Now, let’s compare the matrices for first and second half.

H1 | Own22 | Own22-10 | Own10-50 | Opp50-10 | Opp10-22 | Opp22 |
---|---|---|---|---|---|---|

Own22 | 95.2%% | 4.8% | - | - | - | - |

Own22-10 | 6.3% | 75.0% | 12.5% | 6.3% | - | - |

Own10-50 | - | 10.0% | 70.0% | 10.0% | 10.0% | - |

Opp50-10 | - | - | 10.5% | 63.2% | 21.1% | 5.3% |

Opp10-22 | - | - | - | 10.3% | 82.8% | 6.9% |

Opp22 | 25.0% | - | - | - | - | 75.0% |

H2 | Own22 | Own22-10 | Own10-50 | Opp50-10 | Opp10-22 | Opp22 |
---|---|---|---|---|---|---|

Own22 | 82.9%% | 14.3% | - | 2.9% | - | - |

Own22-10 | 7.4% | 80.9% | 10.3% | 1.5% | - | - |

Own10-50 | - | 18.4% | 68.4% | 13.2% | - | - |

Opp50-10 | - | - | 10.5% | 71.1% | 18.4% | - |

Opp10-22 | 1.2% | - | - | 6.0% | 86.9% | 6.0% |

Opp22 | - | - | - | - | 7.5% | 92.5% |

The most visible difference is noted for the own22 and opp22 zones. Raising the former, the chance of not moving forward from the own22 was 95%. In the second half the All Blacks left their own red zone more frequently. They were also more efficient in the opp22. During the match, the All Blacks have recorded 57 plays within the Wallabies’ 22, 53 of which in the second half.

Other forward transition frequencies were slightly greater in the first half. For instance, the chance for moving forward from the opp50-10 in the first and second half was equal to 21% and 18%, respectively. The test for equality of two frequencies suggested, however, that these frequencies were equal with the probability of 0.808.

Did the red card affect the result? Obviously. But it was not the offence where Scott Barrett was missed the most. Does the defence win championships? Obviously. But had the XV played in the offence like the XIV did, the result might have been different. And I still remember the cold Saturday evening in Sydney, 19. August 2017. Half-time: Australia 6-40 New Zealand.