Dice Randomness

Backgammon Pro now available on Windows Store (Windows 8)! HERE

Please read all this post in full before sending your comments. If, after reading this post, you still have concerns please use the form in this web site to submit your question to support.


Backgammon Pro has three methods of drawing random dice values, depending on the ‘dice setting’:

a) TRUERANDOM: the game has a list of 100,000 values created from the web site WWW.RANDOM.ORG — The site generates truly random numbers from atmospheric electric noise.

b) EXCEL: the game has a list of 100,000 values created using the Excel RANDOM function

c) BALANCED: the game uses the TRUERANDOM list, but tries to balance the number of doubles each player gets

d) MERCENNE TWISTER: uses the most commonly used algorithm to generate random numbers in Backgammon. Used by the best Backgammon games ever built (NEW !)

Every turn, the game ‘takes’ the next value from the table, and assigns it to the player or to the AI that is playing in that turn. There is no more analysis or modification of the dice values.

The game then stores this value in the ‘statistics table, and displays this table in the area where you can see if the dice values are random or not. This information can show if the dice values are favoring one player or the other. The analysis includes number of doubles per player, number of times the player or the AI were blocked from entering from the bar, number of doubles thrown during bearoff, etc. This is helpful in finding out if dice are favoring the player or the AI. This statistics become helpful after at least 1000 dice throws.

Backgammon is a game of skill, but there’s also a component of luck added by dice. There will always be a lucky set of rolls that can cause your opponent to win, even if you play perfectly. The same holds true for you. If you’re honest with yourself and don’t automatically attribute all your wins to skill, you’ll notice lucky sets of rolls that cause you to win when you normally wouldn’t have. However, advanced players know that what matters is their Error Rate and Checker Play Skill Percentage. Advanced players work to improve their moves and lower the number of errors they make instead of focusing on win/loss. If you change how you approach the game and work to lower your Error Rate and increase your Checker Play Skill Percentage instead of focusing on win/loss, you’ll find that over time, your game will improve, and as a result, you’ll win more.


Most backgammon players have an ELO rating of 1600 or lower, so it will be difficult to match the skill vs the level played by the phone in hardest level. Only world class players or players that have ELO rating higher than 1800 can judge accurately judge how well Backgammon Pro on Expert level is playing, that means that anyone stating that they have proof that the dice cheat, would need to be the top players in their country or region to accurately assess if that the dice are rigged.

Backgammon Pro absolutely does not cheat.

It may seem to you as if Backgammon Pro is getting the rolls it needs. Sometimes it can get lucky, as any player can. But often, what appears as luck is actually playing skill. The neural network-based AI has evaluated millions of backgammon positions and can make extremely educated moves in almost every situation. It puts itself in positions that maximize the number of useful rolls for itself and minimize the number of useful rolls for its opponent. Unless you’re a top backgammon player or have evaluated millions of positions like the AI, you can’t expect to play as well as the AI on Expert skill level.

Try playing other top backgammon games such as GNU Backgammon or Snowie. These are world class backgammon games. When set at the hardest skill level, these games play better than almost all human players. If you are not one of the top-ranked backgammon players in the world, then it will be difficult for you to win more than 60% of the times against Backgammon Pro on the Expert level.


Try playing the Easy or Medium level, and set the dice random option to ‘Balanced’. You’ll have a better chance of winning, and you can work on improving your game. Expert level plays better than the majority of backgammon players, will beat you more often than not, and may be frustrating for some.


There are more than 150,000 games played daily of Backgammon Pro, this means, for example, that there are tens of games DAILY where the AI will get 7 doubles in a roll using TrueRandom. Talk about ‘bad luck’. That is all there is, it is bad luck.

All the good backgammon games in iPhone, Android or Windows Phone have lots of negative feedback as there is a perception that the dice are loaded, or fixed, or chosen by the AI to favor the game, see for example Backgammon NJ (Android, iPhone).

Either all the best games cheat, or the in fact most of these games are well written, and the expert level has very strong analytical and statistical analysis that lets is play at a level above most players.


We took great pains to develop a well thought out game with a strong AI that doesn’t cheat.

1) If Backgammon Pro isn’t right for you, please do try other backgammon games.
2) If you want to improve your game, try GNU Backgammon, it is free, and you can see the source code to get the idea of the type of backgammon opponent you will face in Backgammon Pro
3) If you believe firmly on your opinion, contact us through this page (see bottom of the mail).

If you buy this game because you’re looking for a worthy opponent that can improve your game coupled with a great user interface, we thank you very much for your support!


Yes, we can prove that the game does not cheat. We can prove that the dice are immutable, no matter who is playing what, and the game never manipulates the values that are taken from the table.

The random dice value tables, with 100,000 values, is always used, no matter if it is a two player or one player vs the AI.

If you are really interested in seeing that the game does not cheat, write down the last 10 dice throws (exactly as they appear in the bottom of screen, shown with the “HISTORY” setting turned on), stop playing, and send us the values of these last dice throws (the values shown as {n,n} in the bottom of the screen). Let us know your dice setting (TrueRandom or Excel), and version of the game (e.g. 1.7f for Windows Phone). We can send you the next dice throws values. These values will not change even if you are winning (or losing). This will prove that the table of values and the order of the dice never vary, they are precalculated on purpose to eliminate the possibility of the AI ‘cheating’.


If you have written an unfavorable review of Backgammon Pro, or firmly believe that the game is cheating or, we invite you to post on this page.

Please include the detailed explanation of the problem, and include information about how would you fix the problem (for example which statistical test to use to determine statistical anomalies in dice throwing).

We don’t have information to be able to answer comments that only say ‘the game cheats’, without taking into consideration all the information in this blog post.

Dev Team


About the Author:


  1. curt

    Been tracking the hits on open pieces. Works out to approximately 4 to 1 phone to player. Even the standard stats you set up favor the phone although not significantly. The most frustrating thing is the number of combination needed to make a hit, still significantly higher for the phone at almost 5 to 1. The phone hits on 2 of 3 on these, while the player hits on 1 of 7.

    • Curt, I’ve been thinking about your mail. What would I have to measure to analyze this statistic? If this can be done, then probably it can help show if there are deviations/biases.

      • joe

        Measure the percentage of hitting a direct ace shot.. Where you need a one to hit. This direct shot should be hit 11/36 of the time .

    • Jamie, thanks for the comment, that is bad luck indeed!. There are three methods of generating dice (chosen in settings):

      a) TRUERANDOM: The dice are generated from random atmospheric noise (it is true, the numbers are taken from http://www.random.org, which generates it from static electricity in the air). There is no influence from the program. The statistics help show if you are losing because your level is not good enough, or, as it seems it is in this case, really bad luck (really bad!). There are other players that have written to me with the opposite situation, 80% more double sixes for them than for the phone.
      2) EXCEL: 100,000 numbers generated in Excel, already precalculated, that possibly are more uniform than TrueRandom.
      3) BALANCED: in this case the program tries to balance the number of doubles each player gets.

      350 rolls is still a low number, I would recommend two things: 1) keep playing more games (some players have played more than 4000 games) and possibly the randomness of the dice will make you have more doubles (but I can’t guarantee it, as the program has no control to control over the dice values in TRUERANDOM setting). Or 2) change the random setting to BALANCED, and the statistics will hover where you are, but the phone will always report it had ~14 more rolls.

      Hope this helps,

  2. paul

    I still say, no matter what stats say, the dice are in favor of the phone. Phone always gets perfect sets and dice that are way beyond the odds. No matter what the cube is on can never gain more than 2 points per game.

  3. Kate

    I have played “real” backgammon my whole life (my grandmother began teaching me when I was 5), and I have played numerous computerized versions as well. As with any game that involves gambling (whether you’re betting money or points) there is a certain gap between players. Not being 50/50 doesn’t mean its being skewed in favor of the AI intentionally… Its the odds. Its Random… Considering the technology this game uses, I can’t imagine a more realistic game, had I the ability to play with my Grandmother’s ghost. 🙂 If you are uncomfortable with a game of odds, try Chess. Its 100% you to blame if you lose. Good job, guys. Keep it up.

  4. Dimitri Mihaylov

    Did anyone complain that phone sometimes has bad luck too? For me it happens too often LOL… Please, implement the ability for your program to publish it’s laments here automatically concerning it’s bad luck…

  5. Richard Wilson

    The phone plays a very good game. Level four is a great challenge – it closely resembles the play of someone I played over 1500 games with and I experienced similar results to the point of almost losing a friendship! You can win against the phone on level four but only if you concentrate because the phone will extract full value for any opportunities you let it have. The phone is a relentless and remorseless opponent only on level four, on level three it plays very well, however its play is fundamentally limited. Even so the phone’s game is remarkably similar to that of people. Don’t blame the dice – use the cube to advantage. Personally, I’d welcome a level harder than medium but easier than hard!

Leave a Reply