Fair Dice in Online Backgammon: How to Check

Updated 6 November 2026 · Standard formula, shown below

This guide helps you determine whether an online backgammon game uses fair dice—meaning each die roll is equally likely and independent of previous rolls. You’ll learn practical checks you can perform without special software, plus what to do if you suspect foul play. No technical background is needed, just a bit of patience and a simple spreadsheet or notebook.

What You’ll Need

A spreadsheet program (Excel, Google Sheets, LibreOffice) or just pen and paper
Access to the online backgammon game you want to test
Time to record at least 200–300 rolls (more is better)

Step-by-Step

Understand what “fair dice� means. In a fair game, each of the 36 possible outcomes of two dice (e.g., 1+1, 1+2, … 6+6) has an equal 1/36 chance. Also, each roll is independent—the previous roll does not influence the next. Any significant deviation from these probabilities suggests unfairness.
Record your rolls. Start a new game or play normally. Every time you roll, write down the two numbers (e.g., “3-5�). Do this for at least 200 rolls, but 500 or more gives a much stronger test. Be consistent—record every roll, not just the ones you remember.
Count the frequency of each die face. For each roll, note both dice separately. So a roll of 3-5 counts as one “3� and one “5�. Tally how many times each number 1 through 6 appears. With 200 rolls, you have 400 individual die results.
Compare against expected frequencies. For a fair die, each face should appear about 1/6 of the time. With 400 individual results, you’d expect about 66.7 of each number. Calculate the difference between your observed count and 66.7. If any face differs by more than about 20–25%, that’s a red flag.
Check the distribution of sums. Add the two dice in each roll (e.g., 3+5=8). For two fair dice, the most common sum is 7 (probability 6/36≈16.7%), and the least common are 2 and 12 (each 1/36≈2.8%). With 200 rolls, you’d expect about 33 sevens and about 5 or 6 twos/twelves. If you see sevens appearing far more often than expected, or twos/twelves almost never, that’s suspicious.
Look for streaks and patterns. Fair dice are random, so long streaks of the same number, or alternating patterns, are rare but possible. However, if you see a pattern like “always a 6 after a 1� or “never doubles twice in a row,� that’s not random. Write down the sequence of rolls and scan it visually.
Run a simple chi-square test (optional but powerful). In a spreadsheet, list the observed counts for each die face (e.g., in cells A1:A6). In column B, put the expected count (total rolls/6). In column C, compute (A-B)^2/B for each row. Sum column C. Compare this sum to a critical value—for 5 degrees of freedom, a value above 11.07 suggests unfairness at the 5% significance level. (If you’re not comfortable with statistics, skip this step—the frequency check is enough for a basic sanity test.)
Test across multiple sessions and accounts. One short session might be unlucky. Play several games over different days, and if possible, create a second account (if allowed) to see if the pattern is consistent. A fair game should show no difference between sessions.
Check for “human-like� dice. Some sites use algorithms that avoid repeated rolls to feel more random to humans, but that actually makes them less random. If you notice that the same number never appears twice in a row, or that doubles are rarer than expected, that’s a sign of a biased algorithm.
Compare with a real dice simulator. Generate 200 rolls using a trusted random number generator (e.g., a physical die, or a well-known app). Record those results and compare the patterns. If the online game’s distribution looks very different from the simulator’s, that’s evidence of unfairness.

Common Mistakes

Sample size too small. With only 50 rolls, you might see 10 sixes and think it’s rigged, but that’s within normal variance. Always aim for at least 200 rolls, ideally 500+.
Counting rolls incorrectly. Forgetting to record both dice separately, or mixing up the order, will skew your counts. Double-check your tallies.
Ignoring the “doubles� distribution. Doubles (1-1, 2-2, etc.) should occur about 1/6 of the time (6 out of 36 outcomes). If you see doubles far more or less often, that’s a clue.
Assuming a streak means rigged. Random sequences can have long streaks. A run of five 6s in a row is unlikely (1/6^5 ≈ 0.01%) but not impossible. Only a consistent pattern over many rolls is suspicious.
Not considering the game’s rules. Some backgammon variants use a doubling cube or special rules that affect dice rolls (e.g., “Crawford rule� doesn’t affect dice, but some games use a “dice off� for opening roll). Make sure you’re recording only the actual dice rolls, not any modified outcomes.

Tips

Use a spreadsheet formula to tally. If you record rolls as “3-5� in column A, you can use formulas to split and count. For example, in Google Sheets, use =COUNTIF(A:A, “*3*”) to count all rolls that contain a 3 (but that counts 3-3 twice, so be careful). Better: put each die in separate columns (A and B), then use =COUNTIF(A:A,3)+COUNTIF(B:B,3).
Keep a log of the time and date. If you notice a problem, you can report it to the site with evidence. Also, some sites might adjust their RNG over time, so a log helps you see changes.
Play for free first. If you’re testing a site, use free play or low-stakes games to avoid losing money while you collect data.
Look for the site’s RNG certification. Reputable sites often display certifications from independent auditors like eCOGRA or iTech Labs. While this isn’t a guarantee, it’s a good sign.
Trust your gut but verify with data. If you feel the dice are unfair, your feelings might be right, but only data can prove it. Don’t accuse a site without solid evidence.

Troubleshooting

What if the dice seem unfair but the chi-square test says it’s fine? The chi-square test is not perfect—it can miss subtle biases, especially with small samples. Try increasing your sample size to 1000 rolls. Also, check if the bias is in the combination of dice (e.g., certain pairs appear too often) rather than individual faces.

What if the game only shows the total, not individual dice? Some games display just the sum (e.g., “8�). In that case, you can still test the sum distribution, but it’s less powerful. You can also try to infer the individual dice from the board movement—if a checker moves 5 spaces, you know one die was a 5 (unless it’s a double, but you can usually tell). Record both dice whenever possible.

What if the site uses a “live� dice generator? Some sites use physical dice with a camera. In that case, the dice should be truly random, but you can still record rolls and test for bias. If you see a pattern, it might be due to the dice being worn or the camera angle.

What if you suspect the site is rigged but can’t prove it? If you have strong evidence (e.g., 1000 rolls with a chi-square value far above the critical threshold), you can contact the site’s support and share your data. If they don’t respond or dismiss you, consider moving to a different site. There are many reputable backgammon platforms.

Frequently Asked Questions

Q: How many rolls do I need to be sure?
A: Statistically, 200 rolls gives you a rough idea, but 500–1000 rolls is much more reliable. With 1000 rolls, you can detect even a 5% bias with high confidence.

Q: Can a site legally use unfair dice?
A: In regulated jurisdictions, no—online gambling sites must use certified RNGs. However, some unregulated or free-play sites might not be held to the same standards. Always check the site’s licensing.

Q: What if I see a pattern like “always a 6 after a 1�?
A: That’s not random—it suggests a deterministic algorithm. Record 100 such instances and count how often the pattern holds. If it’s 100% of the time, that’s almost certainly rigged.

Q: Does the doubling cube affect dice fairness?
A: No, the doubling cube is a separate mechanism. It doesn’t influence the dice rolls, so you can ignore it when testing.

Q: Are there any tools that can automatically test dice?
A: Some advanced players use software that logs rolls and runs statistical tests automatically. However, you can do the same with a spreadsheet and a bit of effort.

Q: Is it possible that my own bias is affecting the test?
A: Yes—if you subconsciously only record rolls that seem “unusual,� you’ll skew the data. To avoid this, record every single roll without exception, even if it’s boring.

Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.