Every card counting system arrives attached to a spec sheet. Alongside the tag values, the level, and whether the count is balanced, three decimals follow it around the literature: betting correlation, playing efficiency, and insurance correlation. Browse any comparison of systems and the numbers line up like fuel economy ratings on a car lot.
The numbers are real and they measure something, but each one compresses an entire system into a single figure, and the compression loses more than most shoppers realize. A count does three loosely related jobs, steering bet sizes, guiding departures from basic strategy, and calling the insurance bet, and no one decimal rates all three at once.
This piece defines the three metrics, puts the published figures for seven well known systems side by side, and sets out what the comparison can and cannot tell you. First, the context: counting with your head is legal in the United States, casinos are private businesses that may refuse service to suspected counters, and no technique removes risk from the game. The metrics are interesting as engineering; they are not a promise of anything.
The three metrics, defined
Betting correlation, in the standard description, measures how closely a count's tag values track the effect each card rank's removal has on the player's expectation. Removing a 5 from the deck helps the player; removing an ace hurts, and by different amounts. That removal effect is exactly the information a count uses to size bets, which is where the name comes from. The concept underneath it is covered from scratch in effect of removal.
Playing efficiency measures how well the count guides departures from basic strategy. A counter occasionally overrides the chart, standing on 16 against a 10 or doubling 10 against a 10, when the remaining composition justifies it, and this metric rates how faithfully the count signals those moments. Insurance correlation is the narrowest of the three: it measures how well the count tracks the insurance decision specifically, which depends on the proportion of ten value cards remaining, as explained in why basic strategy declines insurance.
All three are conventionally reported on a scale from 0 to 1, where 1 would be a perfect correlate of the underlying quantity. The figures below are the ones commonly quoted for the classical systems, and they should be read as approximate summaries rather than constants of nature: they are computed under stated assumptions about rules and deck count, and slightly different assumptions produce slightly different decimals.
The published numbers, side by side
| System | Betting correlation | Playing efficiency | Insurance correlation |
|---|---|---|---|
| Hi-Lo | .97 | .51 | .76 |
| KO | .98 | .55 | .78 |
| Zen | .96 | .63 | .85 |
| Omega II | .92 | .67 | .85 |
| Hi-Opt I | .88 | .61 | .85 |
| Hi-Opt II | .91 | .67 | .91 |
| Red Seven | .98 | .54 | .78 |
Figures as commonly published for each system. A quick orientation to the systems themselves: Hi-Lo is the balanced level one standard, KO and the Red Seven are unbalanced level one counts that skip true count conversion, and Zen, Omega II and Hi-Opt II are level two systems with finer tag values. Hi-Opt I, the level one half of the Hi-Opt pair, is balanced and leaves aces at zero.
21 TrainerFree on the App Store. The optional Pro upgrade teaches Hi-Lo, KO, Omega II and Zen with counting drills and virtual chips.
Get the app
What the table actually says
The first thing the table shows is that betting correlation is high for every system listed, from .88 to .98. Even the level one counts capture most of the betting signal, because bet sizing mainly needs to know whether the remaining cards run high or low, and a simple plus and minus split of the ranks tracks that well. The unbalanced KO and Red Seven post the highest betting correlation here, .98. Their tags differ from Hi-Lo's only in counting 7s as plus one (every 7 in KO, the red ones in Red Seven), and removing a 7 does help the player slightly, so the extra tag nudges the correlation up.
Playing efficiency varies much more, .51 to .67, and the split follows system level. The level two counts, Zen at .63 and Omega II and Hi-Opt II at .67, all sit above the level one counts: Hi-Lo at .51, Red Seven at .54, KO at .55 and Hi-Opt I at .61. The reason is tag resolution: many playing departures depend on ranks that Hi-Lo lumps together, such as a 5 and a 2, and counts that assign different values to those ranks carry more of the information the departure needs.
The third pattern is the quietest and most interesting. The ace neutral counts, Omega II and both Hi-Opt systems, post the lowest betting correlations in the table, .92, .88 and .91 against Hi-Lo's .97. Per card, the ace's removal effect is among the largest of any rank, much of it because the ace powers 3:2 blackjacks, so a count that ignores aces gives up betting precision on purpose, usually to keep the playing signal clean. That tradeoff is why these systems are so often paired with an ace side count, a second tally that restores the missing betting information at the cost of attention.
The honest limits of the comparison
Now the caveats, because a table like this invites over-reading. These are single number summaries of systems that will be operated by humans under table conditions. A few points of playing efficiency are worth something only if the count behind them stays accurate at dealing speed with conversation in the background, and a harder count is harder precisely where the points are collected. A count lost mid shoe can easily cost more than the gaps in the table are worth, and recovering from that is its own skill, covered in what happens when you lose the count.
For most learners the decision that matters is level one versus level two, and execution dominates the comparison: an accurate simple count outperforms a stumbling complex one. The best system for a learner is the one they can run cleanly, which is the framing our guide to choosing a counting system works through system by system. And none of the metrics changes the basic caution: no counting technique removes risk or promises winnings.
Where 21 Trainer fits
The app teaches four counting systems, Hi-Lo, KO, Omega II and Zen, four of the seven in the table above, with drills that run from single cards up through multi-card hands and true count conversion. Progress shows up in accuracy stats and session history rather than in decimals, which is the right direction: for a learner, measured accuracy at speed matters more than any decimal in the table.
The bankroll simulator runs on virtual chips only, which keeps the practice loop entirely educational. Pick the system that matches your appetite for complexity, drill it until it survives distraction, and let the efficiency metrics inform the choice rather than make it.
Learn a count at your paceFour counting systems, drills, stats and heat maps. No account, no ads, works offline.
Download
Frequently asked questions
What is betting correlation in card counting?
It measures how closely a count's tag values track the effect each card rank's removal has on the player's expectation, which is the information a count uses to size bets. A high betting correlation means the count moves in step with how good the remaining deck actually is. It is scored on a scale where 1 would be perfect, and most common systems score between about .88 and .98.
What is playing efficiency in card counting?
It measures how well a count guides departures from basic strategy, the index plays a counter makes when the deck composition justifies them. Counts with finer tag values, like the level two systems, generally score higher because they distinguish ranks that coarse counts lump together. In the commonly published figures it ranges from about .51 for Hi-Lo to .67 for Omega II and Hi-Opt II.
Which counting system has the best numbers?
No single system wins every metric. In the commonly published comparison, KO and Red Seven post the highest betting correlation at .98, while Omega II and Hi-Opt II lead playing efficiency at .67 and Hi-Opt II leads insurance correlation at .91. Which metric matters most depends on how you intend to use the count, and for most players bet sizing matters more than index precision.
Do you need an ace side count to count cards?
No, several strong systems count aces directly. But ace neutral counts like Omega II and the Hi-Opt systems score lower on betting correlation precisely because aces matter so much to bet sizing, and adding a side count of aces is the standard way to recover that. It costs attention at the table, so it is a tradeoff rather than a free upgrade.
Is card counting against the rules or illegal?
Counting with your head is legal in the United States. Casinos are private businesses that may refuse service to suspected counters, and no technique removes risk from the game or promises winnings. Counting at best produces a small long run edge with large short term swings. This article covers the technique as a skill, using virtual chips only, and nothing here encourages real money play.