Regression is the mirror image of pressing. Instead of adding winnings back to your bets after a hit, you start with a larger bet, collect a profit on the first win, then drop down to a smaller bet for the rest of the roll. The goal is to guarantee yourself a net profit before the 7 can wipe you out. It's not a mathematical advantage — no bet sequence changes the house edge — but it's a legitimate risk-management approach that suits players who'd rather lock in a small profit than chase a big one and risk walking away with nothing.
How Regression Works
The basic pattern is: bet big, collect a profit on the first hit, regress to a smaller bet, continue from a position of guaranteed profit. The smaller bet stays active and keeps paying if the roll continues — but even if the 7 shows up immediately after the regression, you're already ahead.
A common example on the 6 and 8: start with $30 on the 6 and $30 on the 8. Your total exposure is $60. The 6 hits and pays $35 (place bets on the 6 and 8 pay 7:6). Now tell the dealer "regress both to six dollars." Your bets drop to $6 on the 6 and $6 on the 8 — $12 total at risk. You've already collected $35. If the 7 comes on the very next roll, you lose $12 but keep the $35 — a $23 profit regardless of how the rest of the roll plays out.
If the roll continues, those $6 bets keep hitting and each win pays $7. Every additional hit adds to profit that's already locked in.
The Math of a Single Regression
Breaking down the $30/$30 example precisely:
- Opening position: $30 on 6, $30 on 8 — $60 total at risk
- First hit (say the 6 at 7:6): pays $35
- Regress both to $6: $6 on 6, $6 on 8 — $12 total at risk
- Net: $35 collected, $12 still at risk — guaranteed profit of $23 if 7 comes now
- Each subsequent hit on the $6 bets pays $7 — all of it on top of that locked-in $23
Compare that to flat betting $6 on both numbers from the start. Your first hit pays $7 and you still have $12 at risk, so you're only $7 ahead — and a 7-out on the next roll costs you $12, leaving you $5 in the hole. With regression, that same 7-out after the first hit leaves you $23 ahead. That's the core benefit.
The catch is what happens before the first hit. With flat $6/$6, a 7-out on the first roll costs $12. With $30/$30, the same outcome costs $60. Regression assumes the first hit will come — if it doesn't, the larger opening bet costs you significantly more than the conservative flat bet would have.
The Tradeoff — What You Give Up
Regression is not a mathematical improvement. The house edge on a place bet on the 6 or 8 is 1.52% whether you have $6 or $30 on it. What changes is the variance profile — how wins and losses are distributed relative to each other.
With regression you're making a specific trade: you accept higher initial exposure (and a bigger loss if the 7 shows up before any hit) in exchange for a guaranteed profit once that first hit lands. You're not reducing the house's mathematical advantage; you're shifting when the pain and reward land relative to each other.
On a session where shooters frequently 7-out on their first or second roll, regression will cost you more than flat betting would. You'd keep absorbing $60 losses instead of $12 losses. On a session where most shooters hit at least one number before sevening out — which is the more common outcome, given the dice probabilities — regression does what it's supposed to: you walk away from each shooter with a small guaranteed profit rather than depending on the roll continuing.
Common Regression Amounts
You can apply regression at different scales depending on your bankroll and risk tolerance. The principle is the same at any size; the numbers just change.
$30 down to $6 on the 6 or 8: the most common version. The first hit ($35) covers both bets at their regressed size ($12) and profits $23. This is the standard "thirty to six" regression you'll hear called at most tables.
$18 down to $6: a gentler version. Opening exposure is $36 for two numbers. First hit ($21) still covers the regressed bets ($12) and locks in $9 profit. Lower upfront risk, smaller guaranteed profit.
$12 down to $6: minimal regression. First hit ($14) after regressing leaves you $2 guaranteed profit. The upside is limited, but so is the additional exposure compared to starting flat.
One constraint: bets on the 6 and 8 must be multiples of $6 for the 7:6 payout to come out clean. Bets on the 5 and 9 must be multiples of $5 for the 7:5 payout. Any amount that doesn't fit the payout ratio gets rounded down by the house, which works against you. Stick to the right increments for whichever numbers you're on.
Double Regression
Some players extend the approach by regressing twice instead of once — stepping down after the first hit and again after the second.
Example: start with $30 on 6 and $30 on 8. First hit, regress both to $18. Second hit, regress both to $6. At each step you lock in more profit while reducing your exposure further. By the time you're at $6/$6, you've collected enough on the first two hits that almost any 7-out from there still leaves you ahead by a meaningful amount.
The tradeoff is complexity. You need to know your numbers in advance and call the regression clearly and quickly — after a hit, the dealer is moving fast and the next roll can come before you've spoken. Decide your regression plan before the shooter picks up the dice, not after the first hit. If you want to try double regression, practice the math and the phrasing at home before calling it at a live table under time pressure.
Regression on Multiple Numbers
Regression is most naturally applied to the 6 and 8 because they have the lowest house edge among place bets and offer the best starting point for the math to work. But the same logic applies to any place bet or combination of place bets.
You can also apply regression to a full spread across all six numbers. Open with $25 across — meaning a pass-line point covered by your pass bet plus $5 each on the remaining five numbers — then regress to $5 across after the first hit. The broader spread gives you more ways to hit before the 7, which is the condition the strategy depends on.
When you're covering more numbers, verify the math before you sit down: the total you collect on the first hit needs to exceed the total you'll have at risk after the regression for the guaranteed-profit logic to hold. Adding numbers increases both sides of that equation — more ways to hit, but also more at risk if you don't.
When to Use Regression
Regression fits specific situations. It's not a system to run blindly on every shooter — it's a tool for particular conditions.
- Limited session bankroll: if you're working with a modest buy-in and can't comfortably absorb several consecutive 7-outs, regression limits how often you leave a shooter empty-handed.
- Choppy table: if the session has been grinding through short rolls — most shooters hitting one or two numbers and then sevening out — regression suits that pattern. You collect something on each shooter instead of waiting for a roll that never develops.
- Protecting a winning session: if you're ahead and want to stay in action without risking your gains, regressing keeps your bets live while capping downside on each new shooter.
- After a losing streak: regression won't fix the math, but it can slow the bleeding and give you some hits that feel like progress rather than break-even runs.
When Not to Use Regression
There are two situations where regression works against you.
The first is a genuinely hot roll. If a shooter goes on to roll 15 numbers before sevening out, you've spent most of that roll collecting $7 hits on $6 bets. Had you left the $30 bets up or pressed into them, the same roll would have paid multiples more. Regression costs you upside during the rolls it's least needed — the long ones. That's not a reason to never use it; it's a reason to be clear-eyed about what you're trading away.
The second is when the opening bet size is outside your comfortable risk range. If losing $60 before the first hit would meaningfully damage your session — either financially or psychologically — then the opening bet is too large. In that case, start smaller and don't regress at all. Flat betting $6/$6 is a perfectly sound approach and doesn't require the same nerve as opening with $30/$30. The strategy only works if you can execute it without second-guessing the opening position every time the dice go out.