This article originally appeared in the July 1999 issue of GammOnLine.|
Thank you to Kit Woolsey for his kind permission to reproduce it here.
Almost everyone has experience with players who "steam" when
behind on the score sheet. My own living room was positively
moist after I dropped 73 points this evening. However, all
of this time on the outside of the chouette was really beneficial,
because it gave me a chance to consider something that I think
until now has received far too little attention: proper
These days players in the know can calculate doubling windows and doubling doors, market losers and market crashers, and so on ad nauseum. They roll out problems thousands upon thousands of times, memorize match equities and reference postions, and use massive computer power to scientically study whole classes of positions. But in all this new work, I have not seen one article addressing the best way to steam.
In a flash of insight I realized that there ought to be some way to mathematically define how you should change your cube actions based on the score sheet. The most obvious candidate for these adjustments is racing situations. This was a good place to start since these are also the most scientifically studied positions for even money scores, where good mathematical formulae are available.
After some experimentation, the proper adjustment factor was found: the square root of your deficit on the score sheet. The square root was suggested by Joe Freedman, a widely acknowledged genius in the field of steaming. His past experience must have led him to this brilliant suggestion, since on the evening in question he found himself up over a hundred points.
This adjustment is very useful and easy to apply. Let's say you lead a race by two pips, 100 to 102. Down 36 points on the score sheet, you make a 6 pip adjustment, and you have the requisite 8 pip lead for an initial double. If you are doubled down 20 pips in a 100 pip race, you can take as long as you are already behind 50 or 60 points.
For attacking and priming positions, I believe a different adjustment is called for. Here, you should take the logarithm (base 10) of the number of points you are down, and use this number to adjust your or your opponent's board strength in an attacking position, or your or your opponent's prime strength in a priming position.
For example, as soon as you find yourself 10 points down, you can consider your own board to be 1 point stronger anytime you are thinking about doubling. When you get 100 points down, you can subtract two points from your opponent's board strength when you are doubled. Making this adjustment, you will find that you often have a take even when closed out. In the unlikely event that you find yourself 1000 points down, you should usually double after the first roll, since with the six and eight points you already have a broken five prime.
I have found these adjustments very helpful in the short time I've been using them. I would very much like to hear if others try them out, or have any improvements or other ideas on steaming scientifically.