Back counting means observing a blackjack shoe without wagering every round and considering entry only when the remaining-card estimate meets predefined criteria. It is also called Wonging after Stanford Wong, who popularized the approach. It does not guarantee access, advantage or profit.
How it differs from seated counting
A seated counter may place minimum wagers through neutral or unfavorable counts. A back counter can avoid some of that action but sees fewer playable rounds and depends on open seats, table-entry rules and a clear view of exposed cards.
Variables that determine value
- number of decks and penetration;
- counting system and true-count conversion;
- blackjack rules and payout;
- entry threshold, bet size and bankroll;
- number of tables, seats and rounds observed;
- accuracy, variance and risk of ruin.
QFIT’s published discussion emphasizes that hands per hour cannot be generalized because table availability, crowding, dealer speed, shuffles and player skill vary. Profit claims without those assumptions are not portable.
Operational and legal boundaries
Follow table-entry rules and staff instructions. Do not use concealed devices, collusion, marked cards or unauthorized information. Laws and casino rights differ by jurisdiction; this page is not legal advice and does not provide evasion tactics.
A finite-shoe example from observation to exit
Use this as a bookkeeping example, not a universal betting schedule. Assume a six-deck shoe, Hi-Lo, no side count, and a preselected entry threshold of true count +2. The table allows mid-shoe entry and a seat is available.
| Observation point | Running count | Decks remaining | True count | Action |
|---|---|---|---|---|
| Early shoe | +4 | 5 | +0.8 | Keep observing |
| Later shoe | +8 | 4 | +2.0 | Entry threshold reached |
| After one round | +7 | 3.5 | +2.0 | Remain only if the tested plan still permits it |
| After another round | +3 | 3 | +1.0 | Exit or stop wagering under this example policy |
The conversion at the entry point is +8 ÷ 4 = +2. Recalculate before every betting decision with the same rounding or truncation rule used by the underlying analysis. Do not enter simply because the running count is +8; with six decks remaining it would represent a much smaller true count.
Define the policy before watching
- The counting system and tag values.
- How decks remaining are estimated.
- The exact true-count conversion convention.
- An entry threshold derived from the actual rules and tested betting model.
- A separate bankroll-based wager or ramp.
- Exit conditions: count below threshold, shuffle, seat loss, rule change, requested departure or an operational problem.
A +2 threshold is only an example. It does not establish a player edge at every table. Blackjack payout, dealer soft-17 rule, doubling, penetration, errors and the rounds still available determine whether entry has positive expectation after practical costs.
Exposure and missed-round accounting
Back counting reduces the number of neutral and negative rounds wagered, but the player can miss favorable rounds while waiting for a seat or moving between tables. Measure observation time, successful entries, hands played after entry, average stake, travel between tables and any forced flat betting. Comparing only the hands actually played can overstate the hourly result.
If mid-shoe entry is prohibited, the example cannot be executed at that table. If the count falls below the predetermined threshold immediately after entry, follow the exit rule instead of staying to justify the time already spent watching. A stopped shoe, early shuffle or changed table condition ends the original calculation.
A worked example of the hour that includes waiting
Suppose one hour contains 40 minutes of observation and movement and 20 minutes of actual play. During the playing time, you make 25 initial wagers of $40 each. Assume, solely for this illustration, that every wager has a 0.5% player advantage measured per initial wager under a correctly executed model. This is not an estimate for a real table or for the +2 entry threshold above.
Initial action is 25 × $40 = $1,000. The modeled expected gain is $1,000 × 0.005 = $5 before costs for the complete hour. Looking only at the 20 playing minutes and extrapolating would suggest $15 per hour. That figure assumes three times as much playable action as was actually available. If this hypothetical hour also costs $3 in expenses, its modeled net expectation is $2.
None of those averages predicts the result of that hour. The actual loss could be much larger than $5, and the bankroll must cover ordinary variation and permitted extra split or double wagers. Keep the edge’s denominator consistent: a percentage per initial wager already incorporates the modeled outcomes of those extra actions. Do not apply it again to extra cash committed. Side bets require a separate model.
Track observed, eligible and played rounds separately
Record three different counts: rounds you can observe completely, rounds meeting the tested entry criteria, and rounds you actually wager. An eligible round is not a played round if the seat is unavailable or entry is prohibited. Record the elapsed session time alongside them rather than reporting only the favorable segment.
Suppose you observe 80 complete rounds, classify ten as eligible and play four. The practical record is 80 observed, ten eligible and four played, not ten played. That distinction helps identify whether the main restriction is the game conditions, table access or your ability to follow the cards. It does not prove that any of the four wagers had a positive expectation.
What to do when the count is no longer reliable
If a blocked view, distraction or missed round prevents you from following all relevant exposed cards, do not treat the last reliable count as current. Stop using that incomplete estimate for entry or wager decisions. Restart from a known shuffle under the chosen method, or use another expressly validated method that accounts for the missing information.
A favorable number is useful only while its assumptions remain valid. Reaching a threshold once does not authorize a whole session at that stake. Review the true-count explanation for unit conversion and the risk-of-ruin guide before treating a modeled average as a bankroll plan.
Common errors
- entering from a running count without a correct true-count conversion;
- using strategy or betting thresholds from different rules;
- ignoring missed rounds, heat, travel time and variance;
- treating a positive estimate as a guaranteed session result.
Further reading
Use reproducible assumptions and conservative risk limits; counting does not remove variance.
