Editorial illustration for the lesson on spreads, bankroll, risk of ruin, in the Mayfair Casino School.
Spreads, bankroll, risk of ruin
Risk of ruin is the question every counter must answer honestly before they sit down. The maths is not complicated, but the numbers can be uncomfortable.
AC
Annabel Cavendish
Editor in Chief · Reviewed 14 May 2026
Annabel
0:000:00
Welcome to the lesson on spreads and bankroll.
I'm Annabel, and this lesson is where the arithmetic of counting becomes practical in the most direct sense.
Having an edge is not sufficient.
You need enough capital to survive the variance while the edge pays off, and you need your bet spread sized correctly relative to that capital.
Get either wrong, and the edge is theoretical, not real.
Let's start with variance, because it's the thing that surprises most people who approach counting for the first time.
The standard deviation of a single blackjack hand is approximately one point one five units, accounting for the effects of splits and doubles.
The variance per hand is the square of that: approximately one point three two.
For a session of five hundred hands, the standard deviation of the total outcome is one point one five times the square root of five hundred, multiplied by your unit size.
At a twenty-five-pound unit, that's approximately six hundred and forty-four pounds.
Your expected loss in the same session at half a percent house edge is twenty-five times five hundred times zero point zero zero five, which equals sixty-two pounds fifty.
The standard deviation is more than ten times the expected value.
You can play perfectly and still be down six hundred pounds at the end of a session.
That's not a malfunction.
That's the maths behaving exactly as expected.
This is why a counter needs to think in terms of long-term capital rather than session-by-session outcomes.
The spread is the ratio of your maximum bet to your minimum bet.
It determines how much of the count's positive fluctuations you actually capture.
A one-to-four spread barely extracts the available edge.
A one-to-twenty-four spread extracts more but invites immediate attention at any London casino floor.
In UK conditions, a spread of one to eight is moderate and carries manageable heat risk with good cover.
A spread of one to twelve is the upper boundary of what many experienced counters consider viable over multiple sessions at the same venue.
The design of the bet ramp matters as much as the spread.
The ramp should track the true count gradually, not jump binary from minimum to maximum.
One unit at true count plus one and below, two units at plus two, four units at plus three, six units at plus four, eight to twelve units at plus five and above is a reasonable structure.
The key is that the ramp looks like natural bet variation from the outside, not like a mechanical counter script.
The Kelly criterion gives you the theoretically optimal bet size.
It says bet a fraction of your bankroll equal to your edge divided by the variance of the game.
But full Kelly also produces dramatic drawdowns that are psychologically difficult to sustain.
Half-Kelly, betting half that amount, retains roughly seventy-five percent of the long-run growth rate while halving the variance.
Half-Kelly is the practical standard for working counters.
The risk of ruin question: how much bankroll do you actually need?
For a counter with a genuine effective edge of zero point seven five percent, playing a one-to-twelve spread with a twenty-five-pound minimum unit and a three-hundred-pound maximum, risk of ruin modelling suggests two hundred to three hundred maximum-bet units to reduce ruin probability below five percent.
That's sixty thousand to ninety thousand pounds.
This is the technically correct answer.
It's also the answer that stops most recreational counters cold.
The practical implication for a part-time counter at more modest stakes: if your unit is five pounds, your maximum bet is sixty pounds, and your total dedicated bankroll is three thousand pounds, that's approximately fifty maximum-bet units.
You need to understand which risk category your bankroll places you in, and then decide whether that's acceptable.
Most part-time counters accept a higher risk of ruin in exchange for a lower capital requirement.
That's a legitimate choice if it's made consciously rather than by default.
At Aspers Westfield Stratford or the Hippodrome on Leicester Square, both running six-deck games at standard penetration, the effective edge over all hands, including the majority of hands played at minimum bet during neutral or negative counts, runs approximately zero point five to zero point seven five percent for a well-calibrated counter.
The peak-count edge sounds impressive; the effective overall edge is considerably more modest because you're betting minimum for the majority of hands.
The bankroll framework needs to accommodate venue rotation as well.
Your operational lifespan at any single venue is finite.
The capital has to survive both the variance and the rotation across enough venues to reach statistically meaningful hands.
Size your bets to your bankroll.
Treat the bankroll as a long-term instrument.
Don't improvise the ramp.
Begin with what risk of ruin actually measures, because the definition changes the practical advice considerably.
Risk of ruin is not the probability of losing your session bankroll. It's the probability of depleting your entire intended playing bankroll before the edge pays off. A counter with a genuine 1% edge, playing 1,000 hands per trip, will still lose money on a significant proportion of those trips. The edge is real; the variance is also real, and in blackjack it is substantial. The standard deviation of a single blackjack hand is approximately 1.15 units (slightly higher when splits and doubles are considered). That means that in a 500-hand session, the expected standard deviation of outcome is roughly 1.15 x sqrt(500) = approximately 25.7 units. At a £25 unit, that's a one-standard-deviation range of over £640 around your expected outcome. You can be doing everything correctly and still be down £600 at the end of the session. This is not a malfunction of the mathematics; it's the maths behaving exactly as expected.
The N0 Concept: When Does the Edge Show Up?
N0 (pronounced N-zero) was introduced by Don Schlesinger in Blackjack Attack as a measure of how long a counter must play before their edge is statistically distinguishable from luck. Formally, N0 is the number of hands at which the expected profit equals one standard deviation of results. Above N0, the signal is starting to emerge from the noise; below it, you're in the zone where short-term results dominate.
The formula is: N0 = variance / (expected value per hand squared).
For a counter with an effective edge of 0.75% per hand (including all hands, not just the high-count hands) and a standard deviation of approximately 1.15 units per hand, the variance is 1.32. N0 = 1.32 / (0.0075 squared) = 1.32 / 0.0000563 = approximately 23,400 hands. That is a significant number. At 50 hands per hour for 4 hours per session, that's roughly 117 sessions before the edge is statistically visible above the noise at one sigma. At one sigma, you've established an edge; you haven't proved it conclusively.
The practical implication of N0 is that a counter needs to treat their bankroll as a long-term instrument, not a session-by-session P&L account. Evaluating your counting performance after 20 sessions is statistically meaningless; evaluating it after 200 sessions starts to become meaningful. The bankroll needs to survive long enough for N0 to be reached. That's the design constraint.
Optimal Bankroll for a 1-to-12 Spread in UK Conditions
The relevant numbers for a counter playing a typical six-deck S17 UK game with a 1-to-12 spread are approximately these. The effective edge over all hands (including the approximately 65% of hands played at minimum bet or near-minimum) is typically 0.50-0.75% with strong penetration, falling to 0.30-0.50% with weak penetration (below 70%). The effective edge is much lower than the peak-count edge because you're betting 1 unit for the majority of hands.
For a player with a £25 unit (minimum bet) and a maximum bet of £300 (12 units), the bankroll considerations work as follows. our risk of ruin modelling for blackjack at these parameters suggests that a bankroll of 200 to 300 maximum-bet units (so £60,000 to £90,000 in this case) keeps risk of ruin below 5%. That is the technically correct answer, and it is also the answer that stops most recreational counters cold. Professional counting at serious unit sizes requires serious capital.
The practical implication for a part-time counter playing at more modest stakes: if your unit is £5, your maximum bet is £60, and your bankroll is £3,000 (200 minimum units, 50 maximum bet units), your risk of ruin is not 5%; it's substantially higher. Blackjack Forum Online's analysis indicates that a 50-unit bankroll at maximum bet carries a risk of ruin in the range of 30-40% for a counter with a realistic 0.5-0.75% edge. At 100 maximum-bet units, it falls to approximately 13-18%. You need to understand which category your bankroll puts you in before you sit down at Aspers Stratford or the Hippodrome with a spread.
The calculation at the £5 unit level, which is where most learning counters operate: £5 unit x 200 maximum-bet units = £10,000 bankroll for genuinely low risk of ruin. That is not a small number for a hobby player. Most people learning to count will accept a higher risk of ruin in exchange for the lower capital requirement, and that's a reasonable choice if it's made consciously rather than by default.
Bet Spreading and Operational Longevity
The bankroll analysis above assumes you'll be playing at the same venue indefinitely. In practice, UK casino conditions don't allow unlimited play by identified counters; the operational lifespan at any given venue is finite. This changes the bankroll calculation in an important way: you need your bankroll to survive both the variance and the session count before you've exhausted your welcome at your preferred venues.
The relationship between spread size and operational visibility is direct. A 1-to-4 spread is essentially invisible; it produces very little edge but almost no heat. A 1-to-8 spread is within the range of play variation a good recreational player might show; it produces moderate edge and moderate heat risk over extended play. A 1-to-12 spread, consistently applied with perfect correlation to the count, is what a competent pit boss will identify as counting behaviour within a few hours of observation at a quiet venue. At a busy venue like the Hippodrome during peak hours, the same spread has more ambient camouflage.
The professional answer, detailed in the heat and cover lesson, involves managing your exposure across venues, introducing controlled imperfections in your spread behaviour, and setting a session session-depth limit that keeps individual venue exposure below the threshold of conclusive identification. The bankroll framework needs to accommodate multiple venues, not just multiple sessions at one. The bankroll calculator can model the impact of venue rotation on your effective hourly rate and capital requirements.
Key numbers
Bankroll (max bet units)
Approximate RoR (0.75% edge)
N0 (hands)
Sessions to N0 (50 hands/hr, 4hr sessions)
50
~35%
~23,400
~117
100
~13%
~23,400
~117
200
~5%
~23,400
~117
300
~2%
~23,400
~117
400
~1%
~23,400
~117
Sources: our risk of ruin analysis, Blackjack Forum Online RoR analysis, our calculation.
