How to Choose Blackjack Card Counting Software

For home practice, choose a trainer that teaches skills and exposes mistakes. For estimating the effects of rules, stakes and risk, choose simulation or analysis software. Neither type reveals the next card or guarantees profit.

The practical choice is CVBJ for practice, CVCX for game and betting scenarios, and CVData for advanced custom strategies. BJA is a separate training ecosystem. If you are still learning basic strategy and the running count, you may not need to pay for software yet.

Choose by task, not by a “best software” ranking

Choose the tool that addresses your actual problem: playing decisions, counting accuracy or the consequences of specified rules and stakes. There is no universal winner here, and brands are not ordered by affiliate priority.

These uses are documented on the official pages for CVBJ, CVCX, CVData and the QFIT product family. BJA describes its offerings on its software, training drills and membership features pages.

These are documented features, not findings from hands-on testing of the paid products. We did not run CVCX or log into BJA's paid tools. The examples below cover reproducible arithmetic, preparing assumptions and, separately, the behavior of our inline calculator.

Prices, platforms and official demos

The QFIT catalogue checked on October 12, 2026 lists CVBJ V5.6 at US $115, CVCX V6 at US $125, CVData V6 at US $200, and the CVBJ + CVCX bundle at US $200. Get demos from QFIT's official download page.

The price source is the current QFIT catalogue, not the legacy CVCX page that lists $75. These are dated catalogue snapshots; taxes, delivery, region, store and the final checkout amount may differ.

Original English QFIT catalogue showing CVCX V6 priced at US $125

Source: QFIT's official English catalogue, checked October 12, 2026. The original English image text is preserved.

QFIT's desktop products are for Windows. The official downloads page provides demos and platform information. Native Mac support has not been verified, and emulation should not be treated as a tested solution.

QFIT's separate mobile games and drills are not a mobile version of desktop CVCX. According to BJA's developer, its free public drills are desktop-only; its mobile app and web membership are separate products. No unverified membership prices are listed here.

Paying is unnecessary if your immediate goal is to learn rule-specific decisions, card tags and accurate counting. A paid tool becomes useful when you have a defined need for practice feedback, scenario comparisons or detailed simulation.

Build accuracy before speed

Learn basic strategy for your rules first, then practice the running count, remaining-deck estimation and finally index plays. Increase speed only after accuracy is reliable.

Start with rules and basic strategy

Basic strategy gives the rule-specific best play for your hand against the dealer's upcard. It is not a counting system, a betting progression or a guarantee of profit.

Open the free basic-strategy chart, select the actual number of decks, blackjack payout, dealer drawing rule, double-after-split rule, surrender option and dealer peek setting. Then read the chart for that rule profile.

Dealer stands on soft 17 (S17) means the dealer stops on a 17 containing an ace counted as 11. Dealer hits soft 17 (H17) means the dealer draws another card on that hand. These abbreviations describe dealer procedures, not hard hands.

The chart does not track the running count and is not a CVCX substitute. Use the basic-strategy flashcards for dedicated decision practice.

Practice the Hi-Lo running count

The Hi-Lo count assigns +1 to 2–6, 0 to 7–9, and −1 to 10, J, Q, K and A. The running count is the sum of exposed card tags, before adjustment for decks remaining.

For the exposed sequence 2, 5, 10, A, 6, 9:

+1 +1 −1 −1 +1 +0 = +1.

Start at 0 with a newly shuffled standard shoe and reset after each shuffle. The tags and conversion method are explained in the Hi-Lo mathematics guide.

Convert to the true count

For Hi-Lo, divide the running count by the estimated number of all undealt decks, including cards behind the cut card. Do not count only the cards still available before the next shuffle.

With a running count of +6 and 1.5 decks remaining:

+6 ÷ 1.5 = true count +4.

This is an arithmetic example, not a betting recommendation or a guaranteed win. It gives an exact whole-number result, so no rounding is needed.

If a six-deck shoe is planned to deal 4.5 decks, shoe penetration is 75%, leaving 1.5 decks. Do not divide by 4.5: that is the amount dealt, not the amount remaining. Six standard decks contain 312 cards; 75% is 234 cards before the marker and 78 behind it. Burn cards and completing the final round can make actual card use differ. See the shoe and penetration guide for more context.

Use a consistent deck-estimation and true-count rounding convention. An index number is the count threshold at which a specified hand decision changes from basic strategy. Your Hi-Lo index plays must match the rules and rounding convention; indexes are not universal across incompatible conditions.

Log different errors separately

Track running-count errors, deck-estimation errors and decision errors separately, along with elapsed time. This makes it clear which skill needs more practice.

Slow down when errors increase. Raise the pace only after reliable accuracy. There is no universal number of seconds that certifies readiness, and a strong home drill score does not prove profitable real-world play.

Complete full practice sessions using simulated or virtual credits. Practice does not require a casino deposit.

Prepare a complete scenario before trusting a simulation

Define the rules, shoe penetration, counting system and stake policy before reading the outputs. If a condition is unknown, do not replace it with a favorable assumption.

Prepare this input checklist:

• Deck count and blackjack payout.
• Whether the dealer stands or hits on soft 17.
• Double after split, surrender and dealer peek rules.
• Planned shoe penetration and treatment of hidden or burned cards.
• Counting system, compatible index numbers, deck estimation and true-count rounding.
• Minimum and maximum stakes, plus the actual stake at each true count.
• Your bankroll, whether you play one or multiple hands, and any counts at which you sit out.
• Playing rate and how skipped hands affect that rate.

This is a checklist for preparing an analysis using documented variables, not an untested walkthrough of CVCX menus.

A hypothetical starting scenario

A hypothetical scenario could use six decks, a 3:2 blackjack payout, dealer stands on soft 17 (S17), double after split, late surrender and dealer peek, with 4.5 decks planned to be dealt—75% penetration. Late surrender means surrender after the dealer's blackjack check.

Next, specify Hi-Lo, compatible index numbers, deck estimation and true-count rounding. The example minimum is $25 and maximum is $200, a 1:8 bet spread: the ratio of minimum to maximum stake. A rate of 80 hands per hour is only a scenario assumption, not an observed table rate.

A 1:8 spread alone does not define the stakes at each true count, the average bet or the player edge. Before calculating, supply that stake-by-count mapping, your own bankroll, one or multiple hands and any sit-out conditions. Until then, this is not a complete simulation scenario. None of these example conditions is a recommendation to play or bet.

Compare one change at a time

Change one variable at a time and save the assumptions with the report. For example, compare 50% versus 75% penetration while holding everything else constant, or compare dealer stands on soft 17 (S17) with dealer hits soft 17 (H17).

We did not run these comparisons in CVCX, so no numerical outcomes are claimed. The point is to isolate one condition's effect rather than mix it with other changes.

Read expected value and risk in context

Expected value (EV) is a long-run mean under stated assumptions, not a salary or a forecast for your next session. Risk of ruin also requires a bankroll boundary, a horizon or goal, and a stake policy.

Consider this independent hypothetical example, not a CVCX result or the inline calculator's default:

• Average initial stake: $50.
• Assumed positive player edge: 1% of total initial wagers.
• Rate: 80 hands per hour.

Expected net profit before costs is:

$50 × 0.01 × 80 = $40 per hour.

Here, the edge is expressed per initial wager unit. Total initial wagers exclude additional money committed through doubles and splits, and side bets are excluded. Those additional stakes increase cash exposure; do not mix an initial-wager denominator with a total-cash-turnover denominator.

Variance and standard deviation (SD) describe the dispersion of results. The standard error describes sampling uncertainty in the estimated mean net monetary result per declared unit—hand or wager. It is not a win probability or the frequency of winning hands. More simulated hands reduce sampling error, but they do not correct wrong rules, an incorrect strategy or an unsuitable model. There is no universal software “accuracy percentage.”

Before interpreting risk of ruin, ask:

1. What lower bankroll boundary counts as ruin?
2. Is the horizon a trip, a finite session or unlimited play?
3. Is there a target to reach before ruin?
4. How do stakes change, and when does play stop?

Finite-trip risk, unlimited-horizon risk and the probability of ruin before reaching a target are different measures. Compare like with like, and do not treat 5% as universally safe. Costs, errors and rule uncertainty can reverse a small modeled edge.

For more detail, use the risk-of-ruin guide and blackjack bankroll management guide.

What the inline calculator does—and does not—show

The inline calculator produces outputs from a simplified formula; it does not simulate a particular shoe. Do not use it to certify profitability, choose stakes or determine your personal bankroll requirement.

Important before using it: there are no inputs for rules, penetration, your bankroll or the risk horizon. Its displayed 5.0% risk is not measured for capital you entered—there is no such input. Its profit, bankroll and time-to-goal estimates have not been validated as table estimates.

To reproduce the example, select Hi-Lo in the system dropdown and 6 decks in the deck dropdown. Enter 25 in the minimum-stake field, set the spread slider to 8, and set the hands-per-hour slider to 80. These controls do not specify the other conditions needed for a complete scenario.

English calculator inputs showing Hi-Lo, 6 decks, a $25 minimum stake, a 1:8 spread and 80 hands per hour

Source: an actual English ToolsGambling UI capture from a local localhost run with the fixed example. This is not a production-site check.

With those settings, the full widget displays +0.14%, +$8.54 per hour, 5.0% risk, $8,132,900 bankroll, and 1,172 hours to $10,000.

English calculator results showing +0.14% player edge, +$8.54 hourly EV, 5.0% risk and a $8,132,900 bankroll

Source: the same local English ToolsGambling UI run. The screenshot records displayed outputs, not a verified profit forecast or capital requirement.

Why these outputs do not describe your game

The outputs depend on built-in assumptions rather than your table's rules. We independently reproduced the implementation arithmetic, not the validity of these figures for real play.

The formula fixes the average true count at 1.5, uses a base house edge tied to deck count, and assumes this average bet:

25 × (1 + 7 × 0.28) = $74.

For six decks, its formula produces a player edge of 0.1441936%, then calculates:

74 × (0.1441936 ÷ 100) × 80 = $8.536262 per hour,
displayed as $8.54.

These are the supplied formula values, shown at their stated precision; the UI rounds the hourly amount to cents. This is not a count of a card sequence or a simulation of hands. Although the footer describes simulation averages, the implementation does not simulate individual hands.

The implementation selects 5% risk internally, then derives a required bankroll from it. Changing only the rate from 80 to 160 hands per hour doubles displayed EV to $17.07 and the model bankroll to $16,265,800. That exposes the implementation's dependence on the rate parameter; it does not establish a real capital requirement.

Do not use the widget's outputs, verdict or hours-to-goal figure as profit certification or stake and bankroll advice. For analysis of declared rules and strategies, consider the documented capabilities of CVCX or CVData, with a complete set of assumptions.

Online blackjack, live shoes and automatic recognition

Counting depends on a persistent inventory that depletes as cards are dealt, together with knowledge of exposed cards. The presence of an RNG or a live dealer alone does not establish the shuffle schedule.

An RNG is a randomness source, not inherently a reshuffling procedure. If software reshuffles before each hand or samples cards with replacement, previous hands do not create a persistent shoe count.

A live shoe still requires known rules, penetration, treatment of burned and hidden cards, and the provider's and operator's terms. An early cut card reduces available depth, but there is no universal profitable penetration threshold.

A continuous shuffling machine (CSM) returns used cards to ongoing mixing. Recycling cadence varies by device and approved procedure, so ordinary shoe-count assumptions cannot be copied blindly.

Automatic card recognition does not resolve these limitations. Missed cards, duplicate readings and recognition errors corrupt the inputs. Seeing public cards cannot reveal the unseen order. No AI or live-counter recommendation has been verified here.

Watch for tools that promise or request:

• Guaranteed profit or next-card prediction.
• High accuracy without explaining what was measured.
• An unverifiable provider seal.
• An unofficial installer that asks for account details, payment information or access secrets.

Use official download links. Home training and arithmetic checks do not require a casino deposit.

Electronic assistance and the law

Mental counting and electronic assistance during play are distinct. Training at home does not authorize using software at a casino table; local law and operator terms matter.

Nevada is one specific example, not a rule for the entire United States or other countries. NRS 465.075 prohibits using, or possessing with intent to use, advantage devices or software, including tools for tracking cards and analyzing strategy or probabilities, subject to approved-game and Commission exceptions. The cited official revision is dated April 15, 2026, and includes the 2025 statutes.

Do not automatically apply that example to other jurisdictions. Operator rules also matter, and a casino may limit or refuse play. This guide does not provide ways to bypass restrictions.

Your next step

Start with free practice if you are learning; choose a paid product only when it solves a specific training or analysis problem.

• Learning basic strategy: select your actual rules in the free chart, then practice decisions and the running count without real-money bets.
• Needing structured error feedback: review CVBJ and try its official demo.
• Comparing stakes, rules and risk: prepare complete assumptions for CVCX.
• Developing custom strategies and index numbers: review CVData.
• Considering BJA: distinguish its free desktop drills, mobile app and paid web membership before choosing.

A useful tool answers your question under understandable assumptions. An attractive forecast without those assumptions replaces neither accurate practice nor sound modeling.