Controlled shooting in craps is the claim that a shooter can use dice setting, grip, launch, landing zone, rotation, and repeatable mechanics to change the final distribution of outcomes enough to gain an advantage. A smooth-looking throw is not the relevant test. The relevant test is whether the measured result distribution changes by a useful amount under real casino conditions.
A controlled motion is not automatically a controlled outcome
Many players can make their delivery look consistent. They can set the same faces, place the dice in the same fingers, use a similar arc, and aim at the same landing area. Those are observable motor actions.
The edge claim begins only after the dice leave the hand. They hit felt, bounce, interact with each other, and normally reach the opposite end of the table. Back-wall geometry and random collision make the final state difficult to preserve.
So there are two separate claims:
- I can repeat my throwing routine. This may be plainly true.
- My routine changes final dice probabilities enough to beat the posted wagers. This requires statistical evidence.
Confusing those claims is the source of much of the controlled-shooting debate.
What the shooter would need to change
Under fair random two-dice behavior, seven appears in 6 of 36 ordered combinations, or 16.67%. Six and eight each appear in 5 of 36, or 13.89%. Other totals have their own fixed combination counts.
A shooter trying to improve Pass/Place-style action might claim to reduce the frequency of sevens after a point is established. Another might claim to create correlation between the two dice or keep them on particular axes. A hardway specialist might care about a different distribution shift.
The claim therefore has to specify which outcomes change and in what direction. “My rolls are better” is not testable enough.
The sample-size trap
Short samples are capable of producing impressive-looking streaks by chance. If a shooter rolls 60 times and gets only seven sevens, that is below the random expectation of 10. It still does not establish control.
Even 500 rolls can leave substantial uncertainty for a small claimed edge. The Wizard of Odds discussion of dice-control samples explicitly notes that 500 throws are too few to make a strong case for or against influence, even though they are better than no data.
A tiny advantage is especially demanding. If the alleged skill changes probabilities only slightly, thousands of rolls may be needed before the signal can be separated confidently from ordinary randomness.
Record the hypothesis before the rolls
A credible test starts before the first throw.
If the claim is “I reduce sevens after the come-out,” define that statistic in advance. If the claim is “I increase zero-pitch outcomes,” define how those outcomes will be classified. If the claim is “my 6 and 8 rate is higher,” specify whether all throws count and how come-out rolls are treated.
Then record every qualifying result. Do not discard bad sessions, change the target after seeing the data, or stop the trial the moment the results look favorable. Those practices create selection bias that can turn ordinary variance into apparent skill.
A predeclared test is less exciting than remembering a monster hand, but it is the only way to know what the record actually demonstrates.
Practice-rig results and casino results are different evidence
A controlled shooter may practice on a private table with a chosen surface, familiar bounce, quiet environment, and no time pressure. A casino introduces additional constraints: table dimensions, felt, back-wall material, chips in the landing area, crowd movement, dealer pace, and enforcement of legal delivery.
A technique that appears to create a shift in practice still has to survive those conditions. The casino environment is not an unfair complication added after the fact; it is the environment in which the supposed monetary edge must exist.
This is why Craps Advantage Play Reality treats practical enforceability as part of expected value rather than as a separate issue.
What live-table procedure does to the claim
House procedures vary, but the crew generally wants a legitimate throw to travel to the far end and interact with the back-wall area. Repeated short rolls, sliding, deliberate interference with the dice, or excessive delay can lead to warnings or a change of shooter.
The stickperson and box/floor staff do not need to settle the physics debate at the table. They enforce the rules. If a player’s claimed control requires a delivery the casino will not accept, the technique is not a usable casino strategy at that property.
Conversely, a casino allowing dice setting does not mean it believes the player can influence outcomes. Dice setting can be tolerated because it is common ritual and, for most players, does not create a demonstrated threat.
What the best pro-control argument actually says
The strongest version of the argument does not claim that every skilled throw lands exactly as intended. It claims that a small fraction of throws may retain enough correlation or axis behavior to shift the aggregate distribution.
The Wizard of Odds hypothetical dice-setter model explores that idea mathematically. In that model, even a modest “skill factor” could theoretically overcome the very small blended edge of Pass Line with substantial odds.
That is useful analysis because it quantifies the hurdle. It does not prove that a particular player possesses the modeled skill. The model answers “if the distribution changes this way, what would the edge be?” The shooter must still prove the antecedent.
Why RSR by itself is not enough
Controlled-shooting communities often discuss rolls-to-sevens ratio (RSR). A random shooter averages six rolls per seven in the long run because the seven probability is 1/6. An observed RSR above six may look favorable.
But RSR alone can be misleading if it is taken from a short or selected sample. It also does not fully describe the distribution of non-seven outcomes. Two shooters could have the same seven rate but very different frequencies of points, hardways, and other totals.
The wager choice matters. A distribution shift is only valuable if it improves the expected value of the actual betting portfolio enough to overcome its casino price.
A bankroll is not a laboratory instrument
One of the most dangerous ways to “test” controlled shooting is to increase wagers while the evidence is still uncertain. A player may believe the growing bet size proves confidence, but money at risk does not improve statistical validity.
If the goal is investigation, separate the data collection from the betting decision. Record results, define the test, and evaluate the distribution. If the goal is entertainment, keep the wagering decision within the bankroll you are prepared to lose and do not describe the experiment as a proven edge.
The variance simulator can demonstrate how random sequences produce long rolls and unusually low seven counts without any control.
What would change the verdict
The skeptical position is not “dice influence is impossible by definition.” A strong body of evidence could change the conclusion. It would need repeatable results from a predeclared method, large samples, independent replication, casino-relevant conditions, and a measured shift large enough to create positive expected value after the rules and payouts are included.
That standard is demanding because the financial claim is demanding. If someone says a technique can reverse a casino game’s edge, the evidence should be strong enough to distinguish the effect from ordinary luck and selective memory.
The useful conclusion for players
Controlled shooting remains an unproven bankroll strategy for ordinary players. Dice setting can be a harmless ritual. Practicing a consistent throw may improve etiquette and reduce wild dice. Neither of those benefits demonstrates a positive expectation.
If you enjoy the technique, keep the claim proportional to the evidence. A disciplined throw is observable. A statistically verified distribution shift is measurable. A positive edge is calculable. Those are three different milestones, and skipping from the first directly to the third is where myth replaces analysis.
For the neighboring issues, read Dice Control Myth, Dice Setting Myth, and Can Craps Be Beaten?.