We first assume that the probability of Arsenal meeting Chelsea in the first round is 1/63 as there is an equal chance of Arsenal meeting any one of the other teams.
ccc Round Probability A and C meet and A wins xxx xxxxxxxx Probability A and C don't meet but survive
1 1 63 . 3 5 0.00952 62 63 . 7 10 2
2 1 31 . 62 63 . 7 10 2 . 3 5 0.00933 30 31 . 62 63 . 7 10 2 . 3 5 2
3 1 15 . 30 31 . 62 63 . 7 10 2 . 3 5 2 . 3 5 0.00672 14 15 . 30 31 . 62 63 . 7 10 2 . 3 5 2 . 1 2 2
4 1 7 . 14 15 . 30 31 . 62 63 . 7 10 2 . 3 5 2 . 1 2 2 . 3 5 0.00336 6 7 . 14 15 . 30 31 . 62 63 . 7 10 2 . 3 5 2 . 1 2 4
5 1 3 . 6 7 . 14 15 . 30 31 . 62 63 . 7 10 2 . 3 5 2 . 1 2 4 . 3 5 0.00168 2 3 . 6 7 . 14 15 . 30 31 . 62 63 . 7 10 2 . 3 5 2 . 1 2 6
6 2 3 . 6 7 . 14 15 . 30 31 . 62 63 . 7 10 2 . 3 5 2 . 1 2 6 . 3 5 0.00084
Total probability 0.0314571

The probability of this happening in one year is roughly a little over 3 in a hundred and we raise this to the power 4 to get the probability of it happening in 4 consecutive years. This gives 9.8× 10-7 or roughly 1 in a million. Alternative approach:

Multiplying the probabilities on the tree diagram below we get for each round fk (y)=a+by where a and b are the probabilities on the lateral and vertical stems respectively. The probability of A meeting C and winning is thus
f1 ( f2 ( f3 ( f4 ( f5 (3/5))))).

This is computed to be
1 63 . 3 5 + 62 63 . 7 10 2 ( 1 31 . 3 5 + 30 31 . 3 5 2 ( 1 15 . 3 5 + 14 15 . 1 2 2 ( 1 7 . 3 5 + 6 7 . 1 2 2 ( 1 3 . 3 5 + 2 3 . 1 2 2 ( 3 5 )))))=0.0314571