Why would you count in 3 year slots? You're saying that it could happen on 1950, but not 1951, and not 1952, but again on 1953?
What if there was La Niña on 1952, 1953 and 1954? Your method does not account for that. Natural events and climate do not adhere to someone's made up "mod 3" date statistics.