I’ve always wanted to make my own slot machine game. I’m a decent enough programmer, but the slot machines are more than just a spinning reel with a set number of symbols that randomly stop at different positions.

A random number generator is used to determine the outcome of each spin. Thus, each spin is independent of the last and doesn’t influence the outcome of the next. Each spin has the same chance of winning as any other. The RNG is what makes slots fair and unpredictable. If a player hits the winning combination on the first spin, they could do it again on the next one or not see it for thousands of spins. The odds, on the other hand, are always different as they depend on the game and the house edge.

The trick to making a realistic slot machine is that each matching combination that would result in a payout has better odds the lower the payout is. So if having a single cherry on a pay line was the lowest amount you could win in a spin, it would have the highest odds. Inversely, if getting three 7’s in a row was the jackpot then it would have the lowest odds. This is called probability.

Probability is defined as the extent to which an event is likely to occur. This is measured by the ratio of the favorable cases to the whole number of cases possible. It can be expressed as a fraction, decimal, percentage, and odds. Modern-day slot games can have hundreds of symbols per reel and are programmed to have different probabilities for each symbol to come up. The payouts over a large number of spins are referred to as Return to Player or RTP . The RTP can never be 100%, and it’s the difference between that 100% and the RTP that determines the house edge and your actual odds of winning at a casino slot machine.

RTP is usually used to decide on the payback to players over a period of time, most often every month of a calendar year. Percentages are used to describe the RTP and estimate the wins. The bigger the bets, the higher amount of wins returned to the players. To determine how likely players are to win slot games a, we must first define the house edge and the price of a slot. Most casinos feature a house edge of up to 10%, meaning that the casino expects a 10% gain from all wagers. This value is derived from the adjustable RTP of the slot machine.

An example of this is that a hundred $1 bets with a 90% RTP can expect to win $90 overtime, with the house getting $10 (10%).

So the trick is asking chatGPT in a way that will make it write code.

Here are the things I chose to include:

  • Number of symbols
  • Number of reels
  • Number of rows
  • Number of pay lines
  • Adjustable RTP

Here is my query to chatGPT:

Write a Slot Machine app in JavaScript with 5 reels, 6 symbols, three rows and 9 pay lines. Include the ability to adjust the odds of winning per pay line.

Here’s an example implementation of a slot machine app in JavaScript with 5 reels, 6 symbols, three rows and 9 pay lines:

const REELS = 5;
const ROWS = 3;
const SYMBOLS = 6;
const PAY_LINES = 9;

// Define the symbols and their corresponding odds of appearing
const SYMBOL_ODDS = {
  symbol1: 0.1,
  symbol2: 0.2,
  symbol3: 0.3,
  symbol4: 0.4,
  symbol5: 0.5,
  symbol6: 0.6

// Generate a random symbol based on the odds
function randomSymbol() {
  const oddsTotal = Object.values(SYMBOL_ODDS).reduce((a, b) => a + b);
  let random = Math.random() * oddsTotal;
  for(const [symbol, odds] of Object.entries(SYMBOL_ODDS)) {
    if (random < odds) return symbol;
    random -= odds;

// Generate the reels
function generateReels() {
  const reels = [];
  for (let i = 0; i < REELS; i++) {
    const reel = [];
    for (let j = 0; j < ROWS; j++) {
  return reels;

// Check the pay lines for any winning combinations
function checkPayLines(reels) {
  const payLines = [];
  for (let i = 0; i < PAY_LINES; i++) {
    let payLine = [];
    let (let j = 0; j < REELS; j++) {
      payLine.push(reels[j][i % ROWS]);
  return payLines.filter(payLine =>
    payLine.every(symbol => symbol === payLine[0])

// Spin the reels
function spin() {
  const reels = generateReels();
  const payLines = checkPayLines(reels);
  return { reels, payLines };

// Example usage
const result = spin();
console.log("Reels:", result.reels);
console.log("Pay lines:", result.payLines);

This code generates random reels based on the odds of each symbol appearing, checks for any winning combinations in the pay lines, and returns the result. You can adjust the odds of each symbol appearing by changing the values in the SYMBOL_ODDS object.

Taking the code provided and making minor modifications that allowed me to display some symbols and a button to allow manual trigger of the spin() code, here is the result:

It is clear to me that without the specification of what spots construe a pay line, this iteration is basic at best. This code does not generate any actual wins that I can see. chatGPT needs a nudge in the right direction in this regard. Stay tuned as I ponder on how to get it to adjust its code to accommodate this requirement.