HCPs for game

[bluecalm]

25hcp is good 3NT. 2 tens are almost 1hcp when it comes to NT play. Nines are good but not nearly that valuable.

 

When it comes to suit games you want to be in one with 24hcp in general and with 22pc on 5-4 fit if it's not 4-3-3-3 on 4th trumps side. Singleton with 4+trumps is worth a lot: every time you have trumps 5-5 and singleton somewhere you want to be in game and every time you have void and 4trumps it also applies. Singleton with 4 trumps is good, you don't need many hcp for a game here.

If you have 5-5 and partner has 4 trumps you want to be in game too. Maybe 21hcp is what suffice here in general but you often make on less. 5-5 opposite 3 trumps is not nearly as valuable.

 

Bid a lot of games, try to make. Don't despair if the they don't make, take notice if the opponents let you make but could have beaten it. That will happen a lot.

That's it more or less. Rest will come with experience.

[inquiry]

i doubt this will solve any questions, but here goes... I did a bridgebrowser study of this with the following restrictions. The contract was 3NT, the declaring side held either 24 hcp (first three columns), 25 hcp (columns 4-6), 26 hcp (columns 7-9), or 27 hcp (the last three columns). The first column of each group is the number of tricks won, for instance of the 8,486 3NT contracts with 24 combined HCP, two people managed to win only two tricks. and five times only 3 tricks were won.

 

To make it easy to see, i summed all the percentage of contracts that won 9 or more tricks for each of the 4 HCP examined. The results were

 

24 hcp = 49.07%

25 hcp = 61.68%

26 hcp = 72.84%

27 hcp = 86.20%

 

Bridgebrowser will not let you drill down to tens, but we can try over constraints (maximum in one hand or the other, effect of one or more long suits, starting at five, etc). The usual restrictions apply... this was random online bridge events, there were great declarers on some hands, crappy declarers on others, great defenders on some, and total novices on others. The assumption you have to make is the horrible declarers and the horrible defenders cancel each other out.

 

 

 

 

http://3.bp.blogspot.com/-C9YgoBXJwXM/UCczeUagz-I/AAAAAAAAAKo/w2SO7TZBtbc/s1600/stats.JPG

 

[EricK]

Suppose a pair actually does count tens and 9 in HCP - T=1/2; 9=1/4 (and so use a 43 point deck), what are the % now for 24, 25, 26 "HCP" games (where these are based on the 43 HCP total)? Would this make for more accurate bidding?

[bluecalm]

I remember doing something similar for elite players from vugraph hands and results were slightly above 50% for 24hcp but it was difficult to isolate other factors (good suits, tens etc.) - the problem is that not all 24hcp games are bid and that introduce bias.

[Cyberyeti]

Suppose a pair actually does count tens and 9 in HCP - T=1/2; 9=1/4 (and so use a 43 point deck), what are the % now for 24, 25, 26 "HCP" games (where these are based on the 43 HCP total)? Would this make for more accurate bidding?

I'd do it slightly differently as average spots would contain around a 10 and a 9, so only adjust for less/more than that and that still allows the 40 point pack to operate.

[inquiry]

Suppose a pair actually does count tens and 9 in HCP - T=1/2; 9=1/4 (and so use a 43 point deck), what are the % now for 24, 25, 26 "HCP" games (where these are based on the 43 HCP total)? Would this make for more accurate bidding?

 

Can't help with this... with bridgebrowser, you have a lot of data you can mine, but it only handles AKQJ in data searches... even combinations of those in specific suits.. but not tens or nines.

[TimG]

Mikeh, I thought statistical analysis showed that for balanced hands 4-3-2-1 is pretty close to optimum evaluation, no?

 

I once did a DD analysis for two balanced hands in the game range and found that 4-3-2-1 worked as well as or better than a number of alternate evaluation methods. At least one of the evaluation methods took into account Tens, but none took any other "spots" into consideration.

 

I don't claim my analysis was perfect (and don't remember the exact details), but am comfortable with the conclusion that 4-3-2-1 is just as good as anything when evaluating two balanced hands for notrump play where the high cards are distributed such that one hand might open 1NT (either 12-14 or 15-17) and there are about enough points for game.

[han]

I recently saw either Fantoni or Nunes open 1♠ (15+, forcing 1 round) on something that looked like K109xxxx xx void AKQx (I am approximating).

 

I completely agree with the point you were making. But because there has been something discussion on the BBO forums about the Fantunes system lately, I'd like to clarify that (as far as I know) 15+ for 1S is a bit of an exaggeration. Gerben described it in his old blog as:

 

Their 1-level opening bids are all very solid (14+ HCP or compensating distribution)

 

Which corresponds to what I thought. My impression is that they would open 1S with as little as AKJxx x AJ10x xxx.

[awm]

I remember doing something similar for elite players from vugraph hands and results were slightly above 50% for 24hcp but it was difficult to isolate other factors (good suits, tens etc.) - the problem is that not all 24hcp games are bid and that introduce bias.

 

This... but also the bridgebrowser stats given make no account of shape. I'd expect that 24 hcp 3NT when one hand has a six-card or longer suit would be an excellent contract, and that these are bid a lot more routinely than 24 hcp 3NT on two 4432 hands.

[bluecalm]

I think I accounted for shape (analysing only balanced hands). I can't remember now for sure though. If someone is interested I could try doing it again.

Also if you know Python and want to play with it it should be easy to use my old project.

[inquiry]

This... but also the bridgebrowser stats given make no account of shape. I'd expect that 24 hcp 3NT when one hand has a six-card or longer suit would be an excellent contract, and that these are bid a lot more routinely than 24 hcp 3NT on two 4432 hands.

 

3NT contracts (in a bridgebrowser database) with a combined 24 HCP and a six card suit in at least one of the two hands, made 62.59% of the time (compared to 49.07% with no six card suit). BTW, the first study was for balanced opposite balanced, with 5332 and 5422 called balanced but not 6322. This study required one or both hands to have a six card suit (and no seven card suit).

 

0        1    0.03%
1		
2		
3        2     0.60%
4        9     0.26%
5      58     1.48%
6    199      5.03%
7    490    12.40%
8    734    18.58%
9  1177     29.68%
10  798     20.13%
11   381      9.62%
12   102      2.73%
13     14       0.43%
=================
9+           62.59%

 

[jh51]

For game in a major, tricks, not point count, are what's important. I recently bid (and made) a small slam on just 26 HCP. Having a no-loser 9 card trump fit and a running side suit really helped. With the defense I got, I was on a finesse for an overtrick.

[Antrax]

Congrats.

[Cyberyeti]

For game in a major, tricks, not point count, are what's important. I recently bid (and made) a small slam on just 26 HCP. Having a no-loser 9 card trump fit and a running side suit really helped. With the defense I got, I was on a finesse for an overtrick.

Double fits make a huge difference at slam level, at game level in a suit, recognising crossruff opportunities and double fits can reduce the number of points you need.

 

The thread is much more about no trumps I think, and even then long suits distort the numbers required, a solid 6 card suit and 3 aces is plenty.

[NickRW]

Suppose a pair actually does count tens and 9 in HCP - T=1/2; 9=1/4 (and so use a 43 point deck), what are the % now for 24, 25, 26 "HCP" games (where these are based on the 43 HCP total)? Would this make for more accurate bidding?

 

It definitely does help. I haven't used your exact scale - but you'd want to be in 3N with 26 on your scale - 25.75 probably too - 25.5 would be cutting it a bit fine. But it depends a bit on form of scoring and vulnerability as well of course - I guess 25.25 would still show a long term plus vul at imps.

 

Most people don't want to be bothered with this level of detail. The idea that two decent 12 counts are enough for 3N is simpler for people to think with.

 

Don't quote me on the figures - as I said I haven't worked with that exact scale

 

Nick