Betting progressions change how much you wager after wins or losses. They do not change the probabilities of the next craps roll, and they do not remove the house edge from the underlying bet.
That makes progression systems useful to compare as risk-shaping methods, not as ways to beat the game. Two systems can start with the same $10 wager, use the same Pass Line or Don’t Pass bet, and still create very different bankroll swings because one escalates after losses while another escalates only after wins.
The base wager decides the mathematical price
Suppose a bet has house edge h and you wager amount B.
The expected loss on that wager is:
Expected loss = B × h
If the next bet is larger because a progression told you to raise the stake, the probability structure of the base wager has not improved. You have simply multiplied the same edge by a larger amount.
For example, if a hypothetical bet carries a 1.4% house edge:
- $10 action has expected loss of $0.14;
- $40 action has expected loss of $0.56;
- $160 action has expected loss of $2.24.
The progression changes the dollar exposure. It does not make the $160 wager mathematically stronger than the $10 wager.
This is the central test for every betting system: Where, exactly, does the probability or payout improve? If the answer is “nowhere,” the progression cannot create positive expectation by rearranging stake sizes alone.
Flat betting is the control case
Flat betting means using the same stake repeatedly.
If the player makes ten $10 wagers, total action is $100. The bankroll can still swing, but the player knows that each decision creates the same dollar exposure.
Flat betting is not a winning system. It is simply the cleanest baseline for comparing risk because stake size does not react to the previous result.
Advantages:
- easy to track;
- easy to budget;
- no automatic escalation after losses;
- total action is relatively predictable.
Disadvantage:
- it does nothing to recover prior losses faster;
- it produces no dramatic “press” effect during a winning run.
Those are emotional disadvantages, not mathematical ones.
Martingale concentrates risk after losses
The classic Martingale doubles the next wager after each loss and resets after a win.
Starting at $10, the sequence is:
$10 → $20 → $40 → $80 → $160 → ...
The attraction is obvious. If each win pays even money, one eventual win appears to recover all previous losses plus the original $10 target profit.
But the required stake grows exponentially.
After five consecutive losses, the player has already lost:
$10 + $20 + $40 + $80 + $160 = $310
The next required wager is $320.
After six losses, cumulative loss is $630 and the next bet is $640.
The system therefore converts a series of ordinary small losses into a rare but very large bankroll event.
It does not matter that long losing sequences are less common than single losses. They only need to occur occasionally because the damage rises so quickly when they do.
Table limits create a second hard stop. Even a very large bankroll cannot continue doubling indefinitely if the next required wager exceeds the table maximum.
Loss progressions do not make a bet “due”
A Martingale often feels persuasive because the player thinks, “After four losses, a win must be getting closer.”
That is the gambler’s fallacy.
For independent or conditionally unchanged wagers, the previous losing sequence does not improve the price of the next bet. In craps, the exact probability can depend on which wager is active and whether a point is established, but simply losing several bets does not create a compensating probability bonus on the next one.
The progression raises the stake precisely when the player is already under emotional pressure. That is one reason loss systems can be dangerous even before bankroll mathematics is considered.
Reverse Martingale presses wins instead
A Reverse Martingale, often called a Paroli-style approach, increases the wager after wins and resets after a loss.
One simple sequence might be:
$10 → $20 → $40 → reset
if the player wins each step.
Compared with Martingale, this has an important risk difference: the player is pressing money after favorable outcomes rather than chasing after losses.
That does not improve expected value. The larger wager still carries the same house edge. But the bankroll path can be less destructive because a losing streak tends to keep stakes near the base amount rather than force exponential escalation.
The tradeoff is that a win after a press can return much of the accumulated gain to the table.
1-3-2-6 is a structured positive progression
The 1-3-2-6 system uses a four-step unit sequence after consecutive wins:
1 unit → 3 units → 2 units → 6 units
A loss generally resets the sequence.
With a $10 unit, the stakes are:
$10 → $30 → $20 → $60
The system is designed to use earlier wins to finance later presses. If all four bets win on an even-money wager, the cycle produces a sizable gain relative to the starting unit.
The weak point is the same as every positive progression: the player still gives larger action to the same underlying negative-expectation wager. A loss on a later step can erase much of the cycle’s earlier profit.
The sequence may feel more controlled than open-ended doubling because it has a defined endpoint. That is a bankroll-management feature, not a mathematical edge.
Fibonacci and cancellation systems hide escalation behind arithmetic
Some progressions avoid doubling and instead use sequences such as Fibonacci numbers or cancellation lists.
A Fibonacci loss progression might move through unit sizes like:
1, 1, 2, 3, 5, 8, 13...
This grows more slowly than Martingale, which makes it feel safer.
But slower growth is still growth. A long losing run can produce large stakes and large cumulative losses.
Cancellation systems may begin with a list of numbers and set the next wager from the first and last remaining values. Wins remove numbers; losses add values or extend the sequence.
The bookkeeping creates an impression of control because every wager has a reason inside the system. But the reason is internal to the sequence. It does not alter the dice probabilities or payout odds.
Regression systems lower stakes after success
Craps players also use regression systems, especially on place bets.
A simple version might start with a relatively large amount across several numbers, collect one win, then reduce the remaining exposure.
This is different from a classic progression because the stake decreases after success rather than increasing.
Regression can be sensible as a risk-management choice if the player has already decided to make a larger opening wager. It can reduce future exposure after an early win.
But the opening wager still carried its original house edge. The system does not retroactively improve the price of that first bet.
Pressing and collecting change volatility, not edge
Craps table language often uses “press,” “take,” “same bet,” and “collect.” Those choices create a flexible progression without a formal named system.
A player might:
- collect every win;
- press every win;
- press half and collect half;
- press until a target amount is reached;
- regress after one or two hits.
These choices change how much money remains exposed to future rolls.
A full press creates more upside during a favorable run and more give-back risk if the next resolution loses. Collecting keeps the stake stable and moves winnings away from the active bet.
Again, neither choice changes the probability of the next dice total.
A comparison table shows the real differences
| Method | Stake changes after | Main risk shape | Open-ended escalation? | Mathematical edge created? |
|---|---|---|---|---|
| Flat betting | Neither wins nor losses | Stable exposure | No | No |
| Martingale | Losses | Rare very large loss | Yes | No |
| Reverse Martingale / Paroli | Wins | Gives back part of winning run | Usually limited by chosen sequence | No |
| 1-3-2-6 | Wins | Structured press cycle | No if rules are followed | No |
| Fibonacci | Losses | Slower but persistent escalation | Potentially | No |
| Cancellation | Wins/losses according to list | Complex variable exposure | Potentially | No |
| Regression | Usually after wins | High initial exposure, lower later risk | No | No |
The honest reason to choose among these systems is therefore not “which one beats craps?” It is “which exposure pattern matches the amount of volatility I am willing to accept?”
Expected loss depends on total action generated
Progressions can materially increase total action even when the session starts with the same base unit.
Suppose two players each start with a $10 unit on the same wager.
Player A flat bets $10 for 20 decisions:
Total action = $200
Player B uses a progression and, because of several presses and loss escalations, wagers a total of $520 over the same 20 decisions.
If the wager carries a 1.5% house edge, theoretical expected loss is:
Player A:
$200 × 0.015 = $3.00
Player B:
$520 × 0.015 = $7.80
The second player did not face a worse percentage. The system simply caused more money to be exposed to the same percentage.
This is why tracking total action is more useful than tracking only the starting unit.
Table limits break negative progressions before mathematics does
Every casino table has a minimum and maximum.
A player beginning at $10 on a $10–$500 table cannot continue a strict Martingale indefinitely. The sequence reaches $320 and then requires $640, which exceeds the table maximum.
The system’s recovery promise therefore depends on a wager the rules do not permit.
Players sometimes respond by starting at a smaller base unit. That only delays the limit. Exponential growth eventually catches any finite maximum.
Bankroll limits are even more personal
A player does not need to reach the posted table maximum to be functionally stopped. The bankroll may run out first.
If the player brings $500 and starts a $10 Martingale, five losses consume $310. The next required bet is $320, which the remaining bankroll cannot support.
A progression can therefore fail long before the player experiences what feels like an extraordinary losing streak.
This is the central asymmetry of negative progressions: they aim for many small recovery wins at the cost of occasional severe failure.
Progressions can create misleading win-rate memories
A system may generate many sessions that finish with a small profit. That can make it feel superior even if the long-run expectation is still negative.
For example, a Martingale can end numerous short sequences with a one-unit gain. The player remembers “the system works most of the time.”
But the relevant question is not how often a session wins. It is:
How much is won in ordinary winning sessions compared with how much is lost when the progression fails?
A system that wins $10 nine times and loses $200 once has a 90% session win rate but a net loss of $110 across the ten sessions.
Win frequency and profitability are not the same metric.
Progressions do not repair a high-edge wager
Another common mistake is applying a sophisticated progression to a poor base bet.
If the player uses a loss system on a high-edge proposition wager, the progression multiplies exposure to an already expensive bet.
The first decision should therefore be bet selection, not progression selection.
Choosing a lower-edge base wager reduces expected cost per dollar. Only after that choice does it make sense to discuss whether the player wants flat exposure, presses, regressions, or another risk pattern.
A better way to compare systems
Before using any progression, write down five numbers:
- base unit;
- maximum wager the system can require;
- total bankroll available;
- table minimum and maximum;
- expected total action if the sequence runs normally and if it runs badly.
Then ask:
- What happens after five consecutive losses?
- What happens after five consecutive wins?
- Can the bankroll support the next required wager?
- Can the table legally accept it?
- How much prior profit can one losing step give back?
- Does the system cause me to wager more simply because I am emotional about the previous result?
If those answers are uncomfortable, the progression is not conservative merely because its rules are written on a card.
The useful comparison
Betting progressions are best understood as different ways of packaging variance.
Flat betting keeps stake size stable. Martingale concentrates risk after losses. Positive progressions such as Paroli or 1-3-2-6 concentrate more exposure inside winning runs. Fibonacci and cancellation systems slow or disguise escalation without changing the underlying edge. Regression reduces future exposure after an initially larger bet.
None of those structures changes the dice probabilities. None converts a negative-expectation craps wager into a positive-expectation one by stake sequencing alone.
The progression can change the shape of the ride. It cannot change the mathematical price of the ticket.