Baccarat analyzer numbers explained: what each one counts and how they held up in real shoes

2026-10-13

A big road fills one side of the screen, with a few numbers above it. Banker 17, Player 12, Tie 3. Next to them, a box blinks: next hand Banker 72%.

These numbers are an example of a typical screen.

Most programs and apps sold as a baccarat analyzer look something like this. Because there are numbers, it looks as if something has been worked out. This post first spells out what each of those numbers counts, then runs the same arithmetic on 26,454 real collected shoes.

1.The numbers on an analyzer screen come in three kinds

Analyzers look different, but their numbers mostly fall into three groups.

• Counts and shares: how many times Banker, Player and Tie came up in this shoe, and the percentages
• Derived road colors: the big road redrawn in red and blue to show whether column shapes repeat the columns before them
• Next-hand probability: how often each side followed the current shape earlier in this shoe

The first two are records of past hands, arranged another way. Only the third looks like it says something about what comes next, so it is the center of this post.

Next-hand probability: within one shoe, Banker results after the current shape divided by that shape's count.

2.What does the next-hand probability box count?

We fixed one method and counted with it. Take the last three non-tie hands (eight possible Banker and Player shapes), look back through the same shoe at what followed that shape, and show the Banker share.

Say the last three hands are Banker, Player, Banker, and that shape came up seven times earlier in the shoe with Banker following five times. The screen shows 71%. Tiny samples make the number jump around, so we only showed it once the same shape had at least five cases.

With that rule, the box lit up 513,890 times across the real shoes. First, how often did it look lopsided?

• Showing 70% or more: 18.03%
• Showing 30% or less: 15.88%
• Together: 33.91%

About one look in three, the box showed one side ahead by 70% or more.

Next, we checked whether the side it pointed to actually came up more. For each moment, we recorded whether the very next hand was Banker and grouped those by the number on screen.

After the 92,662 times it showed 70% or more, the next hand was Banker 50.67% of the time. After the 81,627 times it showed 30% or less, Banker came up slightly more, at 51.06%. Between 40 and 60%, the figure was 50.61%, hardly different.

Banker over this whole data set was 50.70%, in line with the clean-shoe reference of 50.68%. Each box has 80,000 to 130,000 cases, so chance alone moves it by about 0.3 points either way. All five sit near that range.

Whether the analyzer said 70% or more or 30% or less, the next hand was Banker about 50.7% of the time.

The number swings because the sample is small. A shoe has a little over 60 non-tie hands, and split across eight shapes that is seven or eight cases each. Five heads out of seven coin flips is common.

Counting with the last four hands (16 shapes) gave the same picture. After 70% or more it was 50.56%, after 30% or less 51.26%, with about 20,000 cases each and a range near 0.7 points.

Results for the 32 five-hand shapes counted across many shoes are in our baccarat pattern recognition post. This post looks only at the number an analyzer shows inside one shoe.

3.Does the running shoe percentage tell the rest of the shoe?

The counts and shares box got the same treatment. At the 30th non-tie hand, we split shoes where Banker stood at 60% or more from those at 40% or less, then counted the rest of each shoe.

Shoes at 60% Banker or more by hand 30 were 20.19%, matching the coin-flip 20.18%. Their remaining hands were Banker 50.50% of the time. Shoes at 40% or less, 16.25% (math 16.12%), went on at 51.00%.

A shoe that leaned Banker over its first 30 hands did not lean Banker over the rest. The gaps of 0.2 to 0.3 points are about the size of the noise range for 130,000 to 160,000 hands.

4.What to look at next to any percentage

Look at the number of hands next to any share. In a game where Banker comes up 50.70% of the time, chance alone commonly gives 20 to 80% over 10 hands, 33 to 70% over 30 hands and 38 to 63% over 60 hands (binomial, 95%).

In the real shoes, the middle 95% of Banker shares at hand 30 ran from 33.3 to 70.0%, the same as the math.

An analyzer does have a use in laying out the record neatly. Questions that separate a record tool from a prediction tool are in our baccarat pattern analyzer post.

Hiddenpick is a Windows program that keeps on its server the betting rules that pick owners do not make public, and runs them as they are on the PC of the person who received them through a referral code. It currently runs on Evolution live baccarat at Stake (stake.com). Whether prediction programs can call the next hand is covered with real shoe numbers in the baccarat prediction guide.

We are recruiting national pickster representatives, and those selected receive benefits worth $12,000, a one-year pass. Apply: hiddenpick.my/pickster