comparing systems
#21
Posted 2013-January-15, 18:09
#22
Posted 2013-January-15, 20:18
Edited to add: forgot I am playing Unassuming Club with one partner now. Please don't take my bad bidding in that system the wrong way, as I am new to it
#23
Posted 2013-January-17, 19:57
<given NS hands>
If your system response is 1N or higher,
If your system response is 1♠ or lower,
#25
Posted 2013-January-19, 00:40
#26
Posted 2013-January-19, 04:16
straube, on 2013-January-19, 00:40, said:
Good hands should pass, especially ones with lots of defense like this. they are going to have a worse than field part score auction unless you bail them out. Pass and lead a club against 1N.
#27
Posted 2013-January-21, 05:46
straube, on 2013-January-11, 10:55, said:
I'm just catching up on the forums after christmas/new years, but I'm interested in this. One thing I'd caution is how you pick hands. I'd deal them randomly selecting all hands that have a strong club for opener and then allow the rest of the hands to fall as they may (including negative responses and opponents interference). Otherwise you get false relations like my system handles the goulash and freaks but falls apart on part scores and interference but the former are "more interesting" so it scores better than it should on this sort of research project.
#28
Posted 2013-January-21, 07:50
Mbodell, on 2013-January-21, 05:46, said:
I look at deals always starting with the North hand. If it has a one club opener, then I pick it. If it opens something else I toss the hand. If it passes, I look at the East hand and also toss the deal if it does something other than pass or 1C. Etc. That's how I'm getting 1C for various seats. I also toss the deal if after opening 1C the next hand does something other than pass. I think there are chances for deal selection bias here but that they're pretty small because I'm really not looking at how the auction might proceed until I've decided whether the deal is suitable or not.
Personally what I'm most interested in which initial responses to a strong club prepare the partnership best for a future contested or uncontested auction. Obviously some systems have more gadgetry than others and I'm also interested in this, but (for example) knowing whether 1D should be 0-7 or 2-way (good or bad hand) or GF any or GF some etc is my primary interest.
#29
Posted 2013-January-21, 11:45
#30
Posted 2013-January-21, 12:49
a.k.a. Appeal Without Merit
#31
Posted 2013-January-21, 12:51
(1) There have been no double negative responses, even though sims indicate they are about 20% of responder hands.
(2) While there were a few occasions where not everyone reached game, there have been no hands where (some) game wasn't good.
Am I wrong about these? They seem to suggest some skew, although perhaps it's something about the model for opener's LHO interfering?
a.k.a. Appeal Without Merit
#32
Posted 2013-January-21, 13:07
awm, on 2013-January-21, 12:51, said:
(1) There have been no double negative responses, even though sims indicate they are about 20% of responder hands.
(2) While there were a few occasions where not everyone reached game, there have been no hands where (some) game wasn't good.
Am I wrong about these? They seem to suggest some skew, although perhaps it's something about the model for opener's LHO interfering?
I chalk it up to a small sample size. I've noticed game/slam on every hand but I'm using the methodology I described and looking at BBO random deals.
#33
Posted 2013-January-21, 13:41
straube, on 2013-January-21, 13:07, said:
I would recommend using a Dealer script for the hand generation and posting it to make everyone agrees that it's representative. It may make sense to build in some bias into the script (like eliminating balanced 16-18 opposite balanced GF).
If you aren't comfortable with the syntax, many of us can help...
#34
Posted 2013-January-26, 16:58
#35
Posted 2013-January-26, 18:19
straube, on 2013-January-26, 16:58, said:
See my post in one of the threads where I always will do this, although often by editing the auction within 5-30 minutes of posting it.
#36
Posted 2013-January-26, 21:34
I hope to use your deals to practice and refine Jasmine ♣, outlined here
IMO, after posting an auction, contributors should take the time to rate possible final contracts at the form of scoring that Straube specifies. I don't think that double-dummy analysis, using the actual East-West hands is important. Much more interesting would be the results of computer-simulation using random defending hands by those with access to appropriate software.
I will edit my posts to give marks out of 10 (a subjective estimate).
But it would be great if somebody could compute, for each contract reached,
- At pairs, the likely percentage score
- At teams, your expectation in imps.
#37
Posted 2013-January-27, 23:44
#38
Posted 2013-January-27, 23:50
relknes, on 2013-January-27, 23:44, said:
I think you're right. At this point I'm just mixing 3rd/4th deals in as they occur, but we are using an entirely different 3/4 structure and I think Justin is as well. If you use a different 3/4 structure, just identify it as such and when I post outcomes you can cross compare them with others.
#39
Posted 2013-January-28, 01:18
straube, on 2013-January-27, 23:50, said:
Our system varies, but only slightly, by seat and vulnerability (which is why I explain 1♣ differently on some hands), but the continuations over 1♣ are basically the same (some point ranges shifted slightly). Basically when 1st and 2nd red, and 4th always, 1nt is 12-14 and 15-17 balanced goes in 1♦, when 1st and 2nd white, and 3rd always, 1nt is 10-12 (10-~13 in 3rd), and 13-15 balanced goes in 1♦. The non-1♣ auctions changes more for us between 1/2 seat and 3/4 seat, but these shouldn't really come up in this series.
#40
Posted 2013-January-28, 13:01
nige1, on 2013-January-26, 21:34, said:
I can do the simulations (DD with random E/W hands) but I think this would miss an important point, which is information leaked in the auction. What would be more convincing would be DD with random E/W hands and single dummy leads (I remember reading somewhere that the opening lead was statistically the single play with the largest deviation from DD-optimality, which sounds reasonable), but this would require writing down all the information given in each auction -- a bit too much work for me right now.