1-3-2-6 Betting System on Stake: Sequence Math, EV and Automation (2026)

The 1-3-2-6 betting system is a four-step positive progression that aims to lock in a guaranteed profit after a short winning streak while keeping downside small on losing rounds. Compared to aggressive negative progressions like Martingale, the 1-3-2-6 caps your exposure at the size of the first unit on every reset, which makes it easier to model, easier to automate, and easier to walk away from with the bankroll intact. This guide breaks down how the sequence works, the math on Stake's even-money games, where it fits in a 2026 betting workflow, and how to wire it into a bot.

What Is the 1-3-2-6 System?

1-3-2-6 is a fixed four-step staking pattern used on near-even-money outcomes. The numbers represent units, not absolute amounts: if your unit is 0.0001 BTC, then 1-3-2-6 translates to 0.0001, 0.0003, 0.0002, 0.0006. You only advance to the next step after a win. Any loss resets you to step one. After completing all four steps with consecutive wins, you bank the profit and start over.

The system was originally promoted for baccarat and even-money roulette bets, but the same mechanics map cleanly onto Stake originals like Limbo at a 2x target, Dice at 50% win chance, and Wheel on the lowest-risk preset. It is a discipline tool, not an edge generator: the house edge does not change.

The Sequence Logic Step by Step

Assume a unit of 1 and an even-money payout. The four steps play out as follows:

• Step 1 — Bet 1. If you lose, repeat step 1 with -1 to bankroll.
• Step 2 — Bet 3. After a step 1 win you have +1; staking 3 risks only 2 of original bankroll. A loss here ends the cycle at -2.
• Step 3 — Bet 2. After two wins you are at +4. Staking 2 guarantees +2 minimum even if step 3 loses.
• Step 4 — Bet 6. After three wins you are at +6. A loss here exits at +0, a win exits at +12.

Why 1, 3, 2, 6?

The drop from 3 to 2 in step three is the trick that locks in profit. By under-betting after the second win, the sequence turns the third round into a free-roll: even a loss leaves you net positive. Step four then leverages that cushion to chase the +12 outcome without risking original capital.

When the Sequence Resets

The system always resets to step 1 after any loss. There is no chasing, no doubling, no recovery row. That single rule keeps drawdown bounded and makes 1-3-2-6 attractive for a bot: the state machine has only four states plus a reset condition.

Math and Expected Value

Even-Money Bets and House Edge

On Stake Dice at 50% win chance the multiplier is 1.9802x, meaning the effective payout is slightly below 1:1. The 1% house edge is unchanged regardless of staking pattern. Across an infinite number of cycles, expected value per unit wagered remains -0.01. Progression systems redistribute variance; they do not bend probability.

Limbo at 2x has a 49.5% win chance and the same 1% edge. Baccarat banker sits near 1.06% house edge after commission, and European roulette red/black is 2.7%. The 1-3-2-6 is structurally identical across all of these — only the underlying probability shifts how often you complete the sequence.

Win Probability Across Four Steps

At a 49.5% per-bet win probability, completing all four steps has probability 0.495^4 ≈ 6.0%. The remaining ~94% of sequences end with at most a -1, -2, or +2 outcome. The full per-cycle distribution on a true 50/50 with 1.98x payout is roughly:

• All four wins (≈6.0%): +12 units locked.
• Win-Win-Win-Loss (≈6.1%): 0 units net.
• Win-Win-Loss (≈12.4%): +2 units locked.
• Win-Loss (≈25.0%): -2 units.
• Loss on step 1 (≈50.5%): -1 unit.

Multiply each outcome by its probability and the cycle EV converges to roughly -0.04 units per cycle on a true 50/50 with no edge — and slightly worse once Stake's 1% edge is applied. The system is mathematically a small loser in the long run, like every other progression. What it buys you is a smoother equity curve than Martingale at the cost of giving up rare jackpot wins.

Where 1-3-2-6 Fits on Stake

The system needs near-even-money outcomes and clean win/loss states. That maps to a specific subset of Stake games:

• Limbo at exactly 2.00x target multiplier.
• Dice with the slider set to 50% win chance.
• Wheel on the 1.5x risk profile (1.5x slot = win, 0x slot = loss).
• Baccarat banker or player bets, ignoring tie pushes.
• European roulette red/black, odd/even or high/low.

Slots, Mines, Plinko and Crash above 2x are poor fits because the win/loss distribution is not binary or the payout is not even-money. Forcing 1-3-2-6 onto multi-state games breaks the sequence math.

Bankroll Sizing and Risk

Maximum exposure during a single cycle is 6 units on step four, but the bankroll only ever risks 2 original units before locking profit, and the worst-case losing run still draws down at -1 per cycle. A safe rule for casual play is to set a unit equal to 0.5% of session bankroll: that gives you up to 200 lost step-1 cycles before exhausting the roll. For bot operation, set unit size relative to total bankroll, not session bankroll, and run a stop-loss anchored to a fixed percentage drop.

Automating 1-3-2-6 with SSPilot

1-3-2-6 is a strong candidate for full automation because the state machine is trivial. A bot only needs to track four states, a unit value, and global stop-loss/take-profit. Inside SSPilot you can express the system using on-win and on-loss conditions with bet multipliers fixed to the four sequence values.

• State 1 → bet 1 unit. On win, advance to state 2. On loss, stay.
• State 2 → bet 3 units. On win, advance to state 3. On loss, return to state 1.
• State 3 → bet 2 units. On win, advance to state 4. On loss, return to state 1.
• State 4 → bet 6 units. On any result, return to state 1.
• Global guard rails: stop-loss at -X% bankroll, take-profit at +Y% bankroll, max session duration.

Logging every cycle is more valuable than the strategy itself. Over a few thousand cycles, SSPilot's session logs show whether realized variance matches the theoretical distribution, which is how you catch a misconfigured unit, an unexpected target multiplier, or drift caused by network re-bets.

Limits and Common Mistakes

• Treating 1-3-2-6 as a positive-EV system. It is not. The house edge is unchanged.
• Running it on non-binary games (Plinko, Crash above 2x, Mines with multiple tiles) where the math no longer holds.
• Skipping the step-3 under-bet and pushing straight to 6 units after two wins, which breaks the lock-in property.
• Increasing unit size mid-session after a losing streak. That defeats the bounded-drawdown property entirely.
• Ignoring stop-loss because individual cycles feel small. Volume kills bankrolls slowly.

Conclusion

1-3-2-6 is not a way to beat Stake — no betting system is. It is a structured way to size bets on even-money outcomes with bounded downside per cycle and a clean automation profile. Used with a fixed unit, a hard stop-loss, and a session log, it keeps play disciplined and variance readable. The house edge still applies; treat sessions as paid entertainment with a strategy layer, not as an income strategy, and play within limits you can afford to lose.