Terminology

 "Volatility"

 From: Michael J. Zehr Address: michaelz@michaelz.com Date: 15 June 1998 Subject: Re: JellyFish "equity" and "volatility" evaluations: explain please. Forum: rec.games.backgammon Google: 3585CFF4.89731E4@michaelz.com

```Leo Bueno wrote:
> Could someone please explain the meaning and use of JellyFish's
> "volatility" and "equity" evaluations, particularly with respect to
> cube decisions?
>
> Thanks.

Tom Keith's glossary covers equity
(http://www.bkgm.com/gloss/lookup.cgi?equity) as pointed out by Claes
Thornburg, but it doesn't give a rigorous definition volatility.

JF uses the variance of equity as volatility.  For each of the 36 rolls
(21 unique rolls, but the non-doubles count twice in the equation) find
the equity resulting from the making the best move.  Find the equity
change for each move, square that, and add them all up, divide by 36 and
take the square root of that.

This is similar to the concept of market losers.  When you count market
losers you count the number of rolls with a positive equity swing
greater than some constant (the difference between the cash point and
your current equity).  For those that exceed this value you count 1, for
those that don't you count 0.

Volatility is similar except that every roll contributes something to
the variance with the ones that cause the greatest equity swing
contributing the most.

It's useful because for some positions one can get an approximate
double/no-double decision by calculating aE + bV (where E and V are
Equity and Volatility, a and b are factors) and doubling if the result
is greater than some constant.  (a, b, and the constant are left as
exercises for the reader.... though E+V > .60 means an almost certain
double in non-gammon, non-contact races.)

It's been written that if you're close to your cash point, a few market
losers is enough to make it a double; if you're farther away, you need
more market loser (see, among other sources Bell, "Winning with the
Doubling Cube," Gammon Press).

The formula above is mathematical way of saying that.  The more market
losers you have ("cashers" in the references source) and the bigger they
are, the greater V.  So if you're close to your cash point (E is .55)
then a few market losers (V >= .05) is enough to make it a double.  If
you're far from your cash point (E is .40) then you need many big market
losers (V >= .20) to make it a double.

[E + V > .6 is a simplified version of a more accurate formula.  But
there is no linear function of E and V that says whether or not to
double with 100% accuracy, even for very simple races.  Remember that
while neural nets might get V accurate, they don't do so well a job on E
(Snowie and JF sometimes disagree by .2 on equity, even if they both
agree on the best play in a position).  Furthermore, many doubles based
on this formula are easy takes, so while in theory it might be correct,
in practice it might be better to wait and give your opponent a tougher
decision to make.]

-Michael J. Zehr
```

### Terminology

Alphabet soup  (Tom Keith, Apr 2004)
"Anchor and guard" position  (Chase+, Apr 2010)
"Back game"  (Marty Storer, Jan 2004)
"Baffle box"  (garyo+, Mar 2005)
"Bagai position"  (Timothy Chow, Dec 2012)
"Banana split"  (Rich Munitz+, June 2011)
"Banana split"  (Adam Stocks+, Sept 2004)
"Beavers"  (Sander van Rijnswou, May 1999)
"Beavers"  (Shuman Lloyd Lee, Aug 1991)
"Blunder", "whopper"  (Raccoon+, July 2005)
"Bot"  (Pit Bull+, Mar 2004)
"Bronstein" clock setting  (rew+, Sept 2012)
"Calcutta auction"  (Roland Scheicher+, Dec 2001)
"Chouette"  (Roland Scheicher+, Mar 2002)
"Cube provocation play"  (Chuck Bower+, Apr 2007)
"Dance"  (William R. Tallmadge, May 1998)
"Dropper"  (Robert D. Johnson, Sept 1996)
"Duplication" and "diversification"  (Simon Woodhead, Nov 1991)
"Equity"  (Gregg Cattanach, Aug 2000)
"Equity"  (Gary Wong, Dec 1998)
"Equity"  (Chuck Bower, Oct 1996)
"Equity"  (Michael J. Zehr, Mar 1996)
"Equity", "volatility", "claim", "market"  (Erik Gravgaard, June 1995)
"Freeroll"  (montygram, Nov 2005)
"Gammon price"  (Ron Karr, Aug 1996)
"Gammon rate", "gammon price"  (David Montgomery, June 1995)
"Gammon-go" (GG) and "gammon-save" (GS)  (Mary Hickey, Feb 2004)
"Gammon-go" (GG) and "gammon-save" (GS)  (Marty Storer, Oct 2002)
"Gammon-go" (GG)   (Chuck Bower, Jan 2004)
"Golden point"  (Daniel Murphy, Dec 2004)
"Holding game"  (Alan Webb+, Dec 1998)
"In the box"  (Ken Bame+, Sept 2012)
International phrase dictionary  (David Allen Sorensen, Sept 1997)
"Joker"  (Richard Divdesman, Sept 1998)
"Kamikaze play"  (Bill Patterson+, June 2011)
"Kauder paradox"  (Carl Tait+, Nov 1995)
"Latto paradox"  (Jean-Pierre Seiman+, July 2004)
"Lose your market"  (Shuman Lloyd Lee+, Aug 1991)
"PRaT"  (Raccoon+, Jan 2007)
"Phantom double hit"  (Marty Storer, May 2010)
"Polish prime"  (Jason Lee+, Jan 2006)
"Pure play"  (Daniel Murphy, Nov 2000)
"Pure play"  (Casey Forrest+, Feb 1996)
"Raccoon"  (Steven Keats, Feb 2011)
"Root number"  (Ken Bame, June 2004)
"Russian Bridges"  (leobueno+, Mar 2013)
"Short play"  (AJ+, July 2012)
"Speed board"  (Gregg Cattanach, June 2004)
"Splot"  (mamabear, Apr 2007)
"Squeeze", "trap play"  (Philippe Michel+, Feb 1997)
"Suicide play"  (Brian Sheppard, Aug 1997)
"Swing tournament"  (Carlo Melzi+, Mar 2006)
"Table stakes"  (Carlo Melzi, Sept 2002)
"Technical play"  (Adam Stocks+, July 2002)
"Thematic"  (Beauregard+, Aug 2009)
"Thorp count"  (Stephen Turner, June 1996)
"Time," "timing," "checker," "dancing"  (Marty Storer, Apr 1992)
Turkish names for rolls  (Lars Soezueer, Mar 1997)
"Vigorish"  (Anthony R Wuersch, Feb 1995)
"Volatility"  (Michael J. Zehr, June 1998)
"Wash"  (Hardy Hübener+, Sept 2004)
"Wash"  (Brian Sheppard, July 1997)
"Weaver"  (Alan Webb+, May 2000)
"Zone" of attack  (Matt Reklaitis+, Dec 2007)