*** rec.gambling.craps FAQ ***
- From: "SteveF" <stevefry@xxxxxxxxxxxxxx>
- Date: Mon, 2 Jul 2007 10:13:17 -0800
-------------------------------------------------------------------------------
Frequently Asked Questions about Craps
This is the rec.gambling.craps Frequently Asked Questions (FAQ) list.
It should be viewed with a FIXED WIDTH font (such as Courier) in order to
appreciate the diagrams and graphs.
Changes or additions to this FAQ should be submitted to:
stevefry@xxxxxxxxxxxxxx
Page last modified: 7-2-07
-------------------------------------------------------------------------------
Table of Contents
Q01 How is Craps played?
Q02 What special terminology is used at the Craps table?
Q03 What are "Odds"? Should you bet them?
Q04 What is the house advantage on 10x odds?
Q05 What are "Come" and "Don't Come" bets?
Q06 What are all those other bets?
Q07 What are the odds for all these bets?
Q08 What is the "fire" bet?
Q09 What is a good book to teach me the basics about Craps?
Q10 What is a sim? How can I run one?
Q11 Does the house make money on the "odds" bets?
Q12 The house is favored to win the "odds" bet; why don't they make money?
Q13 What is the "Gambler's Fallacy"?
Q14 How long is the "long term"?
Q15 Can one exploit trends to improve one's chances?
Q17 What is "dice setting"? Does it work?
Q18 But does SSDD dice setting really work?
Q19 How can SSDD give the shooter an advantage?
Q20 How does the Craps NewsGroup (rec.gambling.craps) work?
-------------------------------------------------------------------------------
Q01: How is Craps played?
A01: (Dave Decot, Frank Irwin, Alan Mintz, Ken Elliott III, Jeffrey Osier)
Casino craps is played completely against the casino, on a big felt layout set
inside a large table with high rails around the side so you can bounce the dice
off 'em and still keep 'em on the table. Up to between 12 and 16 people can
play at once, depending on the size of the table (obviously). The layout looks
something like:
Dealer stands Boxman sits Dealer stands
here here here
#######################################################################
# __ ____________________ $$$$$$$$$$$$$ ____________________ __ #
# | d||DC| 4| 5| 6| 8| 9|10| +---------------+ | 4| 5| 6| 8| 9|10|DC||d | #
# |Po||__|__|__|__|__|__|__| |any seven 4-1| |__|__|__|__|__|__|__||oP| #
# |An| ____________________ +===============+ ____________________ |nA| #
# |St|| C O M E | |hard 4 | hard 6| | C O M E ||tS| #
# |Sp||____________________| |-------+-------| |____________________||pS| #
# |La| ____________________ |hard 10| hard 8| ____________________ |aL| #
# |Is||2 3 4 9 10 11 12| +===============+ |2 3 4 9 10 11 12||sI| #
# |Ns||_______FIELD________| |two | three| |_______FIELD________||sN| #
# |E |_____________________ |-----HORN------| _____________________| E| #
# | don't pass bar 12 | |eleven | twelve| | don't pass bar 12 | #
# \________PASS_LINE______| +===============+ |________PASS_LINE______/ #
# | any craps | #
# +---------------+ #
# /dice\ #
#####################################################################
Stickman stands
here
Each player bets a minimum amount determined by the table.
Each die has six sides, each side with a different number of spots from one to
six.
Two such dice are rolled by one player called the "shooter". The shooter must
place a "pass" bet or a "don't pass" bet in order to be eligible to roll the
dice. Exception: the shooter can let his Hot Babe (TM) roll the dice for him if
he has a pass or don't pass bet down.
The total number of spots on the tops of the dice after the shooter has rolled
is called the "roll".
A game consists of a series of rolls.
A roll of 2, 3, or 12 is called "craps".
The first roll by the shooter during a game is called the "come-out roll".
If the come-out roll is 7 or 11, the game is over:
Bets on the "Pass line" win 1:1.
Bets on the "Don't Pass line" lose.
If the come-out roll is craps, the game is over:
Bets on the "Pass line" lose.
Bets on the "Don't Pass line" win unless:
The "Don't Pass" line says "Bar " and the roll is the indicated value, in
which case the bet pushes.
Otherwise, the come-out roll becomes the "point", and a large white marker is
placed on the number representing the point (4, 5, 6, 8, 9, or 10).
For each roll in a game subsequent to the come-out roll:
If the roll is the point, the game is over:
Bets on the "Pass line" win 1:1.
Bets on the "Don't Pass line" lose.
If the roll is 7, the game is over:
Bets on the "Pass line" and lose.
Bets on the "Don't Pass line" win 1:1.
The turn of the "shooter" is over.
Otherwise, the game continues and the shooter rolls again.
During a game, bets on the Pass line cannot be removed; they can, however, be
increased. Bets on the Don't Pass line may be decreased or removed, but not
increased.
When a game is over:
If the game was over on the come-out roll, or because the point was rolled
again, the shooter may continue to be the shooter for another game, or
pass the dice on to the player just clockwise, who becomes the new
shooter.
Otherwise, the shooter must pass the dice on to the player just clockwise,
who becomes the new shooter.
Note: The other bets that can be made and resolved are not detailed above for
purposes of saving space. A description of when these bets win or lose is given
in question Q02.
-------------------------------------------------------------------------------
Q02: What special terminology is used at the Craps table?
A02: (Steve Jacobs, Ken Elliott III, Dave Everett)
Craps Terminology:
any craps
a bet that the next roll will be 2, 3, or 12. This bet pays 7:1 and has a
house edge of 11.1%.
3-way craps
a bet made in units of 3 with one unit on 2, one unit on 3, and one unit
on 12. This is a horn bet without the bet on 11.
any seven
a bet that the next roll will be 7. This bet pays 4:1 and has a house edge
of 16.7%.
big 6
a bet that a 6 will be rolled before a 7 comes up. This bet pays even
money, and has a house edge of 9.1%. A place bet on 6 pays 7:6 but is
identical otherwise. The place bet is preferred, having a house edge of
1.5%
big 8
a bet that an 8 will be rolled before a 7 comes up. This bet pays even
money, and has a house edge of 9.1%. A place bet on 8 pays 7:6 but is
identical otherwise. The place bet is preferred, having a house edge of
1.5%
buy bet
giving the house a 5% commission in order to be paid correct odds for a
place bet. The buy bets on 4 and 10 allow the player to reduce the house
edge from 6.67% to 4% on these bets. Some casinos collect the commission
only on winning bets, while others collect it at the time the bet is made.
come bet
A "virtual pass line bet"; a bet made after the come-out roll but in other
respects exactly like a pass line bet. See question C5 for more details.
come-out roll
the first roll of the dice in a betting round is called the "come-out"
roll. Pass bets win when the come-out roll is 7 or 11, while pass bets
lose when the come-out roll is 2, 3, or 12. Don't bets lose when the come-
out roll is 7 or 11, and don't bets win when the come-out roll is 2 or 3.
Don't bets tie when the come-out roll is 12 (2 in some casinos; the "bar"
roll on the layout indicates which roll is treated as a tie).
dice pass
The dice are said to "pass" when the shooter rolls a 7 or 11 on the
come-out roll. The dice "don't pass" when the shooter rolls a 2, 3, or 12
on the come-out. If the come-out roll is a 4, 5, 6, 8, 9, or 10, this roll
sets the "point", and the shooter continues to roll until the point is
rolled again or a 7 is rolled (see "seven out"). If the shooter rolls the
point before rolling a seven, the dice pass. If the shooter sevens out,
the dice don't pass and the shooter loses control of the dice. NOTE: in
this context, "pass" does NOT mean that the dice to given to the next
player. Control of the dice is transferred only when the shooter "sevens
out" or when the shooter has completed a game and no longer wishes to roll
the dice.
don't come bet
A "virtual don't pass bet"; a bet made after the come-out roll but in
other respects exactly like a don't pass bet. See question C5 for more
details.
don't pass bet
a bet that the dice will not pass. This bet can be placed only immediately
before a "come-out" roll. One result (either the 2 or the 12, depending on
the casino) will result in a push. House edge on these bets is 1.40%. A
don't pass bet can be taken down, but not increased, after the come-out
roll.
double odds
an odds bet that is about twice as large as the original pass/come bet.
Some casinos offer higher odds, such as 5X or even 10X odds.
field bet
a bet that the next roll will be 2, 3, 4, 9, 10, 11, or 12. This bet pays
even money for 3, 4, 9, 10, and 11, and usually pays 2:1 for 2 or 12. Some
casinos pay 3:1 for either the 2 or 12 (but not both), and some casinos
may make the 5 instead of the 9 a field roll.
hard way
a bet on 4, 6, 8, or 10 that wins only if the dice show the same face;
e.g., "hard 8" occurs when each die shows a four.
hop bet
a bet that the next roll will result in one particular combination of the
dice, such as 2-2 (called a "hopping hardway") or 3-5. 2-2, 3-3, 4-4, and
5-5 are paid the same as a one-roll 2; other hop bets are paid the same as
a one-roll 11.
horn bet
a bet that the next roll will be 2, 3, 11, or 12, made in multiples of 4,
with one unit on each of the numbers.
horn high bet
a bet made in multiples of 5 with one unit on 3 of the horn numbers, and
two units on the "high" number; e.g., "$5 horn high eleven": $1 each on 2,
3, 12, and $2 on the 11.
lay bet
a bet that a particular number (4,5,6,8,9, or 10) will NOT be rolled
before a 7 comes up. The casino takes 5% of the winnings on these bets.
The 5% commission is usually taken up front, but some casinos take the
commission after the bet wins.
lay odds
after a point has been established, the don't pass bettor can place an
additional odds bet that will win if the original don't pass bet wins. The
odds bet is paid at the correct odds for the point, and is a fair bet with
no house edge. This also applies to a don't come bet. Making this bets is
referred to "laying the odds" for your don't bet.
line bet
a bet on the "pass line" or the "don't pass line" is called a "line" bet.
These bets are placed at the beginning of the game, before the "come-out"
roll. The shooter is required to make a line bet in order to shoot the
dice.
odds off
odds bets that are "not working". Odds bets can be called "off" by the
player at any time, but are left on the felt until the bet is resolved.
Also, come odds bets are usually "off" during the come-out roll, unless
the bettor asks to have the odds bets "working". Come odd bets that are
"off" will be returned to the player if the line bet loses on the come-out
roll. Don't come odds generally work on the come-out roll.
pass bet
a bet that the dice will pass, also known as a "pass line" bet. This bet
is generally placed immediately before a "come-out" roll, although you can
make or increase this bet at any time. House edge on this bets is 1.41%.
place bet (to win)
a bet that a particular number (4, 5, 6, 8, 9, or 10) will be rolled
before a 7 comes up. These bets are paid at slightly less than correct
odds, giving the house an edge of 1.52% on 6/8, 4% on 5/9, and 6.67% on
4/10.
place bet (to lose)
a bet that a 7 will be rolled before the number you are placing
(4,5,6,8,9, or 10) comes up. The casino requires you to lay slightly more
than the correct odds, giving the house an edge of 3.03% on 4/10, 2.5% on
5/9, and 1.82% on 6/8.
point
if a 4, 5, 6, 8, 9, or 10 is rolled on the come-out roll, then this number
becomes the "point". The shooter must roll the point again, before rolling
a seven, in order for the dice to "pass". A "come point" is just the
number that is serving as a point for a come bet.
put bet
1. A bet made on the pass line after the come-out roll. This is allowed in
Las Vegas and at Turning Stone, but not in Atlantic City and not at
Foxwoods. This is not recommended, as 45% of your pass line wins are made
on the come-out roll. 2. A bet made directly onto a come point number.
E.g., "Put $5 and $10 odds on the six." Not recommended for the same
reasons given in 1.
right bettor
a player who bets that the dice will pass.
seven out
when the shooter rolls seven after a point has been established. Control
of the dice is transferred to the next shooter. Another term for this is
"miss out." You will sometimes hear players call this something else, but
we can't print those things here. This is often incorrectly called "crap
out."
shooter
the player who is rolling the dice. The shooter must place a "line" bet
("pass" or "don't pass") in order to be eligible to roll the dice. Of
course, the shooter can place other bets in addition to the required
"line" bet. Most shooters (and players) tend to play the "pass" line. Note
that shooters who make "don't pass" bets are not betting against
themselves, they are simply betting that the dice will not "pass".
single odds
an odds bet that is about as large as the original pass/come bet. Some
casinos allow "double odds", or even larger odds bets.
take odds
after a point has been established, the pass/come bettor can place an
additional odds bet that will win if the original pass/come bet wins. The
odds bet is paid at the correct odds for the point, and is a fair bet with
no house edge.
two ways
a phrase appended to a hardway or proposition bet to indicate that the
player is betting one chip for the dealers along with his own bet. A $2
bet two ways is $1 for the player and $1 for the dealers; a $6 bet two
ways is $5 for the player and $1 for the dealers; a $10 bet two ways is $5
for the player and $5 for the dealers. E.g., "Hard 6, two ways" or
"Two-way hard 6."
working
bets that are "live" (i.e., can be resolved with the next roll) are said
to be working. Generally, place bets, buy/lay bets, and come odds bets do
not work on the come-out unless you tell the dealers to "make them work."
All other bets (e.g., hardways) work unless you call them "off" (i.e.,
tell the dealers you do not want them to "work").
world bet
a bet that the next roll will be 2, 3, 7, 11, or 12, made in multiples of
5, with one unit on each of the numbers.
wrong bettor
a player who bets that the dice will not pass.
-------------------------------------------------------------------------------
Q03: What are "Odds"? Should you bet them?
A03: (Ken Elliott III, Alan Shank, Steve Fry)
Casinos allow a player to place "odds" on pass, don't pass, come, and don't
come bets after a "point" has been established. If the bet on which odds are
placed wins, the odds bet is paid fairly. This means the odds on pass and come
bets are paid 2-1 for the 4 and 10, 3-2 for the 5 and 9, and 6-5 for the 6 and
8. The odds for don't pass and don't come bets are paid 1-2 for the 4 and 10,
2-3 for the 5 and 9, and 5-6 for the 6 and 8 (this is called "laying" odds).
The player should make odds bets that can be paid exactly, or the dealer will
pay off by rounding down, called "breakage". Odds on the pass line and come
bets should be a multiple of 5 if the point is 6 or 8 and a multiple of 2 if
it's a 5 or 9. For a $5 bet example, see the following:
If you are betting $5 Pass or Come and your point is:
* 6/8 place $5 to win $6, $10 to win $12, $15 to win $18, etc.
* 5/9 place $6 to win $9 (single odds), $10 to win $15, etc.
* 4/10 place $5 to win $10 (single odds), $10 to win $20, etc.
If you are betting $5 Don't Pass or Don't Come and your point is:
* 6/8 lay $6 to win $5, $12 to win $10, $18 to win $15, etc. You bet
in units of $6 and your winnings in units of $5. The limit on odds
bets is compared to the amount you can WIN, so double odds would be
$12 to win $10, etc.
* 5/9 lay $9 to win $6 (single odds), $15 to win $10, etc. The ratio is
bet 3 to win 2.
* 4/10 lay $10 to win $5 (single odds), $20 to win $10, etc. The ratio
is bet 2 to win 1
Casinos advertise the maximum odds bets they allow as the maximum amount
"times" the original bet the odds bet may be (for don't pass and don't come
bets, it's the maximum amount "times" the expected win). You can increase your
odds bet over this advertised maximum only enough to allow you to make an odds
bet that can be paid exactly.
A player can modify his odds bets at any time. That is because the bet is a
fair bet, with no advantage to either player or house; it pays off exactly
inversely to the odds of the bet winning. For example, when placing an odds
bet on a point of 4, the odds are 1 to 2 that you will lose but you get paid
2 to 1 when you win.
When you take odds you do not change the expected value of the flat bet, and
the expected value of the odds bets is always zero, since the casinos pays
true odds on those bets. So, the question is not about expected value at all,
but about volatility. Taking odds adds considerable volatility, which means
you can expect wider bankroll swings, either way. If your luck is good you
will win more and if it's bad you will lose more. That goes for pass, come,
don't pass and don't come bets. Higher volatility also increases the
probability that you will break even or better, at the cost of higher risk.
-------------------------------------------------------------------------------
Q04: What is the house advantage on 10x odds?
A04: (Steve Fry)
First one must understand what is meant by "house advantage", the negative of
which is the "expected value" (commonly known as EV) of a bet. There are two
types of EV, calculated as the expected return on a bet (number of units won or
lost) divided by the initial amount of the bet (called EVI), and an alternate
type is calculated as the expected return divided by the total amount placed at
risk during the course of the resolution of the bet (called EVR.) In addition,
some types of EV consider a "tie" resolution as a non-event (called EVI-NT or
EVR-NT for no ties) and others consider the initial bet as money at risk
(called EVI or EVR.)
The convention with BlackJack players and literature is to use EVI as the
measure of EV. This means that only the initial bet is used in the denominator
of the EVI and not any additional bets on double down or splitting. It also
means that the initial bet is considered at risk even if it results in a push
with the dealer or one split hand wins and one loses.
The convention with most Craps literature is to use EVR-NT when speaking of
"EV". This means that additional free odds bets are included as money at risk,
even though there is no advantage with these bets. Also it means that when the
don't-pass player catches the bar number, his action is not counted as money at
risk.
The following table lists the house advantage for a pass bet and don't pass bet
with the indicated amount of odds. The different types of EV as described
above are shown above the column it represents.
HOUSE ADVANTAGE
Pass bet Pass bet Don't Pass Don't Pass Don't Pass
Odds (-EVI) (-EVR) (-EVI) (-EVR-NT) (-EVR)
------ ---------- ---------- ---------- ---------- ----------
0x 1.4141% 1.4141% 1.4026% 1.4026% 1.3636%
1x 1.4141% 0.8485% 1.4026% 0.6914% 0.6818%
2x 1.4141% 0.6061% 1.4026% 0.4588% 0.4545%
3x 1.4141% 0.4714% 1.4026% 0.3433% 0.3409%
4x 1.4141% 0.3857% 1.4026% 0.2743% 0.2727%
5x 1.4141% 0.3263% 1.4026% 0.2283% 0.2273%
6x 1.4141% 0.2828% 1.4026% 0.1956% 0.1948%
7x 1.4141% 0.2496% 1.4026% 0.1710% 0.1705%
8x 1.4141% 0.2233% 1.4026% 0.1520% 0.1515%
9x 1.4141% 0.2020% 1.4026% 0.1367% 0.1364%
10x 1.4141% 0.1845% 1.4026% 0.1243% 0.1240%
100x 1.4141% 0.0209% 1.4026% 0.0135% 0.0135%
-------------------------------------------------------------------------------
Q05: What are "Come" and "Don't Come" bets?
A05: (Dave Decot, Frank Irwin, Alan Mintz, Ken Elliott III, Steve Fry)
Other bets can be made during the game after the come-out roll by anyone,
called "Come" and "Don't Come" bets. These are made by placing the bet on the
"Come" box or the "Don't Come" box; these bets are regarded as Pass (Don't
Pass) bets, but as if the very next roll of the dice were the "come-out" roll
of a new game. For example, if a come bet is made and if the next roll is 7 or
11, the Come bet wins immediately; if the next roll is 2, 3, or 12, the Come
bet loses immediately; otherwise, the number rolled is the point for that Come
bet (called a come point). Such a Come bet is moved onto the area of the table
where its point appears, awaiting a roll of either its point or seven. The game
for a Come bet always continues until this happens, even though the shooter
rolls the point for the Pass line, even though the shooter begins a new game
for the Pass line, even though another shooter begins rolling, as long as the
termination conditions for that Come bet have not yet occurred.
Note that rolling a seven always terminates all Pass, Come, Don't Pass, and
Don't Come games on the table; since it results in immediate win or loss.
The payoffs for Come and Don't Come bets are the same as for Pass and Don't
pass bets.
It is possible to place odds bets on the points of your own Come and Don't Come
bets by handing the bet to a dealer and stating that you want "odds on my
number". Unless you specify otherwise, odds bets on Come are declared
"not working" on a come-out roll after a point is made. However, odds on Don't
Come bets are usually working by default.
The following table shows the House Advantage for the Come bet if the odds are
not working on a come-out roll.
HOUSE ADVANTAGE ON COME BET
Odds always working Not working on come-out
Odds (-EVR) (-EVR)
------ ---------- ----------
0x 1.4141% 1.3636%
1x 0.8485% 0.9107%
2x 0.6061% 0.6716%
3x 0.4714% 0.5320%
4x 0.3857% 0.4404%
5x 0.3263% 0.3757%
6x 0.2828% 0.3276%
7x 0.2496% 0.2904%
8x 0.2233% 0.2608%
9x 0.2020% 0.2367%
10x 0.1845% 0.2166%
100x 0.0209% 0.0251%
-------------------------------------------------------------------------------
Q06: What are all those other bets?
A06: (Dave Decot, Frank Irwin, Ken Elliott III, Steve Fry)
Other bets are possible:
Place bets (to win):
Bets that an indicated number will be rolled before 7 is rolled (come-out
rolls are ignored for the purpose of determining this, unless otherwise
specified by the player making the bet).
Place bets (to lose):
Bets that a 7 will be rolled before the indicated number is rolled
(come-out rolls are ignored for the purpose of determining this, unless
otherwise specified by the player making the bet).
One roll bets:
Bets that a certain roll, or a certain pair of dice faces, or one of
several rolls, will appear on the next roll of the dice. Such bets may be
made before any roll. These all pay higher than 1:1, This includes "Any
craps", "eleven", "seven", "Horn", and "field" bets.
Field:
A special case of one-roll bet. This pays 1-1 whenever 3, 4, 9, 10, or 11
is rolled, and 2-1 when 2 or 12 is rolled (sometimes 3-1 for a 2 or 12).
Midway:
A special case of one-roll bet. This pays 1-1 whenever 6, 7, or 8 is
rolled, and 2-1 for a hardway 6 or 8.
Hardways:
Bets that a certain pair of dice faces will appear before 7 is rolled, and
before any other pair of dice faces with the same total value are rolled.
For example, a bet on "hard 4 (2 and 2)" loses when (1 and 3) is rolled,
because this is an "easy way" to roll 4. A bet on "hard anything" loses
when 7 is rolled.
Hopping hardways:
Bets that a certain pair of identical dice faces will appear on the next
roll. These all pay 30:1 (or sometimes higher or lower).
Horn bets:
Basically, just betting on the 2,3,11, and 12 at once. This requires 4
units, since you are really making 4 bets.
-------------------------------------------------------------------------------
Q07: What are the odds for all these bets?
A07: (["Winning Casino Craps" by Edwin Silberstang], Ken Elliott III)
Note that some casinos offer different payouts for prop bets, lower
(promotional) commission on buy/lay bets, and collection of buy/lay bet
vigorish only on winning bets, all of which change the "Casino Advantage"
listed below. The numbers below represent those most commonly seen in Las
Vegas.
Bet Casino Payoff Casino Advantage
Pass-Line 1:1 1.41%
With Single Odds 1:1 + odds 0.8
With Double Odds 1:1 + odds 0.6
Come 1:1 1.41%
With Single Odds 1:1 + odds 0.8
With Double Odds 1:1 + odds 0.6
Don't Pass 1:1 1.40%
With Single Odds 1:1 + odds 0.8
With Double Odds 1:1 + odds 0.6
Don't Come 1:1 1.40%
With Single Odds 1:1 + odds 0.8
With Double Odds 1:1 + odds 0.6
Place Numbers (to win)
4 or 10 9:5 6.67%
4 or 10 (bought) 2:1 (-5% commission) 4.76
5 or 9 7:5 4.0
6 or 8 7:6 1.52
Place Numbers (to lose)
4 or 10 5:11 3.03%
4 or 10 (laid) 1:2 (-5% commission) 2.44
5 or 9 5:8 2.5
6 or 8 4:5 1.82
Big 6 and Big 8 1:1 9.09%
Field
With 2 and 12 paying 2:1 1:1 except 2 & 12 5.55%
With 2 pay 3:1, 12 pay 2:1 1:1 except 2 & 12 2.77%
With 2 pay 2:1, 12 pay 3:1 1:1 except 2 & 12 2.77%
Midway 1:1 except hard 6 & 8 5.55%
In practically all casinos, odds on proposition bets are quoted as "x for y",
which means that the casino takes your winning "x" bet and pays you "y", in
contrast to what is done for other winning bets (e.g., if the "Any 7" bet is "5
for 1", when making a $1 bet and winning the casino will take your $1 and give
you $5, for a "real" payoff of "4 *to* 1"). The numbers below are quoted as
"x:y", not "x for y".
Proposition Bets
Bet True Odds Casino Payoff Casino Advantage
Any 7 5:1 4:1 16.67%
Any Craps 8:1 7:1 11.1
2 or 12 35:1 30:1 13.89
29:1 16.67
3 or 11 17:1 15:1 11.1
14:1 16.67
Hardways
4 or 10 8:1 7:1 11.1%
6 or 8 10:1 9:1 9.09
-------------------------------------------------------------------------------
Q08: What is the "fire" bet?
A08: (Steve Fry)
The fire bet is a special multi-game bet that is made before the come-out roll
of a new shooter. If the shooter rolls points of 4, 5, 6, 8, 9, and 10, and
makes all six points (in any order) without a single seven-out then he wins a
large bonus. This bet was introduced in the 90's as a promotional gimmick to
increase interest in Craps and make it more exciting (seeing as how most Craps
tables root for the shooter to roll points and dread the seven out.) Back
then the shooter did not need to make any bet (and it wasn't called "fire"),
but if he made all six points he got paid $4000.
As currently offered in casinos, the "fire" bet requires one unit bet before
the first come-out roll, then when the shooter sevens out if he has made four,
five, or six out of all six points, he is paid according to a schedule. The
following table shows the probabilities and expected value of the fire bet
given two different payout schedules:
four points five points six points
------------- ------------- ------------- Expected Value
Probability: .00879537 .00164077 .000162784 ----------------
Units Paid: 10 200 2000 -25.8 %
Units Paid: 25 250 1000 -20.7 %
When the promotional bonus required no bet, this was a positive EV situation
for the shooter (probably why it isn't offered any more) but this "fire" bet
is one of the worst bets in the casino. Only the keno game is worse than this.
-------------------------------------------------------------------------------
Q09: What is a good book to teach me the basics about Craps?
A09: (Mason Clarke)
Download and install the shareware program WINCRAPS.
*
http://www.cloudcitysoftware.com/
*
Read the general information in the help file and use the game to further
understand the game in action.
-------------------------------------------------------------------------------
Q10: What is a sim? How can I run one?
A10: (Steve Fry)
"sim" is short for simulation -- the running of a computer program that
simulates what happens in real life, where one can save any desired information
about the state of the simulation. After the sim has been run, this
information can be analyzed in order to predict the actual outcome given the
same real inputs as those inputs fed into the simulation. There are two types
of simulation programs: those that attempt to model all relevant physical
parameters and processes to some high degree of accuracy, then use the computer
program to simulate how things change with time; then there are simulation
programs that model random processes in order to determine the probabilities of
certain events occurring. The first type of sim program relies on its accuracy
and completeness of data and the difference engine that simulates time; such a
program would be used to predict weather patterns or to design a nuclear bomb.
The second type of sim program relies mainly on its random number generator
(RNG) code to produce random outcomes; then it's just a matter of running the
sim with enough cases to yield a confident answer. Such a program is perfect
to simulate what happens at a Craps table; and the thing that is simulated is
a gambling session. A session is defined as the bettor starting with a given
bankroll amount and a precise description of the bets he will be making and the
criteria used to halt the session. The question before the bettor is: what is
the likely outcome of an "average" session? In order to answer this question
the program simulates many sessions, each a random series of games. The
outcome of each session is recorded in bins, where each bin accumulates the
number of sessions that ended with the amount designated for that bin; where
this amount is the number of betting units left from the starting bankroll.
For example, bin 1 would correspond to zero units (bankroll busted), bin 2
would represent one unit, etc. When the bins are plotted on the horizontal
axis with bars for each bin where the height of each bar represents the number
of sessions in that bin, one has a histogram plot of the Probability
Distribution Function (PDF). When normalized, this plot represents the
probability of achieving any given outcome. By analyzing the shape of this
curve one can gain great insight into which outcomes are more likely to occur.
By modifying the inputs (one's gambling behavior) a more enjoyable gambling
experience can be achieved (for example, one can get more successes and fewer
failures by reducing odds bets or by increasing the bankroll.) See the plots
in the answer to Q11 for examples of different PDFs when odds are used or not.
The best Craps simulator you can get is WINCRAPS, which is a shareware
program well worth the $24.95 cost of registration. You can get WINCRAPS at:
*
http://www.cloudcitysoftware.com/
*
Using the Hyper-Drive feature of the program and the statistics it provides
will be the easiest way for the average craps player to run a sim. Be sure
to read the help file to understand the required inputs.
There is another, much simpler, craps simulator called GAMBLSIM, which is
available for free at this location:
*
http://members.dslextreme.com/users/stevefry/gamblsim.htm
*
For the Fortran programmer, the source code for the simulator is also included
at the above site. The full version of GAMBLSIM (scroll to the bottom in the
above web site) was used for the simulator output in the answers to Q11 and Q12
below.
-------------------------------------------------------------------------------
Q11: Does the house make money on the "odds" bets?
A10: (Steve Fry)
No. The odds bet is called the "free odds" bet for a reason -- it's free. In
other words, it carries no vigorish, no house advantage. The house makes money
on the flat bet, but not the odds bet (see the tables in Q04 to see how the EVI
does not change with the addition of odds.) There are logical reasons why this
is so: since the payoff odds are inversely proportional to the winning odds,
the result is a draw. Then there are common sense reasons why odds are free:
the house lets you add and take away these bets as you wish while it will not
let you take away a bet that is in their advantage nor will they let you add a
bet that is to your advantage (however they will almost always allow the
reverse.) Also, since the laying of free odds is exactly the reverse of
placing the odds, if the house did have an advantage over the bettor placing
odds, then the bettor laying odds would also have a positive expectation, and
Don't players would be making a killing on this bet.
Failing logic and common sense, the computer simulator should always show that
the house will gain no money from the free odds bet. Using GAMBLSIM, let us
set up three scenarios. One scenario has the gambler playing pass line with
no odds, and his starting bankroll is 120 units (at a $5 minimum table, that
is a bankroll of $600.) The second scenario has the gambler playing pass line
and always placing 10x odds, and since his commensurate risk at 10x odds is
7.667 times the flat bet (because odds are only taken 2/3 of the time), his
bankroll is 920 units. The third scenario is the same gambler as in the second
(10x odds), except his bankroll is 120 units like in the first scenario (he is
under funded because of the increased risk.) All scenarios comprise sessions
that only end when bankrupt (gambler's ruin) or when 200 games have been played
(which is about 6 to 8 solid hours of play.)
Here are the simulation results for scenario one:
GAMBLSIM by Steve Fry
-------
Simulation of Craps Pass Line Wagers
Odds Multiplier . . . . = 0
Session Bankroll . . . = 120.00
Session will not quit at any win goal
Max. Decisions to quit = 200
No. Sessions simulated = 20000000
Starting Random seed . = 8447913
------------------------------------
Simulation Results per Session
------------------------------------
Avg. No. games played . = 200.00
Avg. No. games won . . = 98.59
Avg. No. games lost . . = 101.41
Avg. No. dice rolls . . = 675.14
Avg. Total amount bet . = 200.00
Bankroll was busted . . = 0.000% of the time ( 0)
Win goal was met . . . = 0.000% of the time ( 0)
Bankroll decreased . . = 55.152% of the time
Bankroll increased . . = 39.330% of the time
Avg (mean) end bankroll = 117.17 (change of -2.83)
Median ending bankroll = 117.09 (change of -2.91)
Std-dev ending bankroll = 14.14
Skew of ending bankroll = 0.00
Kurtosis end bankroll = -0.01
Std-dev # games played = 0.00
Median # games played = 100.00
Average tot expectation = -1.413%
Probability Distribution Function of Ending Bankroll:
* *
* *
* *
* * *
* * *
* * * *
* *** *
** * **
** * **
** * * **
** * * **
** * * * **
** * * * **
** * * **
** * *
** * **
* * **
** **
** * *
** *
* **
** **
** **
** * *
**************-----------------------******************
X-axis is change in bankroll, from -80.000000 to 82.000000 units
Y-axis is the relative probability of achieving that ending bankroll
As expected, the bankroll PDF shows the classic "bell curve" normal
distribution centered about the mean. The expectation, -1.41%, printed out in
the statistics area, is as expected.
Now here are the results for scenario two:
GAMBLSIM by Steve Fry
-------
Simulation of Craps Pass Line Wagers
Odds Multiplier . . . . = 10
Session Bankroll . . . = 920.00
Session will not quit at any win goal
Max. Decisions to quit = 200
No. Sessions simulated = 20000000
Starting Random seed . = 8117957
------------------------------------
Simulation Results per Session
------------------------------------
Avg. No. games played . = 200.00
Avg. No. games won . . = 98.59
Avg. No. games lost . . = 101.41
Avg. No. dice rolls . . = 675.15
Avg. Total amount bet . = 200.00
Bankroll was busted . . = 0.000% of the time ( 0)
Win goal was met . . . = 0.000% of the time ( 0)
Bankroll decreased . . = 50.848% of the time
Bankroll increased . . = 48.894% of the time
Avg (mean) end bankroll = 917.17 (change of -2.83)
Median ending bankroll = 914.72 (change of -5.28)
Std-dev ending bankroll = 152.87
Skew of ending bankroll = 0.04
Kurtosis end bankroll = -0.01
Std-dev # games played = 0.00
Median # games played = 100.00
Average tot expectation = -1.413%
Probability Distribution Function of Ending Bankroll:
****
* *
** **
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* **
** **
*****************-----------------------------------*******************
X-axis is change in bankroll, from -768.28885 to 821.86870 units
Y-axis is the relative probability of achieving that ending bankroll
Because this gambler sized his bankroll to match the commensurate risk of
scenario one, his PDF is also bell shaped but is much wider, as is seen in the
standard deviation number, which went from 14.1 in scenario one to 153 in
scenario two. However, the mean change in bankroll (-2.8 in one and -2.8 in
two) remained the same -- which means the addition of the odds bet had no
effect on the mean outcome. This is also seen in the expectation for
scenario two, which is still -1.41%
What happens when the under funded gambler faces a significant risk of ruin?
This is seen in the simulation results for scenario three:
GAMBLSIM by Steve Fry
-------
Simulation of Craps Pass Line Wagers
Odds Multiplier . . . . = 10
Session Bankroll . . . = 120.00
Session will not quit at any win goal
Max. Decisions to quit = 200
No. Sessions simulated = 20000000
Starting Random seed . = 8667913
------------------------------------
Simulation Results per Session
------------------------------------
Avg. No. games played . = 162.44
Avg. No. games won . . = 80.07
Avg. No. games lost . . = 82.37
Avg. No. dice rolls . . = 548.35
Avg. Total amount bet . = 162.44
Bankroll was busted . . = 39.898% of the time ( 7979584)
Win goal was met . . . = 0.000% of the time ( 0)
Bankroll decreased . . = 56.769% of the time
Bankroll increased . . = 43.046% of the time
Avg (mean) end bankroll = 117.68 (change of -2.32)
Median ending bankroll = 78.16 (change of -41.84)
Std-dev ending bankroll = 132.65
Skew of ending bankroll = 0.89
Kurtosis end bankroll = -0.06
Std-dev # games played = 54.87
Median # games played = 199.17
Average tot expectation = -1.429%
Probability Distribution Function of Ending Bankroll:
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
* ****************
+*****----------------*****************************************************
X-axis is change in bankroll, from -120.00000 to 843.00000 units
Y-axis is the relative probability of achieving that ending bankroll
The first thing to notice is that this scenario has the bettor busting his
bankroll 40% of the time, whereas with the first two scenarios he never busts.
Next notice the bankroll PDF is not bell shaped, but has a big spike
at -120 units (which accounts for the times the gambler is ruined), then there
is a gentle hump that peaks around +46 units and extends far to the right.
Looking at the expectation (-1.4%) and the mean ending bankroll (-2.3 units)
we see that it is basically the same as if odds were not taken -- in other
words, the odds bet had no change on the player's expectation even when faced
with significant risk of ruin, and thus the house made no extra money on the
odds bet.
It may not be intuitive at first since so many sessions (40%) bust at -120
units, but if one examines the PDF, those sessions that go way beyond +120 and
out to +843, although rare, are enough to swing the balance such that the
addition of odds bets has no overall effect on expectation. The addition of
odds bets clearly increases variance, as is seen in the standard deviation
results, but if the bettor has sufficient bankroll he can avoid the increased
risk of ruin that comes with an increased amount of money bet.
-------------------------------------------------------------------------------
Q12: The house is favored to win the "odds" bet; why don't they make money?
A12: (Steve Fry)
Although on average the house wins more odds BETS, they don't win more MONEY
on the odds bets. That is because the payoff for the odds bets is EXACTLY
opposite of the odds on the bet (see Q03.) For example, odds on the 4 or 10
are 2:1. That means that when you place the odds on the 4, on average you will
lose two times for every one time you win, but when you do win you are paid off
double your bet. In the long run it evens out to zero -- for every two times
you lose a dollar there will be one time you make two dollars.
Of course, the exact opposite is true if you lay odds (odds behind a Don't
Pass or Don't Come). You will win twice as often as you lose, but you are
paid off only half your bet. Again it is a wash monetarily.
Another effect of placing odds (or any other fair bet that has uneven odds)
is that because you will lose more bets than you win, for sessions that
comprise only a few number of bets, in general you will have more losing
sessions than winning sessions even though monetarily there is no gain or loss.
As the session length increases, this effect is diminished until you have as
many winning sessions as losing sessions.
To see the long-term effect of placing the odds on the 4, GAMBLSIM can be used
to examine how a session is likely to turn out. When you run GAMBLSIM you can
pick bet type "4" -- a Buy bet on the 4. Enter "0" for the commission, and
this is the same as simulating just the odds bet on a point of 4 (the passline
bet is ignored.) To simulate a rather long session, let's use the same as in
Q11 (representing 6 to 8 hours at the craps table), which was 675 rolls of the
dice. Here are the results of the simulation:
GAMBLSIM by Steve Fry
-------
Simulation of Buy bet on 4/10
0.00% Commission charged only on a win
Session Bankroll . . . = 100.00
Session will not quit at any win goal
Max. No. rolls to quit = 675
No. Sessions simulated = 20000000
Starting Random seed . = 341107
------------------------------------
Simulation Results per Session
------------------------------------
Avg. No. games played . = 169.50
Avg. No. games won . . = 56.50
Avg. No. games lost . . = 113.00
Avg. No. dice rolls . . = 678.00
Avg. Total amount bet . = 169.50
Bankroll was busted . . = 0.000% of the time ( 0)
Win goal was met . . . = 0.000% of the time ( 0)
Bankroll decreased . . = 49.294% of the time
Bankroll increased . . = 48.541% of the time
Avg (mean) end bankroll = 99.99 (change of -0.01)
Median ending bankroll = 99.33 (change of -0.67)
Std-dev ending bankroll = 18.41
Skew of ending bankroll = 0.06
Kurtosis end bankroll = 0.00
Std-dev No. dice rolls = 3.46
Median No. dice rolls = 337.50
Average tot expectation = -0.004%
Probability Distribution Function of Ending Bankroll:
****
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
* *
** *
* **
******************-------------------------------******************
X-axis is change in bankroll, from -99.000000 to 101.00000 units
Y-axis is the relative probability of achieving that ending bankroll
The bottom line of the statistics, expectation, is zero as it should be, but
notice that 49.3% of the time a session ended a loser while 48.5% of the time
a session was a winner (so 2.2% of the time the bankroll didn't change.) This
difference of 0.75% between losing and winning is larger for sessions of fewer
rolls (a session of 225 rolls has 1.24% more losing sessions) and smaller for
longer sessions (a session of 1350 rolls has only 0.49% more losing sessions.)
-------------------------------------------------------------------------------
Q13: What is the "Gambler's Fallacy"?
A13: (Mason Clarke, Alan Shank)
Mason writes:
-------------
Here is a very good definition:
*
http://www.fallacyfiles.org/gamblers.html
*
At this site one can find definitions for both the Gambler's Fallacy and
the Reverse Gambler's Fallacy.
Both fallacies are based in the same error ...
presuming a repetitive sequence of events
buried in a past random process is a trend.
Alan writes:
------------
The Gambler's Fallacy is based on a misunderstanding of the difference between
an infinite universe (like craps and roulette) and a finite one (like card
games). Quoting from Epstein, "The Theory of Gambling and Statistical Logic":
The law of large numbers has frequently been cited as the guarantor
of an eventual head-tail balance. Actually, in colloquial form, the
law proclaims that the difference between the number of heads and the
number of tails thrown may be expected to increase indefinitely as the
number of trials increases, although by decreasing proportions. Its
operating principle is "inundation" rather than "compensation."
-------------------------------------------------------------------------------
Q14: How long is the "long term"?
A14: (Alan Shank)
There is no real answer. There is no dividing line between short term and long
term; it is a continuum. The more bets, the more chance that the difference
between actual results and expectation will be very small, compared to the
total amount of the bets. You can see this very clearly from the fact that the
expected value increases at the same rate as the number of bets, while the
standard deviation increases with the square root thereof.
1 $5 passline bet: ev: -$.0707 SD: $4.9995
60 $5 passline bets: ev: -$4.24 SD: $38.69
600 $5 passline bets: ev: -$42.42 SD: $122.35
6000 $5 passline bets: ev: -$424.24 SD: $386.91 (NOTE: ev is now > SD)
60000 $5 passline bets: ev: -$4242.42 SD: $1223
600000 $5 passline bets: ev: -$42,424.24 SD: $3869
For one bet, you have a .4929 chance of coming out ahead, the chance of winning
a passline bet. Assuming you don't run out of money, the chance of coming out
ahead or breaking even is dependent on the ratio of the ev to the SD; that
tells you how lucky you have to be to overcome the inherent disadvantage the
casino enjoys over you. For example, if the ev is -$4.24 and the SD is $38.69,
the ev/SD is .109. This means that your result for the session has to be about
a 9th of a standard deviation better than expectation. The probability of that
can be calculated, and is the same for any standard deviation. You can read it
from a table of "Z values" in any statistics textbook. The probability is .456.
For the other sample sizes:
600 ev/SD = -0.347 probability of breaking even or better: 0.367
6000 " = -1.096 . . . . . . . . . . . . . . . . . . . . 0.136
60000 " = -3.46 . . . . . . . . . . . . . . . . . . . . 0.0003
600000 " = -10.97 . . . . . . . . . . . . . . . . very, very, small
So, make any definition you want of "long term," in term of a player's chance
of breaking even or better, and you can calculate how many bets that would be.
(Of course, it's different for different bets.)
A 1% chance of being ahead corresponds to 2.33 SD better than expectation. To
calculate the number of passline bets yielding an ev/SD ratio of 2.33, we can
do:
(ev/bet) / (SD/bet) * sqrrt(number of bets)
For example, for 6000 bets, we'd have:
-.0707 / 4.995 * 77.46 = 1.096
To solve for the number of bets, then, we'd do:
2.33 = (-.0707) / 4.995 * sqrrt(n)
2.33 = -.01415 * sqrrt(n)
-164.66 = sqrrt(n)
n = 27,114
So, if you define "long term" as the number of passline bets after which only
1% of players are expected to break even or better, the answer is 27,114 bets.
-------------------------------------------------------------------------------
Q15: Can one exploit trends to improve one's chances?
A15: (Mason Clarke, Alan Shank)
Mason writes:
-------------
There is no replicable evidence of a predictive trend in random and independent
events. Such replicable evidence could prove that these events are not random
or not independent or neither random nor independent.
Human beings perceive trends in series of random and independent events.
Based on the evidence, trends exist in rolls of the dice but they offer no
information as to future events. The error lies in presuming a repetitive
sequence of events buried in a past random process is a trend that will extend
into the future.
Interested gamblers can further satisfy their avowed curiosity about trends at:
*
Patterns Exist in Gambling, But They're Meaningless
http://krigman.casinocitytimes.com/articles/5288.html
*
You Can Prove Whatever You Want about Streaks and Trends
http://krigman.casinocitytimes.com/articles/14815.html
*
In the Casino, Seeing Shouldn't Always Be Believing
http://krigman.casinocitytimes.com/articles/5508.html
*
What Can Gamblers Learn from Superstitious Pigeons?
http://krigman.casinocitytimes.com/articles/6113.html
*
Alan writes:
------------
Every betting pattern is "chasing" a certain pattern of dice rolls. A
"flip/skip" system is chasing a P DP P DP pattern, a progressive system is
chasing consecutive wins, a Martingale does well on a choppy table. If you
follow a perceived "trend," you are chasing a continuation of that trend. A
flat passline bettor is chasing a preponderance of passes over misses,
regardless of order.
No matter what pattern you are chasing, whether you win or not is determined by
whether the dice rolls are favorable to that pattern. In all cases, this is
just a matter of luck, which I define as variance that is "in rhythm" with
one's betting pattern.
-------------------------------------------------------------------------------
Q17: What is "dice setting"? Does it work?
A17: (Steve Fry)
"Dice setting" is a very vague term that has many meanings. In the old days it
simply meant that the shooter turned the dice on the table so that the desired
number was showing on the up faces; then he would grab the dice and either
smack them on the table, blow on them, say some incantation, or some other
ritual, and then toss the dice down the table. This was purely a superstitious
action designed to bring good luck so that the dice produce the winning
numbers.
Then in the 70's the superstition grew into a pseudo-scientific theory. What
is the difference between superstition and scientific theory? It depends upon
the underlying mechanism describing the cause behind the results that are
observed. If that mechanism depends on pure faith or strong belief, then it is
a superstition. However, if that mechanism can be described within the current
state of mechanics and physics, then it becomes a theory that can be tested and
verified (or falsified) by scientific experiment, replicated by any scientist,
and, to varying degrees of precision, simulated by computer programs.
The new "dice setting" theory assumed that if the shooter does exactly the same
thing every time -- dice oriented the same, thrown in the same manner, same
force, to the same place on the table, that this will produce repeatable or
less-than-random results. The names "dice control", "dice influencing", and
"precision rolling" began to be used. Presumably, once you notice which
numbers are thrown more often you can bet accordingly, or maybe you can alter
your "set" to produce fewer sevens or whatever you desire. The problem with
this theory is that the dice are simply tossed without any order so as they
tumble there is no preference for any side of a die to show more than any
other.
The first scientific theory behind non-random dice throwing is called Body
Stabilized Dice Derandomizing (BSDD). This theory assumes that if the dice
can travel to their destination while maintaining their orientation in space
then the die faces that are showing (set) initially will also show when the
dice land. This mechanism of stabilizing a body is used with very massive
satellites orbiting the earth in the vacuum of space, but it is doubtful a
tiny plastic cube has enough mass to avoid even the influence of the apparent
wind as it is thrown, not to mention the imperfections of the thrower or the
bouncing off the felt and rubber pyramids. If this theory had any validity
then it would be a simple matter to set for double sixes and demonstrate that
the double sixes remained oriented up the whole time. No such demonstration
has ever been performed.
The next scientific theory is called Spin Stabilized Dice Derandomizing (SSDD).
Instead of body-stabilized dice (which rely on simple mass inertia to remain
stable) the spin of the dice (angular inertia combined with rotational speed)
is used to stabilize the dice such that they remain oriented roughly the same
in the presence of perturbations. Before the 80's, space satellites were not
as massive or as adept as today's, and they would spin to remain stable. How
might one use spin stabilization? If one sets a die with the six-face up and
spins it about an axis that goes through the center of the die perpendicular
to the six-face, then the die should be stable so that six always faces up.
This is called Vertical SSDD because the spin axis is vertical (with respect to
the floor of the Craps table.) There is a cheating dice roll called the "whip
shot", or
"stacking"
or "sliding" that relies on Vertical SSDD to spin the dice while sliding them
to the back wall. Unfortunately, a legal toss must pass through the air and
rebound off the back wall. When the vertically spinning die collides with a
flat surface at an angle, one of the sharp corners on the die will catch the
soft felt or rubber and disrupt the spin, probably beyond its ability to remain
stable.
Instead of spin-stabilizing the die to always show the six-face, what if we
spin it horizontally instead of vertically such that the six-face (and
correspondingly the one-face) never shows. This idea (arrived at independently
by several people in the 80's and 90's) has less control over which faces show
(instead of one face, now four faces can show with equal chance), but the
mechanism allows for the spin axis to remain stable as long as all reflections
are off of flat surfaces and the plane of reflection (the plane perpendicular
to the reflecting surface through which the motion of the die passes) is always
perpendicular to the spin axis of the die (which is always perpendicular to
the two die faces that one desires not to show.) If this SSDD condition can be
maintained then in theory a shooter can produce non random results with his
dice rolls, from which he can presumably profit by placing the proper bets that
take advantage of the derandomizing nature of the rolls.
Does it work?
That is unknown because nobody has yet produced evidence of the skill required
for SSDD. Such evidence would be easy to produce with a camera capable of
observing the dice with sufficient detail -- if the shooter is spinning the
dice about the 1-6 face axis when launching them, then they should come to rest
while still spinning horizontally about the same 1-6 face axis. Lacking a
camera, one could still produce evidence that this skill exists by
demonstrating many verified dice rolls where the one or six face do not show
(assuming they don't show because the dice are spinning about the 1-6 face
axis.) However, no such evidence is forthcoming.
Of course one cannot be expected to have perfect SSDD control. What if the
efforts of the dice roller were only partially effective? Would it still
work? This question is answered in Q17 with the concept of Spin Effectiveness
(SE), which is a measure of precisely _how_ partially effective the shooter is.
It is this author's opinion that it is not humanly possible to both spin the
dice with sufficient rotational speed and keep the spin axis horizontal to the
table enough to overcome the perturbations associated with reflections off the
floor and pyramid wall. However, it may be quite feasible to design and build
a machine that is capable of such a feat. This machine would merely prove that
SSDD is possible, but not that a person could do the same. There is a way to
"falsify" the claim that SSDD works: simulate it. This massive computer
simulation would require man-years of effort, but if each of its components is
verified by experiment, it should be able to prove whether it is easy,
incredibly hard, or virtually impossible for a human to demonstrate SSDD.
-------------------------------------------------------------------------------
Q18 But does SSDD dice setting really work?
A18: (Mason Clarke, Stacy Friedman, Van Lewis)
Mason writes:
-------------
A simple, scientific testing model for the existence of dice control is not
difficult to create.
Some number of individuals roll the dice for an equal number of rolls under
conditions that are as physically equal as possible. Half are dice controllers
and half are not. Those recording the rolls do not know how the rollers are
identified. The recorded sets of rolls are marked with codes that tie to the
individuals, but the links are not recorded along with the roll records.
The roll recordings are examined. They are divided by some set (or sets) of
criteria into groups of rollers whose roll records evidence "dice control" and
groups of rollers whose roll records do not.
A statistically significant difference from a chance mix of dice controllers
and non-dice controllers selected by any criteria used can be a first
indication that an identifiable phenomenon might exist. Of course, the absence
of any significant variation from chance does not prove that the phenomenon
doesn't exist.
Several tests using this protocol by different testers and in different
environments would be required before any hypothesis regarding the existence of
the phenomenon of dice control could be supported. This is a simple
application of the scientific method to the question whether or not there is
some phenomenon for which a cause should now be determined.
"Gregg Cattanach" wrote:
One test I'd like to see is slow-motion video tape of a 'precisionWhat you are suggesting is equivalent to an investigation into the theoretical
roller' doing his thing and determine how many times the dice (both
dice) actually manage to keep the horizontal axis constant. Having
that axis spin 180 or 360 degrees wouldn't count (that would be in
the just getting lucky category). I bet he couldn't do it 1 time
in 20 with a legal roll hitting the back wall. This would tell me
far more than any statistical test.
basis of for a perpetual motion machine without first confirming that the
energy output of the device exceeds the input.
First things, first! Without a phenomenon, no confirmation of the theoretical
mechanism of its cause is required.
-------------
The hypothetico-deductive method basically involves ..
(.) putting forward a hypothesis
(.) conjoining it with a statement of 'initial conditions'
(.) deducing from the two a prediction
(.) finding whether or not the prediction is fulfilled.
It is generally not possible to say which came first, the observation or the
hypothesis. However, the observations and their relation to the prediction are
what give the process weight.
What is missing from this process in "precision rolling" is credible
observations.
There are also two problems inherent in the hypothetico-deductive method that
are relevant to "precision rolling" ...
(.) The Problem of Alternative Hypotheses
The hypothetico-deductive method holds that a positive instance of an
observational prediction of a given hypothesis confirms that hypothesis. The
problem is that this same observational result also confirms many other
hypotheses that are incompatible with the given one.
(.) The Problem of Statistical Hypotheses
The second problem comes as a result of the limitations of deduction.
Specifically, when you have hypotheses that make statistical claims,
observational consequences cannot be deduced from them. For example, consider
the hypothesis that vitamin C will tend to shorten the duration of a cold. When
you infer that the average length of a cold for a group of people taking
vitamin C will be smaller than that of a group not taking the vitamin, the
inference is inductive, not deductive.
Several posters on this NG who actually think have attempted to set up
experiments that would meet the difficulties inherent in the
hypothetico-deductive method that you are attempting to apply to
"precision rolling".
What is forthcoming from the "precision rolling" believers is anecdotal
evidence. This subjective truth is fine for belief or mysticism. It will
never be adequate as evidence because of the several well known limitations to
anecdotal accuracy. This is why anecdotal evidence is rejected by the
scientific method.
-------------
STAN, THE CRAPS BUNNY, controls all the dice rolls in the universe. There is
no dice roll that is outside the MIND OF STAN. However, STAN is a subtle and
clever bunny. He makes certain that all the rolls of the dice at the crap
table are completely indistinguishable from random and independent trials.
There is a leap of faith required to accept the EXISTENCE OF STAN. However, it
is infinitesimal in length compared to the leap of faith you so blithely accept
in "dice setting" because ALL of the confirmable evidence supports STAN's
existence.
-------------
Dice control is just the current hot grift largely started by a previously
established gambling systems con named Jerry Patterson. Of course, it makes
more "common sense" than simple superstition. It is hard to turn much of a
buck selling what is easily perceived as simple superstition.
Roulette wheel clocking for statistically significant variance from chance was
a better grift than simple trend tracking. Equally, dice control has more
income potential when compared to lessons on how to recognize and/or how to
become a hot shooter.
Patterson first published his dice control grift in 2000.
The dice control bunko bandwagon may have been rolling before that.
The replicable evidence from any source is still lacking.
A wise man proportions his belief to the evidence.
Stacy writes:
-------------
"Chris Wegener" wrote:
But the dice do tumble. They separate in the air and are rotating through
all three axis.
Not always.
Even if they didn't, when they hit the wall the energy of the throw causes
the dice to tumble making the outcome random.
Not always.
Dice have been shown through the centuries to be an outstanding source of
random numbers. They can not be controlled by any human being.
This is amazingly wrong. Only in this century have machining techniques become
good enough to construct dice without bias enough to be considered even a
reasonable source of randomness. The evolution of dice includes bone, wood, and
clay pieces, none of which were precise enough to produce anything close to a
uniform distribution of outcomes.
Moreover, the reason the casinos have the rubber pyramids at the end of the
table is that dice *can* be controlled. Do this: Take two dice and stack them
with the ones on top of each other, then put them in your fist so the ones are
pointing out the end. Now, kneel down and pretend you're bowling, and release
the dice rolling forward when your hand gets to the floor. The dice will roll
forward around the 1-6 axis, and a 1 or a 6 will not show up on either die.
This is called a blanket roll, and it's been around ever since anyone thought
of cheating at dice.
For a more impressive display of dice control, I refer you to Steve Forte's
videos and his use of the whip shot (high-rotational-velocity spin around the
vertical axis) to throw a given dice combination at will. I think in the video
he threw several 12s in a row on command.
For those looking for an analogy, there are champion dart players who can
consistently hit the inner bulls-eye circle from 7 feet, 9.25 inches away. Who
can confidently proclaim that the level of human dexterity required to
influence the outcome of the dice is so much higher than that of dart-throwing
as to make dice-influencing absolutely impossible? But if it's not impossible,
well...
An example of a thing equates to scientific disproof of an opposite conjecture.
If a conjecture states something is impossible, it suffices to show one example
to disprove it.
Van writes:
-----------
No evidence [of effective precision shooting] has been provided to the
"boys" here and as such, ***poof*** any discussion is meaningless.
Speaking only for myself, I would label such discussions theoretical
rather than meaningless.
And, if there are any positive recitals they are magically pixied away
with the end all comment, "Luck".
Anecdotal evidence supports a contention that precision shooting might
be possible, but it does not support a contention that it ->is<-
possible. Clearly a good hand, even multiple consecutive good hands,
can be due, as Star Trek's Spock would say, to random chance operating
in one's favor. To show that positive results are due to a shooter's
skill rather than the operation of random chance a statistical analysis
of a large number of trials, made under casino conditions and documented
by a neutral party, would be required. Such an experiment has yet to be
performed.
Unlike some other participants in this forum I do not dismiss out of
hand the possibility of effective precision shooting for lack of
credible evidence. In fact I believe it ->is<- possible. I'm just not
willing to bet very much on it until there is credible evidence.
This is a skill. A dexterity skill. It is a human skill. That
means some can perform better than others.
That would be true if the skill were documented to exist. Such is not
yet the case.
The Mason's of the world would like you to believe those teaching the
skill are snake oil salesman.
Since the existence of the skill remains questionable those teaching it
may well be selling snake oil. OTOH they may not be. My recommendation
to prospective students is to do a cost/benefit analysis prior to
committing large amounts of time, effort, and money to learning how to
influence the dice. (BTW that analysis should include the very real
possibility that the casinos will change the rules for shooting should
enough shooters develop a level of skill sufficient to defeat them.)
I have no problem with someone attending a dice influencing seminar
primarily for its social aspects. However, the pitches I have seen for
those seminars emphasize profit potential. Caveat emptor.
I have been successful at times in utilizing the skill. I have seen
with my own eyes others do the same.
And I have seen with my own eyes people sawed in half, then reassembled
without any ill effects.
Sometimes the skill is on and sometimes it is not. I have had days
when nothing worked. I have had other days where everything worked.
If it does not work on a particular day that does not negate the
existence of the skill.
Nor does it working on a particular day validate its existence.
-------------------------------------------------------------------------------
Q19: How can SSDD give the shooter an advantage?
A19: (Steve Fry)
To answer this question, not only must we _assume_ that SSDD works, we must
_assure_ that it works by quantifying exactly how good it works. In order to
do this, the mathematical term Spin Effectiveness (SE) is introduced that is a
quantified measure of how effectively the shooter keeps the horizontal spin
axis maintained throughout the roll. An SE of zero means there is no control,
no spin effect, and the die will tumble randomly; in this case the probability
of each side of each die landing up is 1/6, because there are six sides and
each side is equally likely. An SE of one means perfect control, the spin will
be completely effective and the two sides (called the axis sides) will never
show up while four sides (called the hub sides) will show up each equally
likely with a probability of 1/4. For any SE between zero and one, there is a
linear relationship between SE and the probability that either an axis side
or a hub side will land face up. We assign a probability, P, to each side of
the die that reflects its chance of appearing. For the two axis sides,
Paxis=(1-SE)/6 and for the other four sides Phub=(2+SE)/12. Knowing these
probabilities and which of the 3 spin axes on each die have been chosen,
we can construct a 6 by 6 table showing the probability for each dice
combination showing. This table can be used to compute the expected value
for any of the bets at the Craps table.
But first an examination of SE is in order, since this is the most important
variable. Using a computer program (described below) one can generate a
plot of expected value versus SE for any desired bet, but if you do not know
the value of your SE, and how stable it is, the graph won't help tell how you
will actually perform.
There are two ways of determining what the SE value is. The most exact way
would be to monitor precisely the motion of the dice. If the spin remains
stable throughout the roll you have a success. After a few rolls (1000 or so)
you calculate the SE as the number of successes divided by the total number of
rolls. In order to monitor the dice with such precision will require
sophisticated computer sensing equipment and special dice or else
high-speed cameras with auto zoom and track capabilities.
Some people might suggest you merely look at the final numbers showing, thus if
you set both dice for (1,6) on the axis then if a die shows 1 or 6 it is a
failure otherwise a success. This ignores the case where the die is tumbling
randomly and happens (2/3 of the time) to miss the two faces that would mark it
a failure even though it really was a failure. We can, however, take this into
account; thus the second way of determining SE is to record the dice numbers
that show for many thousands of rolls. You will have an average occurrence
rate, R, for each side, which is computed as the number of times that face
showed divided by the total number of rolls. How do you know how many rolls to
throw? Statisticians have fancy formulas that depend on certain factors, but
the simplistic answer is that you keep rolling until whatever you are trying
to measure (the R value for all six sides, in our case) stops changing. What I
would suggest is to compute the R's after each 100 rolls, simply keeping a
running average; the value to use for R is the latest average, plus or minus
the largest change in R over the last 10 changes (1000 rolls). Once the R's
are known, both axis side R's should be nearly the same (or average them) to
give you Raxis, and the four hub R's should be averaged to give you Rhub. If
any R significantly deviates from the norm it indicates an abnormal failure of
the dice setters ability, and the SE cannot be calculated because the analysis
to follow assumes a stable, balanced SE value. Equating R and P for both axis
and hub, then doing a least-squares fit on SE gives us the following equation:
SE = (12*Rhub - 24*Raxis + 2) / 5
Once the SE is known to a good level of confidence, we are ready to see if and
how it can give the shooter an advantage. To address this issue I have written
a computer program called DICESET, which you can download from this location:
*
http://members.dslextreme.com/users/stevefry/diceset.exe
*
For the Fortran programmer, you can download the source code here:
*
http://members.dslextreme.com/users/stevefry/diceset.for
*
There is a more complete discussion about the topic of dice control, including
pictures of graphs using data generated from DICESET showing the expectation
curves for different bets, at this location:
*
http://members.dslextreme.com/users/stevefry/diceset.htm
*
There are three axes about the center of a die that are perpendicular to the
six faces, so there are three possible sets per die. It doesn't matter if the
die rotates clockwise or counterclockwise because the math assumes each of the
four sides is equally likely to show. Since opposite sides of a die always sum
to seven, the three sets are labeled 16 (meaning the one-face and six-face are
axis faces), 25 (for the two-face and five-face), and 34 (for three-face and
four-face.) Since there are two dice and three sets per die, there are six
total possible dice sets, which are always denoted by the two-digit sets for
each die separated by a pound sign. Here are the six dice sets:
16#16 16#25 16#34 25#25 25#34 34#34
Summarizing the analysis of the graphs in the above website:
The first graph shows how the different dice sets perform on the Pass bet with
double odds, a graph of expected value (actually EVR) versus SE. For SE=0 the
dice are random and so all six sets produce the usual EV with 2x odds: -.6%
However, sets 16#25 and 16#34 can yield a 21% player advantage with perfect
spin effectiveness (SE=1).
The second graph is a duplicate of the first except it shows the Don't Pass
bet with 2x odds. The best EV possible here is only 12.5% (about half that
possible with pass) obtained with the 34#34 set, and that doesn't even occur
at the highest SE: the EV drops to 10% at SE=1.
The above charts assumed a constant dice set. What if you used one set for the
come-out roll and a different one, depending on the point, after that? Such a
strategy has been developed, called the SuperPass and SuperDont strategies:
* For SuperPass, set 16#16 on the come-out roll; after that, for a
point of 4 or 10 set 16#34, and for all other point numbers set 16#25.
* For SuperDont, set 25#34 on the come-out roll; after that, set 25#25
for a point of 6 or 8, and 34#34 for all other point numbers.
The third graph shows the SuperPass and SuperDont with 2x odds as well as 3
more bets: Hard 6, Any Craps, and Any 3. The SuperPass behaves like the best
Pass set for low SE values, and at high SE values it surpasses the best Pass
and produces a top EV of 32.6% at SE=1. The SuperDont has an even better
performance than the SuperPass, and it yields a top EV of 46.4% at SE=1.
This is surprising because using a fixed set for Don't is much worse than a
fixed set for Pass, yet with a variable set the situation is reversed. As good
as the Super bets are, they can't beat the 100% EV that the other 3 bets can
produce at SE=1. Clearly if an SSDD expert can exceed a 10% spin effectiveness
then Hard 6 (and Hard 8 also, with the same set) is the most profitable bet.
If he can achieve from 4% to 10% effectiveness then SuperDont is the way
to go. If the SE is less than 4% then we need to increase the odds bet. At
100x odds, the EV breaks above zero with an SE=0.14% This represents the
lowest possible profitable SE.
-------------------------------------------------------------------------------
Q20: How does the Craps NewsGroup (rec.gambling.craps) work?
A20: (Steve Fry)
There are two types of postings to the newsgroup: those by humans and those
by automated spambots. Half of the problem with reading and enjoying the
newsgroup nowadays is being able to ignore the spam (using automated or
self-regulated filters) and ignore the kooks and trolls (using a killfile or
self regulating) if they bother you. Of the actual people who post to the
newsgroup, they fall into many different categories. Some are fleeting
first-timers who are brand new to UseNet and haven't found a regular set of
newsgroups to read, ask some newbie questions, then disappear. Some posters
use the newsgroup to further their business, be it selling books or equipment
or selling seats to a Craps seminar that will teach such things as money
management and dice setting. Of the rest of the real people who post here,
about 90% are here to genuinely help newbies who first come here for
information, about 10% are kooks and trolls, and less than 1% are the actual
newbies who have come to the newsgroup to learn and share.
What is a "troll"? The term comes from the way a fisherman will throw out his
line behind a moving boat in the hopes that unsuspecting fish will bite at the
bait. In the newsgroup it refers to a passing "newbie" who throws out
apparently innocent misunderstandings about the game of Craps, then when some
unsuspecting regular poster rises to the bait and corrects him, the "newbie"
becomes defensive and his remarks more and more inflammatory and audacious.
Eventually the troll tires of these fish and moves to some other newsgroup.
His sole enjoyment is to purposefully disrupt the calm order of the newsgroup
by supporting and pretending to defend controversial topics that are generally
known to be incorrect. Once the newness and excitement of the disruption is
over and people start ignoring him, the troll moves on.
What is a "kook"? Also known as a netkook, imagine a troll at the peak of his
audacious ranting and vicious namecalling stage; that is a kook all the time.
Unfortunately there are two problems with a kook that you don't have with a
troll: kooks don't ever leave the newsgroup, and they actually believe their
inane arguments are correct. This means they can never be corrected no matter
how hard any one tries, and since they don't leave, any unsuspecting newbie who
genuinely wants to help will get embroiled in the same old arguments that
failed to work on the kook the last time someone rose to the bait.
Who are the current kooks? To mention their names would only serve to feed
their egos. Besides, all newsgroups evolve as the group of regular posters is
added to and deducted from, as general topics change, even kooks come and go.
Two classic netkooks from the BlackJack and Craps newsgroups of the distant
past were Doug Grant and John Patrick. There are several signs that identify
a kook and a troll: they refuse to answer direct questions or refuse to
provide references or equations to back up their assertions. They often refuse
on the grounds that they are smarter than everyone else and it's just too bad
that nobody gets it.
What do people in the newsgroup talk about? Generally there are four types of
topics: trip reports, table conditions at certain locations, the mechanics of
Craps and how it operates, and the theory of Craps and gambling in general.
Some people love trip reports (live vicariously) and some people hate them
(useless information). The other topics will be populated by those wanting
to gain knowledge, dispense knowledge, or disrupt the gaining and dispensing
of knowledge.
-------------------------------------------------------------------------------
.
- Follow-Ups:
- FAQ for John B
- From: Clams Canino
- FAQ for John B
- Prev by Date: Re: True odds bets?
- Next by Date: Re: True odds bets?
- Previous by thread: Get $100's In Free Poker Money & Gifts No Deposit Needed.
- Next by thread: FAQ for John B
- Index(es):
Relevant Pages
|
