Roulette Wolfie Ladder version G - Last updated: October 3, 2026

84% win rate based on all 1,486 runs ( SEE results for more details )


this is my trial / work in progress version of the wolfie ladder system. see version D for the full system with recovery.

the goal is trying to get a system with small bets on a elecronic roulette table with min bet $3 and max be $250 on a dozen bet.


update 10/01/2026: removed the waiting for a win for stage 1.5 and stage 2.5 seemed to make no notibale difference in the win rate. also added database to save the runs and display the win rate on the results page. Also removed step 7 to keep the system simple and fast we have a 5% chance of hitting this but we will recover it. see version A for having that step included back in.

update 10/02/2026: removed step 14 to keep the system simple and fast we have a 5% chance of hitting this but we will recover it. see version B for having that step included back in. removed step 22 ( see version C for having that step included back in) if loosing step 21 and not bankrupt reset the ladder and start over on stage 3 . Reason being that step 22 had a 31% chance of winning but the chances of winning the stages in the ladder 10 times is 68% . you have a better chance of getting the money back doing the ladder.

People keep misunderstanding what I actually built, so here is the truth in plain English: I am not a gambler — I am a programmer. I did not sit there playing 10,000 roulette games like some guy with a lucky rabbit foot. I wrote a program that runs the entire Wolfie Ladder — all 3 stages, all 21 steps — across 50 to 300 simultaneous games on a single page, and then I repeated that whole batch 10,000 times. That is how you get a real statistical picture. That is how you reach a ~90% win rate. Not vibes. Not superstition. Not “progressions” like Fibonacci that blow up the moment variance sneezes. The Wolfie Ladder is engineered to survive streaks, recover losses, and beat the distribution curve — and I tested it by running 10,000+ full sequences in under 30 minutes, not by hand playing one spin at a time. If people think I sat there clicking red/black like a zombie, they really do not understand who they are talking to. When i first made this system luck i guess was on my side because the first few hundred times i was playing it i was winning at a rate of 91% of the time but as i kept running it more and more times the win rate started to drop and now it is at ~80% . And here’s the part people really don’t understand: the Wolfie Ladder wins even when the game‑level win rate is losing. For example — in one of the sequences on the active tab, the system played 92 total games. Out of those:
38 wins
54 losses
Including a brutal streak where it lost every single bet from #52 through #67
That’s 16 consecutive losses in the middle of the run.
And despite that? Despite losing more games than it won? Despite eating a variance spike that would vaporize Fibonacci, Martingale, or any “pro gambler” strategy?
It still finished the sequence up +$100.
That’s the point. The Ladder doesn’t care about “win more than you lose.” It cares about sequence structure, distribution recovery, and controlled escalation. It’s engineered to survive streaks, absorb damage, and still exit the sequence profitable — even when the raw win/loss ratio looks terrible.
This is why the system works. This is why the win rate stabilizes around ~78.2% across 10,000 full trials. Not because it wins every spin — but because it wins the sequence.
THE WOLFIE LADDER SYSTEM (Step-by-Step) CLICK HERE IF YOU WANT TO KNOW THE REASONING BEHIND THESE STEPS
STAGE 1:
1️⃣ Bet on two dozens, $3 each (total $6). If win → net +$3, reset ladder, back to Step 1. If lose → Step 2.
2️⃣ Bet on two dozens, $9 each (total $18). Win → net +$3 ($9 minus the $6 lost on Step 1), reset. Lose → Step 3.
3️⃣ Bet $14 on the first dozen. Win → net +$4 ($28 minus the $24 already lost), reset. Lose → Step 4.
4️⃣ Bet $21 on the first dozen. Win → net +$4 ($42 minus the $38 already lost), reset. Lose → Step 5.
5️⃣ Bet $28 on the first dozen. Win → net −$3 ($56 minus the $59 already lost), reset. Lose → Step 6.
6️⃣ Bet $42 on the first dozen. Win → net −$3 ($84 minus the $87 already lost), reset. Lose → set bank_needed_to_reset = current bank + 90, then Step 8.
7️⃣(Removed in Version B. It only increased Stage‑1 probability by ~2%, but added a heavy 45‑unit cost. Cutting it keeps the system simpler, faster, and far less volatile. We still have a ~5% chance of hitting this loss, but the ladder easily recovers it.)
STAGE 2:
8️⃣ Bet $10 on two dozens. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → repeat Step 8. Lose → Step 9. ( net win at this step is $10 )
9️⃣ Bet $16 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 8. Lose → Step 10. ( net win at this step is $12 )
🔟 Bet $24 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 8. Lose → Step 11. ( net win at this step is $12 )
1️⃣1️⃣ Bet $36 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 8. Lose → Step 12. ( net win at this step is $12 )
1️⃣2️⃣ Bet $54 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 8. Lose → Step 13. ( net win at this step is $12 )
1️⃣3️⃣ Bet $81 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 8. Lose → set bank_needed_to_reset = current bank + 300, then Step 15. ( net win at this step is $12 )
1️⃣4️⃣ (Removed in Version C. It only increased Stage‑2 probability by ~2%, but added a heavy 122‑unit cost.
STAGE 3:
1️⃣5️⃣ Bet $25 on two dozens. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → repeat Step 15. Lose → Step 16.
1️⃣6️⃣ Bet $40 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 15. Lose → Step 17.
1️⃣7️⃣ Bet $60 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 15. Lose → Step 18.
1️⃣8️⃣ Bet $90 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 15. Lose → Step 19.
1️⃣9️⃣ Bet $130 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 15. Lose → Step 20.
2️⃣0️⃣ Bet $190 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 15. Lose → Step 21.
2️⃣1️⃣ Bet $250 on the first dozen. Win & bank > bank_needed_to_reset → reset. Win but bank ≤ bank_needed_to_reset → Step 15. Lose → Step 15.

Roulette wheel
This system was build using 80 diffent ai agents across 8 LLM models to test the best possible system for roulette using lupopedia AI platform.
The programmer captain wolfie (Eric) was too board playing just one game at a time so programmed this to play up to 500 games all at once, that way can play thousands of times a minute and find the best stratagy that holds up . .
Want the full story? Check out the Patreon post for the complete dialog between Captain Wolfie, Grok, and the mischievous Lilith!

Why my versions are A, B, C instead of major.minor.patch Because the traditional versioning system — major.minor.patch — is meaningless for what I’m doing here. I’m not shipping software releases. I’m iterating behavioral variants of a probability engine. Each version (A, B, C… all the way to ZZ if I want) represents a distinct ladder logic, a distinct recovery pattern, a distinct volatility profile, and a distinct mathematical personality. These aren’t “patches.” These aren’t “minor updates.” These are forks of logic. Version A behaves one way. Version B behaves another way. Version C might remove a step, add a step, rebalance a stage, or change the volatility curve entirely. Using semantic versioning (4.0.96, 4.0.97, etc.) would imply these are incremental changes to the same system — but they’re not. They’re parallel universes. So I use lettered versions because: They’re simple They’re human-readable They reflect behavior, not “patch level” They let me branch logic infinitely (A → B → C → … → Z → AA → AB → … → ZZ) They match how I actually think about the system In other words: Version A is a creature. Version B is a different creature. Version C is another creature. Letters make more sense for creatures than numbers.

Probability Reality Check

American wheel dozen loss probability stays the same on every spin: about 68.42% to lose, 31.58% to hit your dozen. Even after 10 or 15 losses in a row, the next spin is still roughly 1/3 to win, 2/3 to lose. This ladder doesn't pretend the odds improve—it only adjusts stake size so that, when you do finally win, you aim to recover with the progression. The house edge is still there; this is about controlled exposure, not guaranteed profit.

500-Round Simulation (American Wheel Model)

MySQL connected to collab52_salessyntax on localhost. This run was saved to wolfie_ladder_runs.

Every time this page loads it simulates 500 games using the WOLFIE Ladder exactly as defined here or stops if you win $100+ How did you do? If the run ended in the cliff, click here to try again and generate a fresh 500-game sequence. Saved runs are listed on wolfie_ladder_results.php.

Final Bank
194.00
Net Units
-1,306.00
Wins / Losses
75 / 96
Max Step Reached
21
Run Status
Bank Depleted or Game Over
Table Limit Hit
No
Bank Depleted
Yes

Below is one 500-bet run. Each roll uses the real American wheel odds (12/38 for single dozen). Every time a win hits, we reset or loop based on rules.

# Bet Wheel Result Bank Next Step Mode
1 3 each on two dozens 16 Win 1,503.00 1 Step 1
2 3 each on two dozens 13 Win 1,506.00 1 Step 1
3 3 each on two dozens 33 Loss 1,500.00 2 Step 1
4 9 each on two dozens 0 Loss 1,482.00 3 Step 2
5 14 on first dozen 27 Loss 1,468.00 4 Step 3
6 21 on first dozen 6 Win 1,510.00 1 Step 4
7 3 each on two dozens 35 Loss 1,504.00 2 Step 1
8 9 each on two dozens 34 Loss 1,486.00 3 Step 2
9 14 on first dozen 30 Loss 1,472.00 4 Step 3
10 21 on first dozen 29 Loss 1,451.00 5 Step 4
11 28 on first dozen 20 Loss 1,423.00 6 Step 5
12 42 on first dozen 10 Win 1,507.00 1 Step 6
13 3 each on two dozens 23 Win 1,510.00 1 Step 1
14 3 each on two dozens 9 Win 1,513.00 1 Step 1
15 3 each on two dozens 6 Win 1,516.00 1 Step 1
16 3 each on two dozens 2 Win 1,519.00 1 Step 1
17 3 each on two dozens 29 Loss 1,513.00 2 Step 1
18 9 each on two dozens 1 Win 1,522.00 1 Step 2
19 3 each on two dozens 31 Loss 1,516.00 2 Step 1
20 9 each on two dozens 31 Loss 1,498.00 3 Step 2
21 14 on first dozen 1 Win 1,526.00 1 Step 3
22 3 each on two dozens 3 Win 1,529.00 1 Step 1
23 3 each on two dozens 35 Loss 1,523.00 2 Step 1
24 9 each on two dozens 10 Win 1,532.00 1 Step 2
25 3 each on two dozens 1 Win 1,535.00 1 Step 1
26 3 each on two dozens 34 Loss 1,529.00 2 Step 1
27 9 each on two dozens 6 Win 1,538.00 1 Step 2
28 3 each on two dozens 21 Win 1,541.00 1 Step 1
29 3 each on two dozens 21 Win 1,544.00 1 Step 1
30 3 each on two dozens 20 Win 1,547.00 1 Step 1
31 3 each on two dozens 5 Win 1,550.00 1 Step 1
32 3 each on two dozens 25 Loss 1,544.00 2 Step 1
33 9 each on two dozens 0 Loss 1,526.00 3 Step 2
34 14 on first dozen 5 Win 1,554.00 1 Step 3
35 3 each on two dozens 9 Win 1,557.00 1 Step 1
36 3 each on two dozens 17 Win 1,560.00 1 Step 1
37 3 each on two dozens 25 Loss 1,554.00 2 Step 1
38 9 each on two dozens 35 Loss 1,536.00 3 Step 2
39 14 on first dozen 8 Win 1,564.00 1 Step 3
40 3 each on two dozens 10 Win 1,567.00 1 Step 1
41 3 each on two dozens 28 Loss 1,561.00 2 Step 1
42 9 each on two dozens 21 Win 1,570.00 1 Step 2
43 3 each on two dozens 25 Loss 1,564.00 2 Step 1
44 9 each on two dozens 33 Loss 1,546.00 3 Step 2
45 14 on first dozen 18 Loss 1,532.00 4 Step 3
46 21 on first dozen 0 Loss 1,511.00 5 Step 4
47 28 on first dozen 14 Loss 1,483.00 6 Step 5
48 42 on first dozen 19 Loss 1,441.00 8 Step 6
49 10 each on two dozens 33 Loss 1,421.00 9 Step 8
50 16 on first dozen 9 Win 1,453.00 8 Step 9
51 10 each on two dozens 23 Win 1,463.00 8 Step 8
52 10 each on two dozens 33 Loss 1,443.00 9 Step 8
53 16 on first dozen 36 Loss 1,427.00 10 Step 9
54 24 on first dozen 19 Loss 1,403.00 11 Step 10
55 36 on first dozen 16 Loss 1,367.00 12 Step 11
56 54 on first dozen 27 Loss 1,313.00 13 Step 12
57 81 on first dozen 4 Win 1,475.00 8 Step 13
58 10 each on two dozens 34 Loss 1,455.00 9 Step 8
59 16 on first dozen 29 Loss 1,439.00 10 Step 9
60 24 on first dozen 21 Loss 1,415.00 11 Step 10
61 36 on first dozen 34 Loss 1,379.00 12 Step 11
62 54 on first dozen 30 Loss 1,325.00 13 Step 12
63 81 on first dozen 14 Loss 1,244.00 15 Step 13
64 25 each on two dozens 1 Win 1,269.00 15 Step 15
65 25 each on two dozens 31 Loss 1,219.00 16 Step 15
66 40 on first dozen 25 Loss 1,179.00 17 Step 16
67 60 on first dozen 22 Loss 1,119.00 18 Step 17
68 90 on first dozen 20 Loss 1,029.00 19 Step 18
69 130 on first dozen 23 Loss 899.00 20 Step 19
70 190 on first dozen 00 Loss 709.00 21 Step 20
71 250 on first dozen 15 Loss 459.00 15 Step 21
72 25 each on two dozens 34 Loss 409.00 16 Step 15
73 40 on first dozen 13 Loss 369.00 17 Step 16
74 60 on first dozen 10 Win 489.00 15 Step 17
75 25 each on two dozens 21 Win 514.00 15 Step 15
76 25 each on two dozens 1 Win 539.00 15 Step 15
77 25 each on two dozens 18 Win 564.00 15 Step 15
78 25 each on two dozens 26 Loss 514.00 16 Step 15
79 40 on first dozen 13 Loss 474.00 17 Step 16
80 60 on first dozen 16 Loss 414.00 18 Step 17
81 90 on first dozen 21 Loss 324.00 19 Step 18
82 130 on first dozen 12 Win 584.00 15 Step 19
83 25 each on two dozens 13 Win 609.00 15 Step 15
84 25 each on two dozens 28 Loss 559.00 16 Step 15
85 40 on first dozen 8 Win 639.00 15 Step 16
86 25 each on two dozens 26 Loss 589.00 16 Step 15
87 40 on first dozen 22 Loss 549.00 17 Step 16
88 60 on first dozen 34 Loss 489.00 18 Step 17
89 90 on first dozen 10 Win 669.00 15 Step 18
90 25 each on two dozens 22 Win 694.00 15 Step 15
91 25 each on two dozens 2 Win 719.00 15 Step 15
92 25 each on two dozens 5 Win 744.00 15 Step 15
93 25 each on two dozens 17 Win 769.00 15 Step 15
94 25 each on two dozens 23 Win 794.00 15 Step 15
95 25 each on two dozens 3 Win 819.00 15 Step 15
96 25 each on two dozens 34 Loss 769.00 16 Step 15
97 40 on first dozen 23 Loss 729.00 17 Step 16
98 60 on first dozen 35 Loss 669.00 18 Step 17
99 90 on first dozen 5 Win 849.00 15 Step 18
100 25 each on two dozens 6 Win 874.00 15 Step 15
101 25 each on two dozens 24 Win 899.00 15 Step 15
102 25 each on two dozens 12 Win 924.00 15 Step 15
103 25 each on two dozens 3 Win 949.00 15 Step 15
104 25 each on two dozens 0 Loss 899.00 16 Step 15
105 40 on first dozen 27 Loss 859.00 17 Step 16
106 60 on first dozen 5 Win 979.00 15 Step 17
107 25 each on two dozens 7 Win 1,004.00 15 Step 15
108 25 each on two dozens 19 Win 1,029.00 15 Step 15
109 25 each on two dozens 23 Win 1,054.00 15 Step 15
110 25 each on two dozens 23 Win 1,079.00 15 Step 15
111 25 each on two dozens 26 Loss 1,029.00 16 Step 15
112 40 on first dozen 9 Win 1,109.00 15 Step 16
113 25 each on two dozens 24 Win 1,134.00 15 Step 15
114 25 each on two dozens 9 Win 1,159.00 15 Step 15
115 25 each on two dozens 22 Win 1,184.00 15 Step 15
116 25 each on two dozens 00 Loss 1,134.00 16 Step 15
117 40 on first dozen 28 Loss 1,094.00 17 Step 16
118 60 on first dozen 21 Loss 1,034.00 18 Step 17
119 90 on first dozen 8 Win 1,214.00 15 Step 18
120 25 each on two dozens 26 Loss 1,164.00 16 Step 15
121 40 on first dozen 29 Loss 1,124.00 17 Step 16
122 60 on first dozen 21 Loss 1,064.00 18 Step 17
123 90 on first dozen 31 Loss 974.00 19 Step 18
124 130 on first dozen 6 Win 1,234.00 15 Step 19
125 25 each on two dozens 16 Win 1,259.00 15 Step 15
126 25 each on two dozens 6 Win 1,284.00 15 Step 15
127 25 each on two dozens 24 Win 1,309.00 15 Step 15
128 25 each on two dozens 19 Win 1,334.00 15 Step 15
129 25 each on two dozens 33 Loss 1,284.00 16 Step 15
130 40 on first dozen 36 Loss 1,244.00 17 Step 16
131 60 on first dozen 25 Loss 1,184.00 18 Step 17
132 90 on first dozen 16 Loss 1,094.00 19 Step 18
133 130 on first dozen 0 Loss 964.00 20 Step 19
134 190 on first dozen 21 Loss 774.00 21 Step 20
135 250 on first dozen 10 Win 1,274.00 15 Step 21
136 25 each on two dozens 14 Win 1,299.00 15 Step 15
137 25 each on two dozens 33 Loss 1,249.00 16 Step 15
138 40 on first dozen 29 Loss 1,209.00 17 Step 16
139 60 on first dozen 23 Loss 1,149.00 18 Step 17
140 90 on first dozen 28 Loss 1,059.00 19 Step 18
141 130 on first dozen 25 Loss 929.00 20 Step 19
142 190 on first dozen 15 Loss 739.00 21 Step 20
143 250 on first dozen 13 Loss 489.00 15 Step 21
144 25 each on two dozens 4 Win 514.00 15 Step 15
145 25 each on two dozens 0 Loss 464.00 16 Step 15
146 40 on first dozen 23 Loss 424.00 17 Step 16
147 60 on first dozen 33 Loss 364.00 18 Step 17
148 90 on first dozen 20 Loss 274.00 19 Step 18
149 130 on first dozen 9 Win 534.00 15 Step 19
150 25 each on two dozens 13 Win 559.00 15 Step 15
151 25 each on two dozens 27 Loss 509.00 16 Step 15
152 40 on first dozen 11 Win 589.00 15 Step 16
153 25 each on two dozens 26 Loss 539.00 16 Step 15
154 40 on first dozen 5 Win 619.00 15 Step 16
155 25 each on two dozens 23 Win 644.00 15 Step 15
156 25 each on two dozens 18 Win 669.00 15 Step 15
157 25 each on two dozens 6 Win 694.00 15 Step 15
158 25 each on two dozens 23 Win 719.00 15 Step 15
159 25 each on two dozens 26 Loss 669.00 16 Step 15
160 40 on first dozen 19 Loss 629.00 17 Step 16
161 60 on first dozen 00 Loss 569.00 18 Step 17
162 90 on first dozen 13 Loss 479.00 19 Step 18
163 130 on first dozen 20 Loss 349.00 20 Step 19
164 190 on first dozen 8 Win 729.00 15 Step 20
165 25 each on two dozens 1 Win 754.00 15 Step 15
166 25 each on two dozens 34 Loss 704.00 16 Step 15
167 40 on first dozen 17 Loss 664.00 17 Step 16
168 60 on first dozen 0 Loss 604.00 18 Step 17
169 90 on first dozen 23 Loss 514.00 19 Step 18
170 130 on first dozen 20 Loss 384.00 20 Step 19
171 190 on first dozen 33 Loss 194.00 21 Step 20

All the Math and Lilith and Wolfie Arguing About the System (Updated Canon)

American roulette: 38 numbers. A two-dozen bet covers 24 numbers and loses on 14: about 36.84% to lose, 63.16% to hit. A one-dozen bet covers 12 numbers and loses on 26: about 68.42% to lose, 31.58% to hit. Version G Stage 1 is two two-dozen bets, then four one-dozen bets. Stage 2 is one two-dozen bet, then five one-dozen bets. Stage 3 is one two-dozen bet, then six one-dozen bets. A loss does not change the odds of the next spin.

Stage 1 (steps 1–6)

Losing the whole stage means both two-dozen bets miss and then all four one-dozen bets miss:

P(Stage 1 bust) = (14/38)2 × (26/38)4 ≈ 2.97%

P(Stage 1 wins somewhere in the six steps) = 1 − 0.0297 ≈ 97.03%

Step 1

Bet: two dozens, $3 each (total $6). 14 losing numbers.

P(lose Step 1) = 14/38 ≈ 36.84%

Win → net +$3, reset to Step 1. This is where about 63.16% of Stage 1 sequences end. Lose → Step 2.

Step 2

Bet: two dozens, $9 each (total $18). Same 14 losing numbers.

P(lose Steps 1 and 2) = (14/38)2 ≈ 13.57%

Win → net +$3 ($9 minus the $6 already lost), reset. About 23.27% of sequences win here. Lose → Step 3.

Step 3

Bet: $14 on the first dozen. 26 losing numbers.

P(lose Steps 1–3) = (14/38)2 × (26/38) ≈ 9.29%

Win → net +$4 ($28 minus the $24 already lost), reset. About 4.29% of sequences win here. Lose → Step 4.

Step 4

Bet: $21 on the first dozen. Same 26 losing numbers.

P(lose Steps 1–4) = (14/38)2 × (26/38)2 ≈ 6.35%

Win → net +$4 ($42 minus the $38 already lost), reset. About 2.93% of sequences win here. Lose → Step 5.

Step 5

Bet: $28 on the first dozen.

P(lose Steps 1–5) = (14/38)2 × (26/38)3 ≈ 4.35%

Win → net −$3 ($56 minus the $59 already lost), reset. About 2.01% of sequences win here. The spin pays, and the sequence is still down $3. Lose → Step 6.

Step 6

Bet: $42 on the first dozen.

P(lose Steps 1–6) = (14/38)2 × (26/38)4 ≈ 2.97%

Win → net −$3 ($84 minus the $87 already lost), reset. About 1.37% of sequences win here. Lose → the six bets have cost $129 (6+18+14+21+28+42). Set bank_needed_to_reset = current bank + 90, then Step 8.

Stage 2 (steps 8–13)

One two-dozen bet ($10 each), then five one-dozen bets ($16, $24, $36, $54, $81). Losing the whole pass:

P(Stage 2 pass bust) = (14/38) × (26/38)5 ≈ 5.52%

P(Stage 2 wins somewhere in the pass) ≈ 94.48%

Chance of still being in the pass after each loss:

A win on Steps 8–13 returns to Step 1 only when the bank is above bank_needed_to_reset. Otherwise the ladder repeats Step 8. A loss on Step 13 sets bank_needed_to_reset = current bank + 300 and moves to Step 15.

Stage 3 (steps 15–21)

One two-dozen bet ($25 each), then six one-dozen bets ($40, $60, $90, $130, $190, $250). Losing the whole pass:

P(Stage 3 pass bust) = (14/38) × (26/38)6 ≈ 3.78%

P(Stage 3 wins somewhere in the pass) ≈ 96.22%

That 3.78% / 96.22% pair is the old seven-step ladder (one two-dozen bet and six one-dozen bets). On this version it belongs to Stage 3 only. Stage 1 is the 2.97% bust above.

Chance of still being in the pass after each loss:

A win on Steps 15–21 returns to Step 1 only when the bank is above bank_needed_to_reset. Otherwise the ladder repeats Step 15. A loss on Step 21 returns to Step 15. It does not end the game.