[rhm]

Cthulhu D, on 2014-September-30, 18:22, said:

Sure, but in this case I think you are drawing conclusions from inappropriate data sets. Many boards in the 1NT vs Pass analysis are people opening 9 HCP hands or 10 HCP hands. Those are not interesting when the question is 'do we open balanced 11 counts' you must go into the underlying data - which significantly impacts the sample sizes, and makes the conclusion very questionable - to consider only the 11 HCP hands.

Similarly for the 1C vs Pass analysis, again, we need to consider only balanced 11 counts because that is the question we are asking (is it right to open balanced 11 counts). There are a bunch of 9 and 10 counts and some unbalanced 11 in there, and they need to be tossed to answer the question.

Just using Pavlicek's top level numbers for this will not work because the underlying datasets do not align with the question we are trying to answer.

My limited analysis of the dataset is favourable for opening balanced 11 counts (but not 10 counts! Important!). I could be wrong. I would like a more comprehensive examination. I would also be interested to know the cost of opening a short 1C vs a better minor 1D and the benefit from an unbalanced 1D (which is very hard to answer). I would also love to see what the numbers on a 11-13 1NT looks like - it is very possible that is the 'optimal' (airquotes) NT range.

I'm (relatively) confident in this analysis because it reflects my reflects my at the table experience, but I could very well be wrong.

I find RP statistics interesting because they are taken form real life at top level encounters.
Unfortunately they are hard to interpret in the sense that you like to draw Bridge conclusions.

For example:

Pavlicek provides a statistic over 258 hands from 1996-2012, where one room opened 1♠ and the other passed initially. Overall there is a "win" for the room, which passed initially.

Pavlicek's conclusion:

"Opening light in spades, however, leaves little doubt as being a losing strategy. Why so? Opening light in any suit makes constructive bidding less accurate due to the wider range of opener’s hand. This is offset by the advantage in bidding first. When you hold the highest ranking suit, the advantage in bidding first is minimal (you can usually bid later) and the detriment to constructive bidding is maximal (fewer bids remain for exploration). At least that’s my take on it."

I looked over the hands:

Some 1♠ were outright psyches, usually a spectacular loss, the psyche probably due to the state of the (KO)-match at this point.
If you psyche, 1♠ seems to many more attractive than any other bid.
Even if you discount psyches you will see hands, which contain only rubbish and which few would open, and others which you would expect anyone to open, at least nowadays, and you get a real surprise to see someone decided to pass on them.
In other words the hands under analysis vary a lot in strength by any sensible evaluation method. You start to ask yourself, what is "opening light" actually?
I saw 1♠ openings (not psyches), which were apparently completely artificial, having nothing to do with length in spades etc.
Some of the deals ended in the same end contract (not necessarily spades or declared by the side which opened 1♠) even though only one room initially passed.
The room, which made a trick more counted as a win. Sensible or not is hard to say, since the bidding was of course different. (The IMP result depended on whether this trick was an overtrick or broke the contract)
There were hands with an obvious bidding misunderstanding.

Apparently Pavlicek's methodology is not to filter any hands to avoid any statistical bias.
However, this makes the result not any easier to interpret in Bridge terms and I am not convinced that Pavlicek's conclusion above really holds.
I do have doubts

Rainer Herrmann

[NickRW]

Cthulhu D, on 2014-September-30, 18:22, said:

I'm (relatively) confident in this analysis because it reflects my reflects my at the table experience, but I could very well be wrong.

Well, if we're going to get anecdotal, over the years I've played in 3 partnerships where we took completely the opposite view - i.e. go (very by modern standards) conservative with balanced openers i.e. open only the 13s and good 12s. I haven't added it up, but I'm pretty sure in the cases where the deals are passed out, we have been ahead. Also in the cases where we have a constructive auction, the advantage for responder is obvious. But...., in the competitive auction, we're probably not ahead - which possibly makes our strategy playable at imps and less so at pairs.

I'm not saying you're wrong at all - just that I need a lot of convincing

Nick

[Cthulhu D]

rhm, on 2014-October-01, 05:06, said:

I find RP statistics interesting because they are taken form real life at top level encounters.
Unfortunately they are hard to interpret in the sense that you like to draw Bridge conclusions.

Apparently Pavlicek's methodology is not to filter any hands to avoid any statistical bias.
However, this makes the result not any easier to interpret in Bridge terms and I am not convinced that Pavlicek's conclusion above really holds.
I do have doubts

Rainer Herrmann

This is why I reviewed the samples for 11 HCP balanced vs Pass only, but this has a seriously deleterious impact on sample size. I feel like it is a winner, but I would love a better study.

It's worth noting that he did exclude the psyches and unbalanced hands from the 1NT opener, but not on the suit openers which is a shame.

I feel that it is a winner, but it has huge secondary system impacts and if it's *not* a winner I would change my agreements extensively, so I would love to know one way or another. There is a more general tactical thing here as well - It seems like that generally opening is better than not opening, but again, not enough evidence to prove that statement. I play a variety of things to maximise the % of hands my regular partnership opens, and I would be fascinated to know if they are winners.

[PhilKing]

It's entirely possible that just counting your Milton high card points and opening the bidding when they come to 11 is not a winning tactic.

Not all 11 counts are created equal. On the occasions that one player opened and the other passed (in the Pavlicek data), there is a significant chance that judgement and not system was the reason, so going by the raw stats is not the way to go, although the stats prove that opening a marginal 11 can be a winning tactic.

It might be interesting to see the effect of vulnerability on the data, but it's hardly rocket science to suggest that non-vul is the way to go.

[Cthulhu D]

Well yeah, Milton has limitations and your decision making needs to be more insightful - but it's not bad at evaluating the trick taking power of balanced hands so I feel like for the purposes of discussion it's not a terrible yard stick.

I'm not totally convinced that it's good non-vul and less good vul - but I haven't checked. Based on OKBridge analysis that the danger contracts are 3H, 3S and 2NT, my theory would be that it enables less invitational bids because partner is less likely to have an invite opposite a 11-13 than opposite a 11(+)-14 or 11(+)-14(-). I'd probably need to spend some time with Dealer.exe to check, but I'm pretty sure that is the case )I'm focused on 1C - Balanced 11 here, it's very possible that 1S with a 5-3-3-2 hand is just bad). It might all be given back with more frequent invites opposite the 14-16 range as well.

I don't think the penalty concern is that probable at the table but I have no idea how to check that. I'm trading on personal experience there.

[steve2005]

I suggest NT range split of 12-14/15-16/17-19. Yes REALLY nice balanced 11's are also opened. With a 2 pt range for 1N you wont need many INV bids which seems to be all the rage.

This mean 1♣ is opened as much as possible where you have your best bidding system. The 17-19 range occurs less often 12.7% compared to 15-16 5.9%.

I find 17 pt hands are underrated often worth upgrading.

12-14 is 36.8% ignoring 11's.

not sure if numbers are right , but their relative order is.

[fromageGB]

steve2005, on 2014-October-04, 07:59, said:

I suggest NT range split of 12-14/15-16/17-19 ... With a 2 pt range for 1N you wont need many INV bids which seems to be all the rage.

I have played this for years and if you do have a sequence where opener needs to rebid 1NT with 12-14 and 2NT with 17+, I think you need something funny to split the 19 off from 17/18. My sequence is when responder has denied a 4 card major, so I treat the reverse in hearts - 1♣ - reply - 2♥ as a sort of Kokish demanding 2♠ from partner. Now 2NT is the 19 hand, with normal continuations, while 3♣ shows the strong x4x6 shape.

[steve2005]

fromageGB, on 2014-October-04, 09:08, said:

My sequence is when responder has denied a 4 card major, so I treat the reverse in hearts - 1♣ - reply - 2♥ as a sort of Kokish demanding 2♠ from partner. Now 2NT is the 19 hand, with normal continuations, while 3♣ shows the strong x4x6 shape.

like, but not sure u need to show 4♥if p denied. do you mean you bid 3♣ with 4♥ nut dont promise 4♥?

[fromageGB]

steve2005, on 2014-October-05, 14:42, said:

like, but not sure u need to show 4♥if p denied. do you mean you bid 3♣ with 4♥ nut dont promise 4♥?

What I meant was that if I had a 17+ x4x6 shape I would make a natural reverse of 2♥. I would also bid 2♥ with a balanced 19 count. Partner bids the 2♠ relay, and with the balanced hand I bid 2NT and all normal from there - game, slam, minors or whatever. However, with the reversing hand I bid 3♣ over his 2♠ which tells partner the shape, even though I know there is no chance of a heart fit. This allows him to make a better judgement of whether his holdings are suitable for a NT contract. If not, we play in clubs at some level.

[awm]

I really would not trust Pavlicek's data for this sort of analysis. It has a lot of selection bias, by which I mean the following:

Take the 1nt vs Pass examples. Many of the 1nt openings on 11-counts are hands where an expert chose to "upgrade" into a 12-14 range. These are likely to be much better than average 11s. It follows that this data tells almost nothing about the merits of opening average balanced 11s, nor does it say anything about the merits of an 11-13 range.