Hi-lo guide
Thor Hi-lo: why the next card is never due
There is nothing Norse about the hi-lo on this page; Thor is just the address it has always had. The subject is the set of beliefs people form about runs of cards, and what the arithmetic says about each one.
The model these answers rest on
In this model each next rank is drawn uniformly from 13 ranks, independently of every earlier draw, as a hash-based service does when each card comes from a fresh nonce. Some games use a real shrinking deck instead, where past cards do matter; that case is covered in the hi-lo overview. Everything below assumes independence.
Myth: a rank that has not appeared is due
Missing ranks are normal, not a sign that a rank is overdue. In 13 draws, any one rank fails to appear with probability (12/13)^13 = 35.33%, and on average 4.59 of the 13 ranks are absent. After those 13 draws, the absent rank's chance on draw 14 is exactly 7.69%, the same as every other rank.
Myth: after several low cards, a high card is coming
Say the face card is a 7 and the last four draws were all below 7. The chance that the next card is higher than 7 is 46.15%, as it would be after any history. Four lows in a row is not even unusual: the chance of four consecutive draws below 7 is 4.54%, and over 60 draws the chance that such a run appears somewhere is 79.68%.
Myth: a player can run hot
Wins cluster by chance. Calling higher on a 5 wins 61.54% of the time, so five straight wins happen with probability 8.83%. After those five, the sixth call still wins 61.54% of the time. A streak describes the past and carries no momentum.
Myth: patterns in the cards can be read
Sequences such as alternating high and low, or a rank repeating, feel meaningful because people notice them. Under independence every specific sequence of n ranks has the same probability, (1/13)^n. The sequence K, A, K, A is exactly as likely as 4, 9, 2, J: each has chance 35.01 in a million.
Myth: the switch from higher to lower changes the edge
Switching sides changes the chance and the payout together. With a 1% edge, both sides on any face card return 99.00% of the stake on average. No sequence of choices, however clever, adds up to a positive expectation when each choice has a negative one.