You are ignoring the author of this comment. Click to temporarily show the comment.

We should have the tricks for game, but can we avoid four losers? Depends on partner's cards, obviously, but I think the odds are that our cards fit well.

You are ignoring the author of this comment. Click to temporarily show the comment.

While the hand has fair playing strength, good potential and three first-round controls, it lacks the defensive strength to penalize even a five level bid, and most high cards partner is likely to have will be worthless if we declare. There is little risk of being passed out at 1♥, so the challenge is how best to describe your hand to partner or gain his cooperation in exploring slam.

You are ignoring the author of this comment. Click to temporarily show the comment.

Just for drill I looked up some Vugraph results on BBO. The most relevant available in the recent matches were the six segments from the GNT semi-finals and finals for Kit & Co. That gave me 90 hands. I entered the IMP results from these 90 boards (without regard to who went plus or minus) and “bootstrapped” 1000 simulated 15-board segments, randomly selecting from the 90 results and randomly scoring as plus or minus for the team trying to play catchup. They gained 77+ IMPs once in 1000 segments. Next I substituted a copy of the “fireworks” GNT Finals 2 for the “boring” Finals 1 segment and they won 4 in 1000.

I had been simming 25 segments at a time, repeated 40 times; now I bumped it up to 1000 at a time and repeated 10 times. Using all six segments resulted in 9 come-from behind wins out of 10000; doubling Finals 2 yielded 27 out of 10000; using nothing but Finals 2 yielded 186 out of 10000. I also ran the doubled F2 but dropping the behind team to a 45% chance of going plus on each board and they won 14 out of 10000.

So even the most favorable case I tested, bootstrapping just the wild Finals 2 segment and giving the catchup team a 50% chance to win each swing, yielded less than a 2% chance of winning. Of course we'd like actual way-behind-in-the-final-segment data, but I'd be surprised if it yielded anything wilder than that Finals 2 segment.

You are ignoring the author of this comment. Click to temporarily show the comment.

I tend to open 3♣ on hands that would open a weak two if the long suit were anything else, especially when not vulnerable, but I wouldn't be eager to open this collection in any strain. I probably would if playing Ogust, but no such safety valve exists for 3♣.

You are ignoring the author of this comment. Click to temporarily show the comment.

Does strong mean game-forcing? If so, then 2♥ looks like a standout, showing game-forcing values and hearts at a conveniently low level. If the bid were passable, 2♠ must be right. I usually play this double as showing hearts, but that makes sense only for those of us who allow an up-the-line 1♦ with hearts and less than game values.

You are ignoring the author of this comment. Click to temporarily show the comment.

Is this actually a limit raise? For me a limit raise is 11 to 12 in support, or 3.5 cover cards. The description as given could easily include a flat 9 or 10, or 3 cover cards, with a strong likelihood of wasted values in spades. I voted for game under the assumption this bid was the limit raise.

You are ignoring the author of this comment. Click to temporarily show the comment.

This shape is awkward for 1♣ if we have to rebid the five card minor rather than spades. So I'll just open 1♦ in either system, expecting to rebid or jump raise the spades.

You are ignoring the author of this comment. Click to temporarily show the comment.

Need a simulation. A possible approach would be to find other fourth-quarter results by teams way behind as the base for the swinging strategy (if it appears such teams did, in fact, swing) while using first-quarter results for the non-swing approach. Ideally those might be from the same team in a match expected to be close. While selecting appropriate data would be somewhat subjective, the simulation randomization should make the results fairly robust.

You are ignoring the author of this comment. Click to temporarily show the comment.

And the team playing catch-up will likely be taking anti-percentage actions, so the probability of success is surely worse than a coin-flip. Assume, for example, a 40% success rate – the odds become 1/6103. Also, your 1 in 4 chance of 80+ IMPs includes cases where even more IMPs are available, and more than 8 swing hands. If 9 swings are available, they have to get them all, or luckily avoid one swing, or there have to be so many IMPs they can blow one and still gain 78.

All such estimates depend heavily on many assumptions. I think the best approach would be a bootstrap simulation, but the appropriate sample would be other fourth quarter results from teams down a similar amount in major events, rather than the first three quarters of this match.

You are ignoring the author of this comment. Click to temporarily show the comment.

Right, the assumption that the mean and variance would be the same for the final quarter is not warranted. It would be interesting to analyze a large number of such matches.

You are ignoring the author of this comment. Click to temporarily show the comment.

Seems like that is covered by the strong option – the description does not specify “would've overcalled NT if we were playing that as 18-20.” The crucial point is the strength implied.

You are ignoring the author of this comment. Click to temporarily show the comment.

I used canape with the majors, hearts first on 5-4, spades first on 4-5. I also found the 1♠ relay on 0-7 to be very inefficient, so restricted it to 0-5 and defined other bids for 6-7 hands. After 1♣-1♦; 1♥: 1♠: 0-5 1nt: 6-7, 5+ spades 2♣: 6-7, 4+ clubs 2♦: 6-7, 4+ diamonds 2♥: 6-7, 4+ hearts 2♠: 6-7, exactly 4333 (four spades)

Note that hands with four spades and another suit bid the other suit first, while hands with 4+ hearts bid those first. Opener rebids 2NT with the big balanced hand, of course, or jumps to 3♠ over 1NT with spade support.

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

I had been simming 25 segments at a time, repeated 40 times; now I bumped it up to 1000 at a time and repeated 10 times. Using all six segments resulted in 9 come-from behind wins out of 10000; doubling Finals 2 yielded 27 out of 10000; using nothing but Finals 2 yielded 186 out of 10000. I also ran the doubled F2 but dropping the behind team to a 45% chance of going plus on each board and they won 14 out of 10000.

So even the most favorable case I tested, bootstrapping just the wild Finals 2 segment and giving the catchup team a 50% chance to win each swing, yielded less than a 2% chance of winning. Of course we'd like actual way-behind-in-the-final-segment data, but I'd be surprised if it yielded anything wilder than that Finals 2 segment.

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

All such estimates depend heavily on many assumptions. I think the best approach would be a bootstrap simulation, but the appropriate sample would be other fourth quarter results from teams down a similar amount in major events, rather than the first three quarters of this match.

Paul Hightower

Paul Hightower

Paul Hightower

Paul Hightower

1♠: 0-5

1nt: 6-7, 5+ spades

2♣: 6-7, 4+ clubs

2♦: 6-7, 4+ diamonds

2♥: 6-7, 4+ hearts

2♠: 6-7, exactly 4333 (four spades)

Note that hands with four spades and another suit bid the other suit first, while hands with 4+ hearts bid those first. Opener rebids 2NT with the big balanced hand, of course, or jumps to 3♠ over 1NT with spade support.