How Much to Bet When the Count Is Good: Bet Spreads, Bankroll, and Risk of Ruin for Live-Dealer Blackjack

A true count is not money. It is a forecast. The money comes from one thing only: betting more when the forecast is good than when it is bad.

If you flat-bet the same amount every hand, a perfect count earns you almost nothing. You would still be paying the house edge on the many hands where the shoe is neutral or poor, and you would be under-betting the few hands where it turns in your favor. Counting without a bet spread is a hobby. Counting with one is a strategy — and the spread is also where the risk lives.

This post is about the three numbers that decide whether that strategy survives contact with real variance: how big a betting unit is, how wide the spread is, and how much bankroll stands behind it.

Why the bet has to move

In broad Hi-Lo terms, each point of true count is worth roughly half a percent of player edge. A typical 8-deck, dealer-stands-on-soft-17 live table starts around half a percent in the house's favor. Put those together and you get a rough ladder:

These are rules of thumb, not constants. Rules, deck count, and how deep the dealer cuts all shift them. That is exactly why the calculator shows an Edge % computed from the actual remaining cards rather than a chart. But the shape is always the same: most of the shoe is negative, a minority of hands are positive, and the positive hands only matter if you bet more on them.

So the question is not "what is the count?" It is "how much more do I bet, and when?"

A betting ramp, in plain units

Pick a unit: the smallest bet you will make, usually the table minimum or a little above it. Then decide how many units you bet at each true count. That table is your ramp, and the ratio between your largest and smallest bet is your spread.

A simple 1–8 ramp:

Three rules make a ramp work:

Do not raise before the edge turns. At +1 you are roughly even. Raising there is just paying more house edge at a higher stake. The first raise belongs at +2.

Raise in steps, not leaps. Doubling per true count keeps the bet proportional to the edge, which is roughly what optimal sizing asks for.

Use the true count, not the running count. A running count of +8 with six decks left is a true count around +1. That division is the whole reason the true count exists.

How big should a unit be? (the Kelly idea)

The classic answer is the Kelly criterion. In blackjack terms it reduces to a simple rule:

Bet roughly your edge divided by about 1.3, as a fraction of your bankroll.

The 1.3 is the variance of a single blackjack hand in betting units — splits and doubles make blackjack swingier than a coin flip.

So with a +1.5% edge, a full-Kelly bet is about 1.5% ÷ 1.3 ≈ 1.15% of bankroll. If that is your top bet, and your top bet is 8 units, then one unit is about 1.15% ÷ 8 ≈ 0.14% of bankroll — roughly bankroll ÷ 700.

Full Kelly maximizes long-run growth, but it is a rough ride: big drawdowns are normal. Most serious players bet half Kelly or less. That gives a practical starting point:

One unit ≈ bankroll ÷ 1,000 to bankroll ÷ 1,500.

With a $5,000 bankroll, that is a unit of about $3–5. With a $25,000 bankroll, about $15–25. If the table minimum is bigger than your Kelly unit, the honest answer is that the table is too big for your bankroll — not that you should round up.

Risk of ruin, worked out

Risk of ruin is the chance that your bankroll hits zero before the edge has time to show up. A widely used approximation is:

Risk of ruin ≈ e^(−2 × EV × Bankroll ÷ Variance)

where EV is the average win per hand, Variance is the variance per hand, and all three are in units. You do not need to compute it at the table, but walking through it once shows why the type of table matters more than anything else.

The numbers below are illustrative estimates, not simulation output. They use rough frequencies for how often each true count appears, plus the edge ladder above. The point is the difference between the two tables, not the decimals.

Table A: 8 decks, cut at about 50%

This is the most common live-dealer Classic setup. With a shallow cut, the true count rarely drifts far from zero before the shuffle. Rough frequencies: about 88% of hands at +1 or below, 7% at +2, 3% at +3, 2% at +4 or higher.

• With the 1–8 ramp above: the expected value comes out slightly negative — around −0.06% of a unit per hand. The good hands are too rare to pay for the bad ones.
• Stretch it to 1–12 (12 units at +4 and up): expected value is about +0.07% of a unit per hand, with a standard deviation of about 2.4 units per hand.

Plug that into the formula:

That is not a strategy. That is a very slow coin flip with a large bankroll tied up in it. On a shallow 8-deck shoe, a modest spread barely clears break-even.

Table B: 6 decks, cut at about 75%

Deeper penetration means the count reaches +3 and +4 far more often before the shuffle. Rough frequencies: about 79% of hands at +1 or below, 9% at +2, 6% at +3, 6% at +4 or higher.

With the same 1–8 ramp, expected value is about +0.74% of a unit per hand, and the standard deviation is about 2.8 units per hand.

Same player, same spread, same discipline — and roughly ten times the expected value, with a fraction of the bankroll needed. Penetration is not a detail. It is the game.

That is why the first step is always choosing a table that can actually produce a high count, and why it is worth measuring the cut on a table before you bring money to it.

Spreading on a live-dealer table

Online live tables change a few things compared with a casino floor.

You usually cannot wong. On a land-based table you can watch from behind and jump in when the count is high. Live-dealer tables generally make you sit from the start of the shoe. You play the bad hands too, which is why the ramp's bottom bet matters so much.

Every bet is logged. There is no floor person to charm and no cover play that hides a pattern from a database. A 1–8 spread on a table that keeps hitting +3 looks like a counter whether you smile or not. A 1–20 spread looks like one immediately. Operators can limit your spread, cut the shoe earlier, or restrict the account under their own terms. Counting is legal; being welcome is a separate question.

Minimums can be your friend. Small live minimums let a modest bankroll use a correct unit size. If the minimum forces you to bet more than about bankroll ÷ 1,000 per unit, you are over-betting before the first card.

A practical checklist

1. Decide your bankroll before the session, and only count money you can lose without changing your life.
2. Set your unit at roughly bankroll ÷ 1,000 to ÷ 1,500.
3. Write your ramp down, e.g. 1 unit up to +1, then 2, 4, 8. Do not improvise it at the table.
4. Only sit at tables whose cut lets the count move. If a table never reaches +3 across several shoes, it is Table A.
5. Raise only when the true count says so — not after three losses, and not because +1 "feels hot."
6. Re-size when your bankroll changes. Kelly is a fraction of what you have now, not what you started with. After a losing stretch, the unit goes down.

Let the calculator tell you when

You do not have to hold the running count in your head and divide it by an estimate of the discard tray while a dealer is moving. The calculator tracks every card you tap and shows the true count and the edge on the current shoe. When the edge goes positive, that is your cue to step up the ramp. When it is still negative, the bet stays at one unit — no matter how the last hand went.

Try a full shoe in demo first, without betting, and watch how often the edge actually crosses into positive territory on the table you plan to play. That one shoe will tell you whether it is a Table A or a Table B — before your bankroll has to.

Want to know whether all this is worth the time? Read an honest breakdown of what online card counting actually earns. For the math behind the edge numbers, see expected value in blackjack, and for what the app is counting, the Hi-Lo guide.