← blackjack lab

Blackjack Trainer

Keep the running count as cards come off the shoe, then check yourself.

space deal / pause   c check   r reset
Running count?
0 / 0checks correct
0streak
0best streak
A23456789T-K

How the systems differ, and what counting is worth: counting systems · index plays

A hand is dealt against the dealer. Pick the mathematically best play. The deviations are the Hi-Lo Illustrious 18.

Dealer
You
0 / 0correct
0streak
0best

The count-based exceptions (the Illustrious 18) are laid out in the lab: index plays.

Size your bet to the Hi-Lo count; bet the half-Kelly amount.

bankroll 1,000 units  ·  true count
0 / 0within range
0streak
0best

Half-Kelly bet ≈ ½ × (edge ÷ variance) × bankroll, where the edge is about 0.5% per true count above +1 and a blackjack hand's variance is about 1.3. At or below the +1 pivot you flat-bet the table minimum (1 unit). Rounding the exact half-Kelly up or down both count as correct.

Back-counting with the Hi-Lo count: track the count yourself as the cards come out, decide each round whether to sit in, then see how much of the available edge you captured.

Dealer
Player
You

The true count is the running Hi-Lo count divided by the number of decks remaining in the shoe — that is what tells you the edge. “Edge captured” assumes a flat 1-unit bet each round you enter; the break-even count is about +1 (the ~0.5% house edge off the top is cancelled by ~0.5% per true count), so an optimal back-counter enters at +1 or higher and sits out the rest. Use Reveal count sparingly — the goal is to keep the count in your head.

Read the discard tray: estimate how many decks are left in a 6-deck shoe — the denominator of the true count.

discard tray
how many decks are left in the shoe?
0 / 0nearest ½ deck
0streak
0best

As cards are dealt they pile into the discard tray, so the tray and the shoe always add to 6 decks. Estimating the decks remaining by eye is how you turn a running count into a true count (running ÷ decks left). Landing within half a deck counts as correct.