craps; dice; dice control; probability; hot hand fallacy; illusion of control


Applied Statistics | Engineering Physics

Document Type

Original Research Article


This paper presents our findings from experiments designed to test whether we could use a custom-made dice throwing machine applying common dice control methods to produce dice rolls that differ from random. In earlier research we calculated the percentages of control a craps player needs to break even or beat the house (Smith and Scott, 2018). Using the most common practices of dice control in craps, we established how dice should be configured (i.e., set) and thrown to achieve certain outcomes such as not rolling a seven in the point cycle. We decided to run experiments to see if a dice throwing machine that generally mimics the biomechanical properties of expert craps players (e.g., back spin, on-axis throw, repeatable throwing angle etc.) could achieve at least a break-even level of control. Using our machine (named “Lucky Lil”) on a six foot craps table we filmed dice throws using a Phantom® VEO4K 990s high-speed digital camera captured in 4K resolution. Watching these videos we calibrated the machine to ensure throws were on-axis and flying correctly. We then recorded 7,557 craps throws using various settings. We use chi-squared tests to determine if we were able to produce a statistically significant number of on-axis rolls.

Funding Sources

Monmouth University awarded us a creativity grant to fund the costs of our experiments.

Competing Interests