Luck vs. skill

[UtahPhillips]

Whoever did the little write-up on backgammon on BGA is rather unknowledgable about the game.


they rated luck as 4, strategy as 2

Should be more like strategy 5, luck 2, or 3 at most

[ReelMcKoy]

I think I agree with hervepa on this. Although it can often feel like a simple matter of lucky rolls, it is clear there is a great deal of strategy and skill involved. And for this reason we see the very best players often win. (but the luck still gives a starting player a chance against the best players)

[euklid314]

Since there is no doubling cube on BGA I would suggest the rating "luck 4, strategy 4".

There is tons of strategy and on the long run you will prevail with better play.

But for a single game luck is very (!) important. Compare it with chess: After 1 months, when you have got a basic grip on the tactics you will win 0 out of 100 games against the world champion. After 1 months, when you have got a basic grip on the tactics you will win at least 10 games out of 100 against the world champion (or against XG). It took me a very short time to win approx 30% of the games against XG (that plays better than any human).

The reason is that luck often wins you a game even with suboptimal play. Actually that is one very attractive part of the game!
However: As you try to master the game on a higher level you have to accept the fact that you will often lose a game even with better play. If you cannot accept this fact, the game of backgammon is not yours.

Only in the long run the strategy will give you the reward you might deserve...

[Baricuda]

If you’re rolling dice or dealing cards then luck is still a dominant factor with skill only influencing outcomes ranges. Get two Ai of the same code base and level to play itself and you will see how dice are rolled or cards are dealt will determine the game. Exception are games like poker where it becomes more psychological than about cards etc .

[Romain672]

Let me do my way to see the luck part of the game: the highest elo.
So as usual, a game like Yahtzee is around 350, an high skilled game at around 800.

And backgammon is at... 550.

The point is not to decide if a game is about luck or strategy, the point is to give a good idea of how much you can win a game with perfect strategy against a player who put X hours in the game.

I don't believe the strategy of a simple game should ever goes to 4 or 5 (prove me wrong ty, would be really curious about a game of complexity 2 which got 4 in strategy ).
If I take Hanabi which can have virtually infinite strategy, it's writed at 3, which look fine because most of the strategy can come in some really rare spots.
I'm not playing much high strategy game, but imo, that's mainly a game where, even if you are good in strategy, at first glance, you will have no idea what is the good thing to do. On games like Hanabi or Backgammon, you can totally have a good grasp of what seem good.


edit: Okay I just find one: https://boardgamearena.com/gamepanel?game=thecrew . But meh

[dschingis27]

And backgammon is at... 550.

I don't think that Elo of the highest rated player is a good measure, especially for short games with a lot of luck. There is just a lot of cheating going on or at least some way to abuse the system is often applied. More often than not you will find the highest rated player conducting some "strategy" in boosting Elo. But ok, let's assume that there is no cheating - even then the highest rated player will often have had a large streak of luck and it is highly unlikely that the high rating can be sustained when playing regularly.

Much more reliable to look at average Elo of roughly 11th to 20th rated players. But even this can be misleading..

I don't believe the strategy of a simple game should ever goes to 4 or 5 (prove me wrong ty, would be really curious about a game of complexity 2 which got 4 in strategy ).

I would say that many abstract games fall into this category of low complexity and high strategy. Take Go for example. But of course it depends on how you define complexity.

[dschingis27]

This is a really good source on the topic:
http://freerangestats.info/blog/2016/03/19/elo-pr-luck

[Tuut Tuut tis Luut]

Something else worth mentioning. Remember those ratings on the game page, are clearly meant for new players who have never played the game themselves to get a feel for what the game is like. Not to give a completely accurate calculation of the luck factor.

Regardless of how much strategy you feel backgammon has. There's no getting around the fact that dice rolls determine every move you make. High luck low strategy is simply the best description to show a new player what the gameplay is like. Someone looking for low luck game will not want to play a game that's totally dice based. No matter how convinced you may be the luck ultimately doesn't matter that much.



But also, I love backgammon, it's easily one of my favorite games, but let's not pretend it's something it's not. Luck simply plays a very significant part. That's why the doubling cube is a thing and why you'd never see a serious BG tournament based on single games.

Hell, even the strategy part of the game is entirely based around shifiting the probability in your favor for future rolls. (I.E. increasing your chances of getting lucky rolls.) It's a game of creating your own luck, but that does mean luck is included in the package.

No need to pretend like it's more strategic than it is.
It's not Chess, that's why we love it.

[Asia-Funtasia]

Something I really can't get used in backgammon on boardgamearena is, that there is no difference in how good you win. Victory points should be doubled if you win with a gammon. Gammons make the game more skill and less luck.

The other thing is the possibility of hitting by passing in your homebase. I didn't know, that is possible in official rules. I learned backgammon the persian way: you have to stay, where you hit the opponent, if it has been in your own house. Then you would think twice, if you really want to hit. I found the rules online called Tavla.

(that the dices are ridiculous is obvious)

[buv3x]

Here are some of my thoughts on the topic.

The first one is - this is a really good and informative take, thanks a lot for the link dschingis27 :

This is a really good source on the topic:
http://freerangestats.info/blog/2016/03/19/elo-pr-luck

However, for a case of one-point matches that we have on BGA it's possible to make a more concrete calculation. A metric that can be used is MWC (match winning chances). For any game it starts from 50% for each player and ends up with one having 100%. So the swing is always 50% to a winner, which attributes to two values. Difference in total luck (1) plus difference in total errors (2) made by each player. Both of the values for any given game are a given by engine analyses, and with the current engine quality you can be confident in those values up to maybe 1-2% (most probably even less than that).
So in this case an objective metric of luck vs skill rate for a single game would be an average of 100%*(1)/((1)+(2)) taken over a big set of games.

Now, I've started analyzing my games after I've wrote a parser for BGA replay logs, so that now it takes little time. I haven't gathered any data yet, but I plan to start doing it, so that I can return with concrete numbers. But my rough estimate based on the games' analysis I've checked so far is about 80/20 luck/skill (which means an average a winner has 10% winning difference based on decisions and 40% based on dice rolls). That might sound a bit harsh, but it actually might even be generous for the skill part. Anyway, I'm gonna post an update after gathering data from some 100+ games.