How to analyze saved games

[mweimer]

I need to know how to analyze my saved games.  All I get when I open a saved game (or session) is a replay of the game.  I assumed that "gnubg" or something was going to help me by suggesting moves other than the one I made or providing statistics (ala Snowie) or something else.  Was I wrong?  Is a replay all there is to it?

[don]

First, go to the drop down menu, Analyze/match.

Note that gnu will analyze according to your settings.  The more expert your settings, the longer it takes and the better (in general) the analysis.  Two move look-ahead is normally fine, four takes a while.  Don't do odd numbers here.

After a while, when it's done, drop down Windows/.... get the game record and analysis windoze open.  Move and resize for convenience and move the the game record to follow the game/match

See if that points you in the right direction.

--
don

[mweimer]

Thanks, Don.  I'll try it again.

[nixo]

Hello mweimer,

you can read this article, from Mary L Hickey, where she explains how she does.
It's available for free on gammonline  http://gammonline.com/members/Sep00/articles/hick.htm

Nicolas 

[don]

BTW, when I say look-ahead, I mean your ply settings should be odd numbers (1-3-5).  Seems even numbers are not so good.

--
don

[boomslang]

No, it should be even plies (try Supremo or World Class)

[sixty_something]

Here are my favorite tips and tricks for using GNU to analyze saved games played at FIBS .. these may be more basic than you need, but I've tried to make it as close to step-by-step for beginners as I can, for now .. GNU is a very powerful tool .. on first use, it can be a little overwhelming .. I know from experience .. now, it's becoming quite handy and very helpful

• first, Zorba's on-line GNU tutorial http://www.gammonlife.com/gnu/index.htm is well worth reading and/or revisiting
• View/Game record and View/Analysis panels are best both checked (thanks, Zorba, for that tip)
• Analyze/Analyze match after importing your matches
• moves and cube decisions are then and only then marked in the game record
• marked moves are: green - doubtful, blue - bad, red - very bad
• marked cube decisions are highlighted: gray - doubtful, pink - bad, yellow - very bad
• learn to visually understand GNU's color code markings, mouse over to see them and the Analysis panel for details
• Analyze/Match statistics for a quick overview of the match and each game along with an estimated FIBS rating
• Page Down/Page Up in the game record for move-by-move browsing
• EQUITY IS IMPORTANT - get a basic understanding using whatever metaphor works for you, but ignore MWC at first
• depress the MWC button in the analysis panel to turn it off - MWC means Match Winning Chances
• equity ranges theoretically from -3 to +3, but simply think of it as 1.0 for winning a cubeless/gammonless game
• your equity is similar to your probability of winning a game (thanks, dorbel, for this hint)
• at the start of a game, your equity is essentially 0.50
• negative change in your equity is BAD - the bigger the change the worse the move
• equity changes with each roll, move, and/or cube decision
• marked moves and cube decisions signal ranges of negative equity change relative to GNU's recommended play
• doubtful is a range around -5%, bad around -10%, and very bad around -25%, the latter can approach -100% with blunders
• the same holds for cube decision markups
• very lucky dice rolls are boldfaced, very unlucky rolls are italicized (also based on relative change in equity)
• now, get out your funny money (for educational purposes only, of course)
• wager $1 per game - you have a 50-cent share of equity at the start, a win puts $1.00 in your pocket, a loss costs you a dollar
• nickel, dime, and quarter losses in equity from moves and cube decisions can knock you out of a game and cost you that $1 PDQ!
• GNU, as I use it, helps teach minimizing these nickel, dime, or quarter equity losses whenever playing
• experiment with the Details, Show, and Temp[erature] Map buttons for other neat features .. leave the rest for later
• hope this helps you toward having fun GNUing .. it's a great learning tool

sixty_something

[sixty_something]

My $1 wager for a single game and nickel and dime metaphor for simplifying GNU's use of equity overlooks a tremendous amount of subtlety, complexity, and detail. But, frankly, that's the point. GNU offers so many metrics for analysis and evaluation it's easy to get lost and confused by the raw numbers. Since learning to use GNU for post-game analysis (thanks to dorbel's encouragement), I've been seeking ways to see the big picture and delve into the metrics only as necessary. Besides backgammon is a notorious and ancient money game and probability theory isn't everyone's cup of tea. We all understand a $1 bet on a game. So, nickels and dimes wasted from poor moves and bad cube decisions make a little more sense than a screen full of three digit decimal probabilities and this thing called equity.

When I look at my game records littered, almost always, with marked moves and missed cube decisions, say 3 doubtful, 2 bad, and 1 very bad (on good days), I get a very clear sense of what these mistakes have "cost" me in a game. For example, 3 nickels, 2 dimes, and a quarter's worth of poor choices add to 60 cents.  Of course, the exact amount in equity will vary. But this represents the wasted potential equity from my less than optimal play and yields a quick intuitive way to access it. That 60 cents remember is deducted from the 50 cents I suggest is a player's shared stake in equity at the start of the game.

Since backgammon is a game of strategy and chance, this doesn't mean the game is lost when we waste 50 cents or more in equity, sometimes much more. With each dice roll our equity in the game, our chance of winning or losing the game, goes up and down, sometimes wildly with lucky and unlucky rolls. But a lucky roll played poorly, wastes a potential gain in equity. Just as an unlucky roll well played minimizes equity loss.

This led me to thinking about how equity is gained in backgammon. On reflection, I think equity is gained in only two ways over which we have little to no control. We gain equity in a backgammon game with:

• playable good dice rolls (poor rolls can result in equity loss)
• opponents bloopers

We all know great dice even when played poorly can result in a beginner beating an expert in a single game. Likewise, bad dice played really well can beat good dice played poorly. So, to an extent the whole concept of equity gain and loss, or positive and negative change becomes almost irrelevant except as a measurement of our current stake in the game. So, my whole way of thinking about GNU's pesky markups is changing from lost equity to wasted equity. I can do little about losses, but there is hope in controlling waste.

With all it's metrics and supernatural power, GNU can't do a thing to teach us how to gain equity. Prayers to the dice gods might help, but I'm a skeptic there. Otherwise, we're dependent on opponents blunders to "gain" equity. Genrerally, that's not a good thing to depend on either, but DeaDice says lipstick works. So, what is the point of all this attention on equity?

GNU, used in post-game analysis, focuses almost all its analytical power highlighting ways in which poor moves and cube decisions waste potential equity. Since we have little to no control over how we gain equity, it's in minimizing wasted equity where we have the opportunity to learn and improve. Ben Franklin would have appreciated GNU, "A penny saved is a penny earned." My nickel and dime approach is just an easy way of quickly and visually understanding how you handle equity in a single game.

Finally, regarding my nickel and dime metaphor, here are the bounds in pennies of the default ranges used by GNU in marking moves and cube decisions:

	Wasted Equity
doubtful	Nickel	0.04 to 0.08
bad	Dime	0.08 to 0.16
very bad	Quarter	0.16 plus

Happy GNUing and watch those pennies!

sixty

[donzaemon]

Yeah,  good point for those who are unfamiliar with what equity is.
The equity expressed by bots for a position are usually the "cubeless equity" based on  "how many points you are likely to win in that given game DISREGARDING THE CUBE"  ie  cube position/size is irrelevant, you get 1 point for a straight win 2 for a gammon etc.    so it shows your chances out of 1 of winning there (it can go over 1 for gammons though but the cube doesn't matter)   
ie. if your equity is positive,  you're a favorite to win ,  if it's negative , you are more likely to lose.
There is also "cubeful equity" which brings the cube value into it.  this may be useful for assessing settlements in moneygames etc. but is likely to only confuse you.


How much equity should you have to double ? 
That gets complicated because positions are dynamic but  0.65 is good marker for doubles .... (though looking at market losers is proably a better philosophy for doubling)


Now ... the three numbers for the cube of course involve the cube ,  if you take and lose, you lose 2 rather than 1 so it usually comes down to managing that rather than looking at your cubeless equity so it's almost like using two different systems here and maybe should be mentally separated from the cubeless equity number.
Now on to the cube numbers :
It should be a take if it is still less than 1.   
this means you pass if it is more than 1... so you have technically "lost your market" if you double a position that shows more than 1.000 equity on the take evaluation.
The take/pass point is easy to figure out with logic if you realize passing gives him 1 point exactly (he wins one game)
so if I am giving him more than 1 point from the current position I pass,  if I am giving him less than 1 point from the position I take.

so once you know this,  finding good cube positions for study revolves around finding positions that are very close to 1.000 equity and preferably just under it,  a .999 position is a perfectly efficient double. 



now back to the previous post.


You mentioned that the ways we lose equity are
*playable good dice rolls
*opponents bloopers

While that is true and can't be ignored if you are dissecting games play by play to look at it scientifically, I would offer a slightly different view for practical play and people dedicated to learning the game that sees the game as a race to perfect play and completely ignores the dice rolls.  In other words, you are racing against your opponent striving for perfect play and each time somebody makes a mistake he will lose some ground and the only way to maximize your equity is to play perfectly.
Of course ,  in some cases (especially short matches) your opponent can lose significantly more ground but still win because of the equity given back by the dice gods but to trying to bring the dice into practical thinking can
1. keep you off track from where you can really improve
2. be futile because you have no control over the dice
3. be a little too complex to quantify or clearly think about in your head at times.

*some positions have many more good rolls than bad ,  and the other way around naturally exists as well.
*some are very balanced where good and bad rolls are about equal
*some are very benign where most rolls don't create much of an equity shift
*some are very volatile where the equity swing between different rolls can be huge ,  in a most extreme case it could be something like a certain roll will likely give you a backgammon but another roll will likely get you backgammoned  (imagine getting into this one at 3 away 3 away with $30,000 riding on the game...  might want to think about a settlement here)

These cases can be mixed and balanced in various ways so it can become very difficult to assess what the weight of any given dice roll is in the scope of the total possibilities for that roll.   In short ,  the value of good and bad rolls are much different for many situations. 


So in place of thinking about the roll at hand I will offer an alternative in the form of looking at the 'dynamics of the position'  and what certain rolls can do to it.
*what type of a game are you in
*how volatile is it
*what's your 'game plan'
*how's your race
*how's your positional advantage or disadvantage
*how is the balance of blots or other vulnerabilities

These points can often give you a very good sense of what your equity is for a given position and if you have a choice of rolls and go through the checklist for the different resultant positions you can get a very good sense of the differences they create.


It's very hard to ignore good and bad rolls, especially when they make a difference in the outcome of a game or match. The thing to remember though is in the long long run they will always be equal (assuming there is no cheating going on) and since you have no control anyway it is crazy to attach any emotion to them.

So if you are dedicated to learning the game ,  I would recommend an approach where you re-frame your point of view to consider the rolls basically insignificant and even look at the win/loss outcome as insignificant , and shift your thinking to train yourself to see errors and blunders with the same kind of intensity as you'd normally see a loss.  At the same time you can see a 'difficult' play** where you made the best move as a 'win'.  I imagine anyone who dedicates himself to this type of thinking and uses a bot to enhance this type of real and practical learning and seeing the dynamics of each position instead of just hoping for better numbers next time will see some terrific improvement in a short order of time.

** a 'difficult' play may be a move where the correct move is a very difficult 'hard to see' choice ,  but the alternatives are all big blunders (not just slightly wrong) .  These types of positions come up more than one would think in this crazy game

[sixty_something]

thanks for a terrific and insightful reply, donz .. i think your comment regarding re-framing is right on and GNU is a great tool for doing just that

So if you are dedicated to learning the game ,  I would recommend an approach where you re-frame your point of view to consider the rolls basically insignificant and even look at the win/loss outcome as insignificant , and shift your thinking to train yourself to see errors and blunders with the same kind of intensity as you'd normally see a loss.  At the same time you can see a 'difficult' play** where you made the best move as a 'win'.  I imagine anyone who dedicates himself to this type of thinking and uses a bot to enhance this type of real and practical learning and seeing the dynamics of each position instead of just hoping for better numbers next time will see some terrific improvement in a short order of time.

since using GNU regularly to analyze games, that's exactly what i've been doing .. sometimes i find myself spending as much time using GNU after a match as i spent on the match and i do analyze *every* match .. win or lose, i'm always most satisfied when my blunders are minimal and my error rate is low

i've been keeping track of my error rate in sets of matches for 1000+ moves on fibs and against GNU without tutorial help or hints .. it has been dipping from an Equiv. Snowie error rate well over 20 when i started using GNU .. now, collecting data using GNU's relational database, it is coming down to below 10 on average

while this has almost nothing to do directly with wins and losses, as i sometimes lose matches where i outplayed my opponent and just as regularly am rated ***Awful!*** by GNU and win a match .. yet it seems to have almost everything to do with wins and losses, too, as i keep hitting all time highs in my FIBS ratings .. after being stuck in the 1500's for over a year, i'm a very happy camper right now .. btw, it's been less than a month since i reinstalled and starting using GNU regularly - thanks to dorbel analyzing and commenting on a 3-point match we played .. my FIBS rating then had bounced irregularly from a low of 1480 to about 1620, seldom over 1600 for more than a few days .. today, i just hit 1700 for the first time and seem to be on a sustained stable trend (something i've also never noticed before for this long)

but, true confession time, i've been spending an inordinate amount of time doing this of late .. on occasion, i've also gone days at a time without playing on fibs at all because in warm-ups against GNU that error rate is 40+ .. i always review those especially bad warm-ups .. sometimes it helps .. sometimes i haven't a clue or just am too tired or distracted to see it .. when it doesn't, i've just been stepping away rather than reinforcing bad habits with bad play (i can thank dorbel a lot for pointing out that lesson, too) .. last week i spent all my gaming time playing freecell and hearts .. my backgammon brain had apparently been totally fried and needed a cold boot

part of this nickel and dime approach has also had the effect of re-framing the way i'm beginning to "see" the game .. it closely parallels your concept of a race to be closest to perfect play .. i'm so far from perfect my mind won't even allow me to toy with that concept long .. but as a result, i have far less emotional investment in an individual match outcome or dice roll, just as you've suggested .. maybe it's approaching the Zen of backgammon .. i can get my mind around that as a mindless mindfulness

i've also learned that some of my best thinking (i thought) involved minimizing risk by projecting opponents next dice rolls .. GNU and thinking of nickels and dimes in equity has taught me minimizing risk while maximizing builders is not at all the same as minimizing wasted equity .. indeed such decisions can lead to some really bad equity blunders and easily cost me a game or match .. i don't have better words than yours to describe that yet, but will continue to seek them and carefully re-read what you said quite well

... in place of thinking about the roll at hand I will offer an alternative in the form of looking at the 'dynamics of the position'  and what certain rolls can do to it.

*what type of a game are you in
*how volatile is it
*what's your 'game plan'
*how's your race
*how's your positional advantage or disadvantage
*how is the balance of blots or other vulnerabilities

These points can often give you a very good sense of what your equity is for a given position and if you have a choice of rolls and go through the checklist for the different resultant positions you can get a very good sense of the differences they create.

it's hard to put into words, for me, what i "see" or "sense" in this game .. i've been trying to get a grip on equity since i first heard it mentioned in a whisper while watching a tournament match nearly a year ago, as a matter of fact it was you that mentioned it .. metaphors of understanding and some kind of non-numeric uncalculable sense of equity which i can't begin to put into words are working, for now .. GNU seems quite happy to continue giving computations and voluminous detail and decimal places that help, not hinder or hide, this understanding

thanks again for a great reply and thanks to mweimer for starting this thread

sixty

[dorbel]

Both sixty and donz have given us some stimulating insights here. They demonstrate clearly that improvement at backgammon is only partially about acquiring new technical knowledge. Of much greater importance to 99% of us is finding ways that allow us to apply what we already know, in simple and practical ways that can be used over the board. Sixty's startling improvement in recent weeks is due entirely to application, both to analysis and approach to actually playing.
Note that self-analysis and study often leads to temporary setbacks in playing standard as sixty has observed. I believe this to be due to trying to apply principles that we have just observed to work in one situation to another position that, although apparently the same, is not. Stopping playing, or stopping studying, or both for a few days seems to cure this.

[don]

Nice discussion sixty-one.  But for clarification:

at the start of a game, your equity is essentially 0.50

is wrong.  Even odds (50% winning chance) is 0.0 in terms of equity.

--
don

[sixty_something]

right, don, I'd noticed that too, but hadn't found the words to describe why or a need to .. thanks for forcing the issue

the issue is mixing my coin metaphor, equity (itself a numeric metaphorical measurement), and probability .. yes, each player starts with the same equity and the starting point is zero .. each player starts with the same probability of winning the game, 0.500 .. i guess that is the source of my 50 cents reference .. but the point of my coin metaphor was not intended as a way of seeing either directly .. the coin of this realm is wasted equity

wasted equity, in the game record, is highlighted in color .. these equity coins simply become another way of seeing and measuring that .. for me, it's a simple way of visually interpreting GNU's endless sets of numbers and a visual aid for what donz describes as a race toward perfection

the nickels and dimes are intended to help understand GNU's marked rolls in the game record, that is wasted equity which GNU calls equity error .. i just like the connotation of waste .. but these coins, as tokens of equity, also turn out to be good for visually evaluating our stake in the game which is another way of saying equity .. like poker chips in front of a player across the table and in the pot in any given hand, they represent how we are doing in the game and what is at risk with each decision .. so, since we have a $1 wager in this metaphor, let's put a dollar in coins in front of each of us .. of course, with gammons and backgammons, we'd better have $2, even $3 in our pocket and then there is the cube!

with each dice roll we encounter potential equity change .. with each roll coins move from my equity stack to yours or from yours to mine as a function of whatever is rolled based on GNU's equity calculations  .. however, anything but the best choice, the perfect choice, means more coins than necessary have moved because of a weaker play than GNU with it's supernatural neural net collection of experience suggests

again, the coin metaphor was originally intended for wasted equity, for GNU's color coding of marked rolls highlighting wasted equity, not equity itself .. so, while true that each players equity starts at 0.00, it is equity waste i'm really trying to minimize and measure with the visual and tactile metaphor of coin .. with wasteful play, i unnecessarily give away coins from my stake or equity in the game .. a penny or two may not matter, but i don't want to give you too many nickels or dimes .. giving you quarters makes no sense at all

these equity coins aren't worth a plug-nickel in making supernatural move and cube decisions, except in making the whole concept of backgammon equity more comprehensible and in learning to use GNU .. for me, i was so lost in the blinding flurry of numbers, i was missing the whole point of GNU as an analysis tool

to extend or defend the coin metaphor further requires too much time and currency, both of which i'm short of at the moment, and probably isn't worth it .. but it has been fun thinking about

sixty

p.s. the core of beginning to understand GNU for post-game analysis, for me, came when understanding the markups in the game record with doubtful, bad, and very bad decisions or wasted equity .. unmarked decisions, for GNU newbies who have read this far, are not necessarily indications of unwasted equity .. unmarked decisions include best choices where our choice matches GNU's choice .. but unmarked decisions also include choices where equity change is less than the marking threshhold for doubtful at -0.040 in equity error, metaphorically we don't mark wastes of mere pennies

[sixty_something]

i found another handy way to use this equity coin metaphor in understanding GNU's wealth of statistics .. while reviewing a match with gBOT last night, a little light of understanding came on regarding GNU's Match statistics .. the Overall summary panel for all games in the 5-point match looks like this .. the percents are MWC units, as you can see it's comparable to equity, but we'll ignore MWC for now

                                             gBOT                 sixty_something     
Overall Statistics:
Error rate (total)                -0.743 ( -7.539%)    -1.632 (-17.242%)   [equity cents expressed in dolllars]
Error rate (per decision)**     -13.8 ( -0.140%)     -33.3 ( -0.352%)   
Wasted equity (per decision)      -1.38 cents            -3.33 cents
Equiv. Snowie error rate                -6.8                   -14.8             
Overall rating                          Intermediate           Beginner           
Actual result                            +50.00%              -50.00%           
Luck adjusted result                    +3.08%               -3.08%           
Luck based FIBS rating diff.          +16715.76                               
Error based abs. FIBS rating          1848.4               1534.8             
Chequerplay errors rating loss         191.1                498.6             
Cube errors rating loss                    10.5                 16.7             

[** - Note that the Error rate (per decision) is multiplied by 1000 in GNU's default settings for Snowie 4 comparability (i'm changing mine to equity cents, asap, but haven't found out how yet) .. GNU's error rating always comes out higher than Snowie's, due to details in the calculation but both are based on equity .. this difference, however, enables a meaningful extention of the equity coin metaphor .. i've converted Error rate (per decision) to equity cents and inserted Wasted equity (per decision)  lines above and below .. this makes more sense to me, but the Snowie error rating is the lingua franca of the realm for backgammon player ratings]

the Error rate (total) is the total of wasted equity for all marked moves and cube decisions .. for example

                                                                     gBOT                 sixty_something     

Chequer Play Error rate (total)                     -0.660 ( -6.070%)    -1.434 (-15.740%)   
Cube decision Error rate (total)                    -0.083 ( -1.469%)    -0.198 ( -1.502%)   
                                                            -------------------   -------------------
Error rate (total)                                       -0.743 ( -7.539%)    -1.632 (-17.242%)

so, gBOT gaveaway nearly 75 cents in potential equity and i gaveaway over $1.60 .. guess who won the match?

again, the percent values are the error or waste expressed in MWC (Match Winning Chances) which is GNU's other analysis metric .. if you like percentages MWC may be a better metric for you, but i like equity and most folks find it preferable, especially when first beginning to understand GNU analysis

one other metric makes sense now using the coin metaphor: Error rate (per decision) .. GNU's on-line manual says, "The error rate per decision is the total error rate divided by the number of non-trivial decisions (i.e., the sum of unforced moves and close or actual cube decisions). " .. here are the numbers from this match

                                                    gBOT               sixty_something     

Chequer Play Statistics:

Unforced moves                                   48                      40                 
Error rate (total)                      -0.660 ( -6.070%)    -1.434 (-15.740%)   
Error rate (per move)                -13.7 ( -0.126%)     -35.9 ( -0.394%)         -0.660/48*1000 = -13.7
Wasted equity (per decision)        -1.37 cents            -3.59 cents
Chequerplay rating                          Intermediate             Awful!             

Cube Statistics:

Total cube decisions                               57                     33                 
Close or actual cube decisions                    6                      9                 
Error rate (total)                        -0.083 ( -1.469%)      -0.198 ( -1.502%)   
Error rate (per cube decision)         -13.8 ( -0.245%)       -22.0 ( -0.167%)       -0.198/9*1000 = -22.0
Wasted equity (per decision)          -1.38 cents            -2.20 cents
Cube decision rating                         Intermediate         Casual player       

so, in the give and take of wasted equity or equity error in this match, gBOT is giving away almost a penny and a half per move and almost exactly the same per cube decision .. while i'm giving away three and a half cents per move and a little more than 2 cents per cube decision .. GNU's ratings are based on error rate ranges .. i'm not familiar with looking at the error rate enough to have a feel for the rating system, but giving away nearly 4 cents per move in wasted equity sounds pretty "Awful!" to me, too .. like i said, guess who won the match?

before you answer, one final set of equity numbers demonstrates how that might change .. GNU also tracks and tabulates luck

                                            gBOT                     sixty_something     
Luck Statistics:

Rolls marked very lucky                 3                               1                 
Rolls marked lucky                        0                               5                 
Rolls unmarked                            49                             45                 
Rolls marked unlucky                     2                               1                 
Rolls marked very unlucky              1                               3                 
Luck rate (total)                     +0.803 (+21.993%)    -1.273 (-24.931%)   
Luck rate (per move)                 +14.6 ( +0.400%)     -23.1 ( -0.453%)    -1.273/10*1000 = -127.3
Luck rate (per move)                 +1.46 cents             -2.31 cents              (does not compute)
Luck rating                                None                 Better luck next time

[Luck rate (total): The total luck for this game or match reported both normalised and unnormalised.
Luck rate (per move): The luck rate per move reported both normalised and unnormalised.
not sure why i can't get this to compute .. wonder if it's a bug or me?]

gBOT won the match, but it was due to a really lucky sequence and resultant amazing equity swing in the last game .. with the cube at 2 and match score 1-1, my cubeful equity was +0.927 when from the bar gBOT, with a roll marked very lucky by GNU, hit my slotted ace point in a 5-point home court .. this one roll resulted in an equity swing of +1.483 .. a following roll, marked unlucky, resulted in an -0.627 equity loss for me - no waste in either of those, just gain and loss of equity from the roll of the dice .. gBOT then gradually closed its 4-point home court, picked up a loose straggler, left me dancing on the bar for 10 consecutive rolls, and gammoned me by one roll winning 4 points and the match .. in the currency of equity that two roll sequence comes to over two dollars in equity change with the cube at 2 .. as it turns out that is almost exactly the difference in our luck rate above .. that's likely just a coincidence, but it's no coincidence equity tracks the status of a backgammon game so well .. but, as we all know, the last to roll of the dice wins and a pair of sixes can eat up your opponents equity fast

i played gBOT again this morning seeking revenge .. the difference in Match statistics can swing wildly from match to match .. this is a good example

                                                 gBOT                     You                 
Overall Statistics:
Error rate (total)                       -0.523 ( -6.735%)    -0.433 ( -4.536%)   
Error rate (per decision)               -10.1 ( -0.130%)       -8.7 ( -0.091%)   
Wasted equity (per decision)           -1.01 cents            -0.87 cents
Equiv. Snowie error rate                 -5.2                      -4.3             
Overall rating                             Advanced               Advanced           
Actual result                             +50.00%                -50.00%           
Luck adjusted result                     -5.80%                 +5.80%           
Luck based FIBS rating diff.          -2100.00                                 
Error based abs. FIBS rating           1868.3                   1818.4             
Chequerplay errors rating loss           181.7                    231.6             
Cube errors rating loss                        0.0                       0.0             

just had to show this one .. i lost the match with a legitimate backgammon in the second game with the cube at 2 .. it was a complicated game, but my luck rating was "Go to bed" .. when i lost the game, i felt like i must have played terribly .. GNU's rating was a nice pat on the back especially after being beaten with a backgammon

all this makes me remember the days when i was a kid and first learned to read baseball box scores, racing forms, and stock listings beginning to "see" and "sense" the inside levels of the game and market not too obvious on first glance .. it's nice to have such a tool for backgammon, too

equity cents are making even more sense to me .. thanks for reading all this, if you've made it this far .. let me know whether this coin metaphor has been worth all the bother .. also, should we play you'll know what my new greeting means:

"hey, mister, can you spare a dime?"

sixty

[blitzxz]

Note that self-analysis and study often leads to temporary setbacks in playing standard as sixty has observed. I believe this to be due to trying to apply principles that we have just observed to work in one situation to another position that, although apparently the same, is not. Stopping playing, or stopping studying, or both for a few days seems to cure this.

Here is a important point. I would say that the reason is that gnu doesn't tell what are you doing wrong. It just says you did horrible blunder but it doesn't tell why. The player has to interpret. And often the player (like myself) complitely misunderstands the nature of the mistake and later plays according to that misinterpretation. So when you see gnu suggesting weird looking move stop to think about it for a while. Exactly why was it wrong? Check pip count, position, match-score and most important what is likely to happen in the next couple of rolls. If you still can't figure out the underlying principle ask from somebody else or just forget it and move on.

[sixty_something]

gnu provides a really good way to better understand a difficult position .. i call it a playout .. dorbel was the first to recommend it to me .. he and others say it is a way used before the days when computer analysis tools like gnu to evaluate a position .. according to others, some worldclass players still use the technique today to analyze a position on a real board with real pieces

for me, a playout is, when from any position, one simply begins playing it over and over again (using gnu or a real board) until you can "see" the dynamics of the position and the reasons behind gnu's suggestion

it's quite easy to do with gnu by simply going back to the marked position and playing it again and again keeping track mentally or with pencil and paper of the ultimate outcomes .. while something similar to this can be statistically accomplished with a gnu rollout, i found that by slowly and carefully playing the position a dozen or two dozen times or more, i begin to "see" the folly of my original choice and why gnu has selected the move or decision i did not

now, if i could just learn to remember what i've learned from those situations and apply it across the board .. maybe the answer is returning to the habit and discipline of doing more playouts on difficult positions .. i've found recently having stopped the serious studying that got me to 1800 for the  first time about six weeks ago that my level of play is off and i'm back down to the mid to low 1700's

hmmm, not so simple after all