Opening Rolls

Forum Archive : Opening Rolls

Magriel's Chapter 5

From:   Hayden Alfano
Date:   10 May 2006
Subject:   Magriel Chapter Five - Opening Rolls

I am currently reading Magriel's "Backgammon," and Chapter Five
is about opening rolls. Magriel breaks the opening rolls into five

Category 1: Closing Strategic Points
Roll Recommended move
3-1     8/5, 6/5
6-1     13/7, 8/7
4-2     8/4, 6/4

All of these seem rather obvious, without much room for debate.

Category 2: Bringing Checkers from the Midpoint as Builders
Roll Recommended move
5-4     13/8, 13/9
4-3     13/9, 13/10
5-2     13/8, 13/11
5-3     13/8, 13/10
3-2     13/10, 13/11

The 5-3 move - 13/8, 13/10 stands out as interesting. We saw in
category one the importance of making points in our inner board to
contain our opponents back men. Is the three point really that much
different in value that this move is actually better than 8/3, 6/3?

According to a Kit Woolsey post I found in the FAQ, it's not. He says
that "the simple 8/3, 6/3 is clearly best. The once common 13/10,
13/8 has been found vastly inferior." Assuming Woolsey is right and
that 8/3, 6/3 is the standard opening here, does anyone have any
insight as to why 13/10, 13/8 was once thought to be superior?

Woolsey's post also suggests that 24/20, 13/8 might be a slightly
better 5-4 than 13/8, 13/9, that 24/21, 13/11 is a slightly better 3-2
than 13/10, 13/11, that 24/22, 13/8 is a slightly better 5-2 than
13/11, 13/8, and that 24/20, 13/10 is about equal with 13/10, 13/9 for
a roll of 4-3.

It appears, then (provided Woolsey's assertions are correct), that
splitting one's back men and bringing one builder has gained favor
over bringing two builders. What has happened to change the
conventional wisdom? Splitting seems to be conducive to a running game
- are these simply preferred style of play decisions?

Category 3: Running with One Back Man
Roll Recommended Move
6/5     24/13
6/4     24/14
6/2     24/16
6/3     24/15

Woolsey agrees with 24/13 for 6/5, and doesn't argue much with the
running plays for rolls of 6/3 and 6/4 (although he does mention that
the splitting plays are slightly better). However, he finds 24/16 to be
objectionable for 6-2, arguing that running 8 pips isn't far enough,
and suggests 24/18, 13/11 in its place. What is it about running 9, 10,
or 11 pips that makes this an acceptable move, while running 8 pips is

Woolsey also posits that 8/2, 6/2 isn't far behind 24/14 and 24/18,
13/9 for a roll of 6-4.  Is the two point simply too deep to make this
the default play? If so, why is it dramatically different than the
three point?

Category 4: Creating a Builder and Splitting Your Back Men
Roll Recommended Move
4-1     13/9, 24/23
2-1     13/11, 24/23
5-1     13/8, 24/23

Slotting the five is the only other serious contender for any opening
roll that contains a one, according to Woolsey. Which do you prefer,
and why?

Category 5: Rolling Doubles
Roll Recommended Move
6-6     24/18 [2], 13/7 [2]
5-5     13/3 [2]
4-4     24/20 [2], 13/9 [2]
3-3     24/21 [2], 8/5 [2]
2-2     24/20 [2] or 13/11 [2], 6/4 [2]
1-1     8/7 [2], 6/5 [2]

Doubles on the opening roll seems impossible, but it may be worth
discussing the possibilities for when you roll doubles after your
opponent has already moved.

Woolsey doesn't touch these rolls in his post, but I, for one, was
surprised to see 24/20 [2] recommended over 13/11 [2], 6/4 [2]. The
recommendations for the other rolls place a lot of emphasis on making
key points (such as the 4) and establishing builders. Why does making
your opponent's 5 trump these important concepts?

I appreciate everyone's discussion and input, and hope that all of
this isn't too basic/old hat for anyone. I am not always available to
read and respond quickly, but I will be keeping tabs on this and should
be able to respond to anything within a couple of days. Depending on
how well-received this is, we can continue on to other chapters of

-Hayden Alfano

Raccoon  writes:

> I am currently reading Magriel's "Backgammon," and Chapter Five
> is about opening rolls.

Note the second paragraph in this chapter: "For the more experienced
player, alternate opening moves are possible and may be adapted
according to the particular position. These alternate openings will be
discussed at the beginning of section III."

The "alternate openings" presented in section III are:

(a) slotting with 21 41 and 51
(b) splitting with 54 43 and 32
(c) splitting with 64 63 and 62

(Magriel: "few experts would play 24/16" -- the move presented in
chapter 5)

The second paragraph of chapter 5 continues: "The plays indicated here,
however, may be used with confidence since they are in no way inferior
to the alternates discussed later."

If we substitute "about as good as" for "in no way inferior to," then
thirty years of analysis and data from backgammon computer programs
tells us that Magriel's presentation of the opening moves in Chapter 5
and 15 was essentially correct, with the exception of his
recommendation for opening 53 and his complete omission of the third
alternate play of 8/2 6/2 for opening 64.

The bots, with their precise (but not necessarily "100 certain")
breakdowns of wins, gammons, and gammon losses, have been particularly
helpful in refining our understanding of match play theory (poorly
developed in pre-'Backgammon" years) and of adjusting -- picking and
choosing between the plays recommended in Chapters 5 and 15 -- for
particular match scores.

> According to a Kit Woolsey post I found in the FAQ, it's not. He says
> that "the simple 8/3, 6/3 is clearly best. The once common 13/10,
> 13/8 has been found vastly inferior." Assuming Woolsey is right and
> that 8/3, 6/3 is the standard opening here, does anyone have any
> insight as to why 13/10, 13/8 was once thought to be superior?

Magriel discusses that in the paragraph beginning: "5-3 could also be
used to make the three point."

See also the thread "5-3 opening roll" beginning
July 8 1997. In that thread Fredrik Dahl points out that while
JellyFish rollouts showed that "making the 3 pt is clearly better,
human experts discovered this too, before programs were strong enough
to be trusted."

In the second post in that thread I reviewed the history of the opening
53 play in the backgammon literature, discussed the reasoning behind
13/8 13/10, and the modern discovery, or rediscovery, that "any point
in the home board, even the 3 or 2 point, can be valuable in an early
exchange of blot hitting."

(Today I would amend those comments to acknowledge that moving one
spare from the 13 point to the 8 point is indeed a small improvement.)

That thread can be accessed at:

Yet another discussion of opening 53, from 1996, can be accessed at:

> Category 5: Rolling Doubles
> Roll       Recommended Move

A more thorough discussion of playing doublets in response to opening
rolls might have been useful, but this short section does present the
"standard" plays, with the exception of 33 and half-exception of 22.

> Woolsey doesn't touch these rolls in his post, but I, for one, was
> surprised to see 24/20 [2] recommended over 13/11 [2], 6/4 [2].

Well, Magriel writes: "2-2.: Close the twenty point. Another strong
play would be to close the eleven and four points." He might better
have written: ""2-2.: Close the eleven and four points. Another strong
play would be to close the twenty point." That said, 24/20(2) is indeed
sometimes the best way to play 22.

> The
> recommendations for the other rolls place a lot of emphasis on making
> key points (such as the 4) and establishing builders. Why does making
> your opponent's 5 trump these important concepts?

Magriel discusses that topic in Chapter  20, "The Golden Point."

Adam Tansley  writes:

Hi Hayden,

Magriel's Backgammon was written before the advent of neural-net (n-n)
programs such as jellyfish, snowie, gnu etc.  Despite the fact that the
book set a new benchmark for backgammon literature, it was inevitable
that once the n-ns got their hands on it that they would have
corrections to make.

"Category 1: Closing Strategic Points"

In addition to the three opening rolls you list, 53 should certainly be
used to make the 3 point. Making this point is almost as valuable as
making the 7 point.

The n-ns are even persuading modern players to make the 2 point with
64.  Magriel's objection was that the 8 and 2 points cannot both be
used to make a prime, and that making the 2 point is not flexible.
True enough, but endless rollouts have shown that the extra gammons
that come with making the 2 point compensate.

"Category 2: Bringing Checkers from the Midpoint as Builders"

With the exception of 53, the rolls you list are all candidates for
bringing down 2 builders, although bringing 1 down and splitting is
usually as good or better.

"Splitting seems to be conducive to a running game".  Actually the
point of splitting is not to run, but to aim for a decent anchor.
Deciding whether to split or bring 2 down is really a matter of style,
with rollouts showing that both plays are very close. Bringing 2 down
is more likely to lead to prime v prime games - it creates more point
making combinations, as well as discouraging the opponent from
splitting too far.  Splitting is more likely to lead to holding games
of one sort or another, as well as blitzes.

"Category 3: Running with One Back Man"

65 is the only concrete running number in the list, with 64 being the
only other real contender.

The problem with 62, 63 and 64 as running numbers is that if you're
hit, you've achieved nothing with your opening roll, and even if you're
not hit you still have work to do to safety the runner.
Generally a good a anchor is worth more than getting 1 back man to the
mid-point.  An anchor is a lasting asset, whereas running a man is not.
For this reason the split is generally preferred to running.

"Category 4: Creating a Builder and Splitting Your Back Men"

As I've said, you may well add 62, 63, 64 to this list.
"Slotting the five is the only other serious contender for any opening
roll that contains a one"

Rollouts tell us that for 21 slotting and splitting are pretty much a
dead heat, with n-ns slightly preferring the split for 41 and 51.  41
has as many point making combinations as 21 (although not all make the
5 point), but 51 has the defect of fewer good subsequent rolls.

Again, this is essentially a matter of style.  Splitting is likely to
lead to anchors and hence some sort of holding game.  Slotting is a
more aggressive attempt to dominate by making early key points, but
with real possibilities of complex positions following a blot hitting

Most strong players favour the slot, hoping to either dominate or
out-play their opponent when things get complex.  If you're weaker than
your opponent perhaps you should lean towards splitting.

"Category 5: Rolling Doubles"

Not much to add.

"Why does making your opponent's 5 trump these important concepts?"
There's a lot to say about the 5 point. Making it is a top priority in
the opening.  If you can make your opponent's 5 point you nullify much
of her blocking potential and vastly reduce the number of gammons you
lose.  This means you can play more aggressivley, and it puts pressure
on your opponent to equalise by making an anchor of her own.

Match play bring more complications into the equation.  Sometimes
gammons are more important than usual, sometimes you particularly want
to avoid being gammoned etc.  Worry about that later...


Stick  writes:

No contest opening rolls:
3-1 ... make the 5 pt.
4-2 ... make the 4 pt.
6-1 ... make the bar pt.
6-5 ... run with 24/13
5-3 ... make the 3 pt.

Back in the day people played two down with an opening 5-3 because they
realized the importance of the 4 and 5 pts. and wanted to bring more
builders down to make these points.  What they didn't seem to grasp was
the power of the 3 pt., or any point for that matter.  The 3 pt. was
seen as too deep but in reality making the 3 pt. is the correct play by
a huge margin.  It should be noted that two down is the #4 play for an
opening 5-3, with the split of 24/21 13/8 being ranked #2 and even the
straight run of 24/16 being better than semi stripping the midpoint and
not moving your back men by bringing two down.

For me 6-3 and 6-2 can be added to the list of 'no second choice'.  The
bots have shown extensive rollouts where the commital running plays
simply don't gain enough even when they work so everyone nowadays plays
24/18 13/10 and 24/18 13/11 respectively.

I also believe that slotting with an opening 2-1 to be 'correct'.  I
have extensive rollouts on my site,, that back
this up.  Aside from simply being a bot lover I've noticed the power of
not only going out of your way to make your own 5 pt. in the early
going, but having good sixes to play and unstacking in the opening.
Slotting with an opening 2-1 accomplishes all of these things, enough
good things I cannot say about this slot and will play it at all

For money I have also come to believe that an opening 4-3 should be
played two down from the midpoint, 13/9 13/10.  The splits aren't far
behind but bringing two down creates soooooo many point making numbers
if you aren't hit (powerful points too!) that it's worth semi-stripping
the mid.  On the other hand with an opening 3-2 I don't play two down
but play the larger split, 24/21 13/11 because all your ammo isn't
aimed quite as concisely at the points you want.  Even so, I wouldn't
mind playing two down with an opening 3-2 either, matter of preference
I believe.

Slotting with an opening 4-1 is better than slotting with an opening
5-1 and after playing thousands of games playing both I'm confident of
this.  That's not to say the minor split with either roll is worse,
they're about equal, but with a 4-1 I prefer the slot still, 5-1 I
split.  The reason I don't slot with an opening 5-1 is that in the
ensuing battle/struggle for your 5 pt. you have no new builders if
you're hit and you need to hit back whereas with the 2-1 or 4-1 you
have an extra builder, an extra number, to help you in that battle.

The only roll that the verdict is up in the air for me still is an
opening 6-4.  I feel if you're the weaker player or it's DMP you should
simply run with 24/14.  If you're in need of a gammon the obvious 8/2
6/2 making a point is justifiable.  I prefer the split though extensive
rollouts show them all too close to call.  The split creates a more
complicated game in general and doesn't commit you to a certain type of
game the way making the two point does.

Lastly, 5-2, 5-4 I believe should both be played by splitting.  By
bringing two down you semi strip the mid for too little of a reward.
You don't create that many new point making rolls to make it worth not
splitting the back men.  The rollouts show a preference for the splits
too, but the preference isn't overwhelming.

I believe the next generation of bots will solidify my thoughts even
further on opening rolls and the first three moves of the game are my
specialty I just wish I knew what to do after that... =))

Did you find the information in this article useful?          

Do you have any comments you'd like to add?     


Opening Rolls

At different match scores  (Louis Nardy Pillards, July 2002) 
Average advantage of winning opening roll  (Chuck Bower, Oct 1998) 
Choosing a strategy  (Daniel Murphy, June 2001) 
Early game rule of thumb  (Rich Munitz, Feb 2009) 
Factors to consider  (Kit Woolsey, July 1994) 
How computers play  (Kit Woolsey, Mar 1995)  [Recommended reading]
Magriel's Chapter 5  (Hayden Alfano+, May 2006)  [Long message]
Mloner vs Jellyfish  (Kit Woolsey, Dec 1995) 
Nactating a whole game  (Nack Ballard+, Jan 2011)  [Long message]
Nactation  (Jim Stutz+, June 2010) 
Nactation overview  (Nack Ballard, Oct 2009) 
Nactation--Why use it?  (leobueno+, Jan 2011) 
Opening 1's: Split or slot?  (Douglas Zare, Dec 2003) 
Opening 21: Rollout  (Stick, Mar 2006)  [GammOnLine forum]
Opening 21: Split or slot?  (Dick Adams+, Dec 2003) 
Opening 32: Rollout  (Stick, Feb 2006)  [GammOnLine forum]
Opening 43: In GOL online match  (Raccoon+, Feb 2004)  [GammOnLine forum]
Opening 43: Pros and cons  (Stick+, Jan 2006)  [GammOnLine forum]
Opening 43: Which split is better?  (Peter Backgren+, Aug 2000) 
Opening 43: Which split is better?  (Michael J. Zehr+, Mar 1996) 
Opening 51: Rollout  (Stick, Feb 2006)  [GammOnLine forum]
Opening 52: Merits of splitting  (Peter Bell, Apr 1995) 
Opening 53: Magriel's recommendation  (George Parker+, July 1997)  [Long message]
Opening 53: Split to 21?  (Alex Zamanian, Aug 2000) 
Opening 53: Why make the three point?  (Kit Woolsey+, Feb 1996) 
Opening 6's: Slot the bar point?  (Chuck Bower+, Feb 2000) 
Opening 6's: Slot the bar point?  (David Montgomery, June 1995) 
Opening 62: Could running be best?  (Gary Wong, Sept 1997) 
Opening 62: Split, run, or slot?  (Chuck Bower, May 1997) 
Opening 63: Middle Eastern split?  (Mark+, Apr 2002) 
Opening 63: Slot the four point?  (Dennis Cartwright+, Mar 2002) 
Opening 64: Make the two point?  (William Hill+, Jan 1998) 
Opening 64: Make the two point?  (Darse Billings, Feb 1995) 
Opening 64: Rollout  (Peter Grotrian, Jan 2006)  [GammOnLine forum]
Opening 64: Split to 20?  (Peter Bell, June 1995) 
Opening 64: Three choices  (Brian Sheppard, July 1997) 
Opening 65: Becker on lover's leap  (Jeffrey Spiegler+, Aug 1991) 
Opening 65: Computer rankings  (Chuck Bower, Jan 1997) 
Opening rolls ranked  (Arthur+, Apr 2005) 
Rollouts of opening 21 and replies  (Alexander Nitschke, Oct 1997) 
Rollouts of openings  (Tom Keith+, Jan 2006) 
Rollouts: Expert Backgammon  (Tom Fahland, Aug 1994) 
Rollouts: Jellyfish 3.0  (Midas+, Sept 1997) 
Rollouts: Jellyfish 3.0 level 6  (Chuck Bower, Feb 1999)  [Recommended reading]
Rollouts: Snowie 4.1  (Rene Cerutti, Apr 2004) 
Slotting the four point  (Joe Loria+, Oct 1999) 
Snowie's openers and replies  (rcerutti, Feb 1999)  [Long message]
Splitting versus building  (Dave Slayton+, Aug 2000) 
Splitting versus slotting  (Daniel Murphy, Apr 2001) 
Splitting versus slotting  (Daniel Murphy, Sept 1997) 
Trice's rankings  (Marty Storer, Feb 1992) 

[GammOnLine forum]  From GammOnLine       [Long message]  Long message       [Recommended reading]  Recommended reading       [Recent addition]  Recent addition

  Book Suggestions
Computer Dice
Cube Handling
Cube Handling in Races
Extreme Gammon
Fun and frustration
GNU Backgammon
Luck versus Skill
Magazines & E-zines
Match Archives
Match Equities
Match Play
Match Play at 2-away/2-away
Opening Rolls
Pip Counting
Play Sites
Probability and Statistics
Source Code
Strategy--Bearing Off
Strategy--Checker play


Return to:  Backgammon Galore : Forum Archive Main Page