All comments by Grigol Gogoberidze

.

OR

Benefits include:

- Create and share convention cards
- Create/answer bridge problems
- Connect with other bridge players
- Write and comment on articles

Plus... it's free!

Grigol Gogoberidze

It seems E intended ‘European cuebid’ trying to describe absence of ♦ control, not very practical vs 18-19pc. I think a pair playing ‘American cues’ can not go wrong here. Of cause this does not mean general ad of either method.

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

Suppose there is a 1NT-3NT board, which needs some advanced technique to make, say criss-cross squeeze. In the scoreboard you will see very few experts making (scoring, say, 95%) and a lot of average players going down (scoring, say, 45%).

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

In general I do not understand why BBO should forbid others to run rating systems like bboskill. It does not look fair to me.

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

Just some data analysis. As a pro in this field let me use your table of results in a a bit different way to arrive to some numbers. The idea is the following: the case when only 3 experts match hypothesis is definitely less statistically important then the case when all 5 experts match. In fact what you did in the table, you averaged opinion of 5 experts and compared that number with your hypothesis on binary bases (so there are only 2 results, match, not match). This way you derived numbers 56 out of 65. This averaging in the middle of the procedure does not add anything to analysis and makes it more difficult to interpret the results in numbers (what is the probability that in random experiment designed this way you can get 56 out of 65).

Alternative ways of analysis are as follows. You choose 5 experts to avoid problems related with personal style (say, Forrester is more active bidder then BZ), so you have 5X65=325 choises of ‘randomised experts’. Suggesting BZ are placing bids in random order, what is the probability that 264 (I counted all positives and negatives in your table, so your 4-positive in last column means 4 positive and 1 negative now ) out of all 325 cases will be positive? Now this is very easy. This is described by binomial distribution and basically is equivalent to the question - what is the probability that flipping coin 325 times you will have 264 heads. Here is online calculator http://stattrek.com/online-calculator/binomial.aspx we need probability 1/2, 325 and 264. The result is that the probability that this happened by chance is less than 0.000001.

Alternatively, you can do similar analysis for every expert separately.

And finally, if we use the same way your averaged numbers 56 out of 65 (although after averaging this is not strictly true) we will yield the same less then 0.000001 probability.

PS: I did this analysis in 30 minutes so some small errors are possible.

Grigol Gogoberidze

Let me remind that even Kasparov and Co failed to organize self-sustained separate (from FIDE) ‘world championship’. The fact that many people are dissapointed by WBF position with respect to recent scandals does not mean that parallel structures will be successful. May be it is better to pressure WBF by means of our national NBOs to help them become more adequate.

Grigol Gogoberidze

Grigol Gogoberidze

Grigol Gogoberidze

For people who are interested in details, this is quite popular and simple textbook on statistics for 1st year undergrads http://www.onlinestatbook.com/version_1.html

Chapter 1. Introduction. Section: levels of measurement. Subsection: ordinal scales.

Main (but not the only) problems with Dr Jassem's analysis is that when you adjust some numbers to different answers, this gives you still just ordering (like if you call Jassems categories not 2,1,0,-1,-2, but instead: ‘very positive’,'positive' ,'neutral','negative','very negative'), and not numbers which you can use for computations, say to compute correlation coefficient. The only thing you can do with this kind of variables (either number adjusted or not), to have ordering an state that ‘very positive’ is more positive then ‘neutral’. The key idea is that there is no way to show that distance between ‘neutral’ and ‘positive’ is the same as between ‘positive’ and ‘very positive’. For instance, I can claim that 1 very positive case is more important then 2 just ‘positive’, so we should use 3,1,0,-1,-3 scale. Now all numerical results will change.

In mentioned above textbook this is explained as follows: ‘What if the researcher had measured satisfaction by asking consumers to indicate their level of satisfaction by choosing a number from one to four? Would the difference between the responses of one and two necessarily reflect the same difference in satisfaction as the difference between the responses two and three? The answer is No. Changing the response format to numbers does not change the meaning of the scale. We still are in no position to assert that the mental step from 1 to 2 (for example) is the same as the mental step from 3 to 4 …’