The KO Count (Knock-Out)

Updated 24 August 2026

KO is unbalanced, which is the whole point of it: you never divide by the decks remaining. That removes the hardest part of counting at the cost of a little precision. We computed its betting correlation at 0.970 — against 0.963 for Hi-Lo — from the effect each card has on the game, rather than copying the figure from a book.

How it compares, and how we know

KO drops the true-count conversion entirely. Instead of dividing, you start at a number that depends on the deck count and act when the running count crosses a fixed threshold. For most people that trade is worth taking — a system you execute correctly beats a better one you fumble.

The number that ranks counting systems is betting correlation: how closely the system’s tags track the actual effect each card has on your edge. We computed it rather than copying it — by removing one card of each rank from the shoe, recalculating the entire game, and correlating the result with the tags. The full method is here.

Is it worth switching from Hi-Lo?

Probably not. The gain over Hi-Lo is 0.007 of correlation, which translates into a small fraction of a percent of edge. A single miscount per hour costs more than that.

Who it actually suits: Anyone who finds the division at the table the hardest part, which is most people. It is the standard recommendation for a second system after Hi-Lo, or a first one instead of it.

This is the part the system comparisons usually skip. The difference between counting systems is small. The difference between counting accurately and counting sloppily is not.

Every system, with the numbers

Computed for six decks, dealer stands on soft 17, using the effect of removal for each card rank.

Keep reading

By Martin Kosek · every number on this site is computed by us, not copied.