Matchpoint averages by seat

[bd71]

Stats for my last 631 matchpoint boards on BBO. About 80% of these hands are in ACBL robot tourneys where I get the best hand, explaining the disproportionate number of hands I am declarer.

 

Position - Instances - Avg. MP - % Contracts Made

Declarer - 317 - 51.4% - 65.6%

Dummy - 139 - 55.1% - 75.5%

Lead - 66 - 46.0% - 75.8%

Defense (no lead) - 96 - 54.6% - 47.9%

Passout - 13 - 50.7% - N/A

Total - 631 - 52.1% - 66.9%

 

Questions:

1. What should I make of the higher matchpoint scores when I'm dummy? I suspect others often try to engineer the bidding in individual ACBL tourneys so GIB is not declarer, which I do NOT do...could it be that this leaves me in better contracts and produces better scores? Or is it just that I treat GIB better than others and he plays more solidly with partner's active support?

 

2. Am I really bad on opening leads? My stats knowledge is rusty, but I think the 95% confidence interval for my "on lead" MP% is about 40-52%, just outside the overall average of 52.1%. So I think I can conclude with 95% confidence that I'm a bad opening leader.

 

3. Should I be surprised that my GIB partner makes >75% of his contracts and I make only 66%? Since I have stronger hand all of the time (in the 80% of the hands that are ACBL individual), shouldn't I have a natural advantage over GIB in that my strength is concealed when I declare? Is this grounds to conclude I am a worse declarer than GIB?

 

4. Any other takeaways from these stats that I can use? Have others done similar analysis and found it useful?

[rogerclee]

1) I think your higher scores as dummy reflect that bidding is a strength of yours.

2) We can't really infer too much about your scores as a declarer, except that maybe you are slightly worse at it than other areas of the game.

3) I think % of contracts made is not a good measure to relate to your skill as a declarer, since auctions where GIB is declarer can be very weird. Avg MP score is a much more reliable statistic.

 

BTW your 95% CI on opening lead is consistent with a standard deviation of about 24%.

[jdonn]

You're better when partner declares than when you do. You're better (by a lot!) when partner leads than when you do. I think we can simply conclude your partners (gib?) are better than you are.

 

I have noticed over time gib is quite effective on lead. I suppose running sims based on what we know about the hands of the opponents is a great way to decide a lead.

[Jlall]

You're better when partner declares than when you do. You're better (by a lot!) when partner leads than when you do. I think we can simply conclude your partners (gib?) are better than you are.

This is not a good way to look at it in robot races

[Stephen Tu]

I don't think looking at MP% alone can tell you whether you are worse than GIB at declarer play or not. It's hard to separate out the effects of the bidding, presumably GIB declares approximately equally as well at all tables, so GIB's MP% score is only really indicative that your bidding is better than the field, the contract is reasonable on average so GIB makes it etc.

 

Your own poorer score when declaring might be due also to overbidding not just inaccuracies in the play.

 

To tell whether you are good declarer or not you just have to look at your result vs. the field in same contract, are you consistently taking as many tricks as the best humans or not. BBO is a lot better than live in that you can see card by card how each of your opps in the tourney play. On hands where you take fewer tricks, you should examine the lines of the people who took more tricks on same lead. See what they did differently, try to figure out if their line is better than yours or not, whether there was superior technique involved, or they just got lucky on a poorer line, or if they just guessed some two-way hooks better etc.

[Jlall]

Yeah I mean gib sits in gibs seat, you sit in the humans seat. So if everyone in the field is a 10, but youre a 9, you will be below average usually.

 

Meanwhile all the gibs can be 2s and if you get your gibs to better contracts it will have a higher average.

 

So saying that OP is worse than gib is not necessarily true.

 

Also, how much you hog changes things a lot. Roger and I play a lot of these together and he hogs far more than I do, so as a result I would expect his gibs to have a higher average than me, and I would expect him to have a lower average than me. This does not mean I am better or that his gibs are better obviously.

[helene_t]

I think there are many possible interpretations of these stats.

 

As for the opening lead, it could be that GIB makes better opening leads than you do but it could also be that when you are on lead, the human will often be on lead also at other tables, and you may be relatively poor at opening leads compared to other humans. Or it could be that you are better at reading GIB's leads than other humans are.

 

Similar considerations for declarer/dummy. Maybe bidding is your strongest asset and you are close to average in terms of declarer play, and it is indeed quite plausible as you say that other humans hog the contract at the expense of sometimes playing inferior contracts.

 

Finally the standard error on the difference between your opening lead MPs and your non-opening-lead defense MPs is probably around 4% or maybe a little less, so it is borderline w.r.t. statistical significance.

 

Anyway, interesting stats. Would be nice if other people generated similar stats so that we can see if we can find differences in specific qualities of players, rather than just some players being better than others!

[hotShot]

Years ago I made some stats counting the number of cards played, that lost a trick compared to double dummy play.

The average GIB lost one trick in 4 boards, the leading GIB close to one trick in 3 boards.

So I'd say that the lead is a very difficult part of the game.

[hanp]

I think there are many possible interpretations of these stats.

Ditto.

[junyi_zhu]

Stats for my last 631 matchpoint boards on BBO. About 80% of these hands are in ACBL robot tourneys where I get the best hand, explaining the disproportionate number of hands I am declarer.

 

Position - Instances - Avg. MP - % Contracts Made

Declarer - 317 - 51.4% - 65.6%

Dummy - 139 - 55.1% - 75.5%

Lead - 66 - 46.0% - 75.8%

Defense (no lead) - 96 - 54.6% - 47.9%

Passout - 13 - 50.7% - N/A

Total - 631 - 52.1% - 66.9%

 

Questions:

1. What should I make of the higher matchpoint scores when I'm dummy? I suspect others often try to engineer the bidding in individual ACBL tourneys so GIB is not declarer, which I do NOT do...could it be that this leaves me in better contracts and produces better scores? Or is it just that I treat GIB better than others and he plays more solidly with partner's active support?

 

2. Am I really bad on opening leads? My stats knowledge is rusty, but I think the 95% confidence interval for my "on lead" MP% is about 40-52%, just outside the overall average of 52.1%. So I think I can conclude with 95% confidence that I'm a bad opening leader.

 

3. Should I be surprised that my GIB partner makes >75% of his contracts and I make only 66%? Since I have stronger hand all of the time (in the 80% of the hands that are ACBL individual), shouldn't I have a natural advantage over GIB in that my strength is concealed when I declare? Is this grounds to conclude I am a worse declarer than GIB?

 

4. Any other takeaways from these stats that I can use? Have others done similar analysis and found it useful?

Well, the leading thing doesn't mean much. Often you find a killing lead and gib switches to another suit to ruin the defense. Gib is just very very bad at leading and switching IMO. Against NT, gib very rarely lead long suits. Against suit contracts, gib would lead stiff or doubleton trump honor if possible. And very rarely, gib would lead a suit with honors. Against 6NT, gib very often lead a long suit with honors. All these point to a very weak opening leader by nature. When gib leads, human player tend to blow way less tricks than gib does.

Also it's not very difficult to reach the conclusion purely from the technique to know whether one is better in declaring than Gib. Gib's declaring skill is better than good club players, but nowhere close to expert level.

[ggwhiz]

The declaring thing vs dummy just means you bid better.

 

Years ago after a 7 day Canadian team trials, one guy in the bar pulled out the stats and told his partner "I'm plus 87 imps as declarer and you are minus 40".

 

His pard just said "Your bidding in a supporting role sucks".

[bd71]

Stats for my last 631 matchpoint boards on BBO.  About 80% of these hands are in ACBL robot tourneys where I get the best hand, explaining the disproportionate number of hands I am declarer.

 

Position - Instances - Avg. MP - % Contracts Made

Declarer - 317 - 51.4% - 65.6%

Dummy - 139 - 55.1% - 75.5%

Lead - 66 - 46.0% - 75.8%

Defense (no lead) - 96 - 54.6% - 47.9%

Passout - 13 - 50.7% - N/A

Total - 631 - 52.1% - 66.9%

For those who were interested before, here is an update of my latest 679 matchpoint hands on BBO...again, the vast majority are in ACBL robot tourneys so I'm declarer more often. This sample has no overlap with the first one.

 

Position - Instances - Avg. MP - % Contracts Made

Declarer - 318 - 51.5% - 68.0%

Dummy - 170 - 54.6% - 72.9%

Lead - 79 - 50.9% - 67.1%

Defense (no lead) - 95 - 51.2% - 53.7%

Passout - 17 - 50.0% - N/A

Total - 679 - 52.1% - 68.0%

 

 

The combined samples are:

 

Position - Instances - Avg. MP - % Contracts Made

Declarer - 635 - 52.1% - 67.5%

Dummy - 309 - 54.8% - 74.1%

Lead - 145 - 48.67% - 71.0%

Defense (no lead) - 191 - 52.9% - 50.8%

Passout - 30 - 50.3% - N/A

Total - 1310 - 52.1% - 67.5%

 

 

Tentative conclusions:

 

1. The declarer vs. dummy differential is very stable and similar. I am pretty much convinced this difference is driven by bidding (likely my tendency NOT to hog contracts from GIB, presumably unlike many of my competitors) and has nothing to do with my dummy play.

 

2. The lead vs. non-lead defense differential in matchpoints largely disappeared, but there's still a large difference in % of contracts made. Not sure what to conclude here. Were it not for the big gap in % of contracts made, I would conclude that the big difference from the first sample was a statistical fluke. But I have no idea what to make of it now.

 

3. Currently, I am a somewhat better player than the typical player in BBO's ACBL robot tournament fields. Relative to the competition, my strongest area is bidding, then declarer play, then defense (not trying to differentiate leading vs. other components of defense). I am better than the field in all 3 areas, although the lower volume/credibility of the defensive data may mean I am merely average here.

 

4. BBO's ACBL tournaments -- which emphasize bidding and declarer play -- are likely the worst possible forum for me to play in if I want to focus on my relative weakness (defense).

 

5. For someone who thought he was an "advancing" player, the lack of improvement in average MP score shouldn't make me very happy or confident of the presumed advancement.

[hanp]

What have you been doing to try to improve? Maybe people will have suggestions. I find your data very interesting but without knowing more it is hard to comment.

 

For becoming a better defender I would make these general recommendations (which may or may not apply to you):

 

1. Become a better declarer.

 

2. Read books on defending, especially books with puzzles.

 

3. Play with good players and discuss your defensive hands.

 

4. Go over the hands you defended very carefully.

 

I don't think the difference between the hands you play and the hands you defend is particularly large btw, and this may easily be attributed to bidding style instead of play.