UNLV study reveals that gamblers can’t detect slot machine payout percentages
July 16, 2019
It's a common sight on casino floors: patrons jumping from slot machine to slot machine before eventually hunkering down at a game that’s due for the next big payout. But can players — even the regulars who frequent a particular property — really tell the difference between the house edge on one game from that of another?
Nope. At least not according to a series of recent studies led by Anthony Lucas, a UNLV Hospitality College professor and former gaming industry operations analyst.
For the past several years, Lucas and colleague Katherine Spilde from San Diego State University have taken to casino floors on multiple properties in the U.S., Australia, and Mexico to investigate. Their results contradict long-held beliefs by casino operators about a player’s ability to detect differences in how much — and how often — a slot machine pays.
"I think some operators are naturally and understandably cautious of new information that challenges traditional industry practices," said Lucas. "But we must consider how we know what we know. This is where our work takes on a Moneyball-like aspect — questioning the wisdom of widely held beliefs when data show that a new way of thinking may be better."
In their latest study, the UNLV-led research team compared two pairs of reel slot games at a "locals" casino in suburban Sydney, Australia, where all wagering occurs on electronic gaming devices.
Their process is relatively straightforward: take two identical slot machines, positioned in similar places on a casino floor, but vary the par — the percent of total coin-in that the machine keeps over time. For example, if the par on a game is set at 10 percent, the machine would be expected to retain $10 of every $100 wagered, on average, over the long term. But in the short term, this rarely happens, increasing the difficulty of par detection.
For this study, researchers compared the daily performance of pairings for the games "Tokyo Rose" and "Dragon’s Fortune X" over a nine-month period. The pars within each pairing ranged from 7.98 percent on the low end to 14.93 percent on the high end.
Researchers measured daily coin-in for each machine as well as its T-win, a formula that multiplies coin-in and par to calculate a machine’s expected value, or its theoretical win. If, over the course of the nine-month test, regular players could detect a difference in the pars, this comparison would reveal whether (and how much) players migrated from higher par to lower par games.
As Lucas predicted, differences between the high and low par games remained stable throughout the length of the study, which meant that there was no statistically significant indication of play migration.
And while the lower par machines had more coin-in over the course of the study period, the T-win was greater on average for the high par machines. The positive impact from the elevated T-win on revenue for the higher par machines more than compensated for the decline in coin-in on those machines.