Hello Robot Battlers,

My bosses are big believers in performance metrics. Their thought is that you should be able to measure everything that is job related, such as the number of defects closed, the number of customer service cases worked, and so on.

When it comes to robot battles, a change in what system you play should have an effect on your long-term average percentage score, other things being equal. If I was to play 5000 hands playing a certain system, and 5000 hands playing a different system, the skill and luck factors should cancel out over that long of a time.

I started keeping track of net monthly average score in October. My preferred game is matchpoints but you could do the same thing with IMP scoring.
October 2018: 60.6556
November 2018: 60.1736
December 2018: 60.2520

January 2019 is not complete but it is 60.2230. Over this four-month period I played over 5000 hands using pretty much the same system. As expected, playing a consistent system yielded consistent results over a large sample size.

Starting February 1 I am going to experiment with opening notrump ranges:

With 10-12 HCP I will open 1 of a suit and rebid 1NT (or pass a 1-level suit bid if suitable)
With 13-15 HCP I will open 1NT. (currently 14-16)
With 16-17 HCP I will open 1 of a suit and rebid 2NT (currently 17-18)
With 18-20 HCP I will open 2NT. (currently 19-20).
With more points I will open same as before.

The next ACBL robot championship is scheduled for March 16-18, 2019. Hopefully I will have enough information to see if this system change is effective before then. But I am cutting back on robot play this year, to two sessions a day (720 hands a month), so I may not be able to tell one way or the other before then.

Cheers,
Daniel