In another thread, Frances has said that:
Quote
I can't give a logical justification for this, because I think it 'ought' to be correct to pass in 4th seat on this pile of rubbish. But I'll open it anyway.
In fact, I have looked at Pearson points for some time using BridgeBrowser and have not found them to be relevant (supporting Frances’ observation above). But to answer the question do they matter, (or at least is 15 the magic number to open on) I have performed the following experiment. I used the largest online database for bridgebrowser (with more than 23 million hand plays). This is from a non-BBO site, but I choose it for two reasons. 1) Typically they have many comparisons on the same hand (often more than 50 tableS), and 2) they have as many matchpoint hands as imp hands.
I set the criteria for the dealer and next three hands as having 4 to 12 hcp, and that they must all pass (thus no forcing pass systems, and no flake accidently passing a 15 hcp hand). The fourth hand, I varied the hcp and number of spades. HCP went from 11 to 14, spades went from 0 to 3. The fourth seat hand (for these studies) never had a six card suit, and was five-five only the times I tested with a spade void (so either 0445 or 0355). Then I determined the result the player in the fourth seat would get for passing.
Let’s start with 14 hcp (and thus spade void).
Passing out the hand with these 14 Pearson points occurred to rarely to be of significance (only 4 times out of 23 million hands and out of 751 that matched the search criteria – but passing there was still a huge loser), So I looked a passing out with spade void with 13, 12, and 11 hcp (and pearson points), In each case, Passing out was a loser despite having as few as 11 pearson points. With 13 hcp and spade void, passing earned -5.09 +/- 0.72, and 34.87% (+/1 8.3%). But again, so few people passed with 13 hcp and void to be not significant. So with 12 hcp, passing was worth -1.47 (+/- 0.81) and 40.31% (+/- 3.97), only with 11 hcp and a void did we see some trend where passing was better, and even then only at matchpoints. With 11 hcp and spade void, pass earned -0.46 (+/- 0.35) at imps (did better bidding), but 50.21% (+/- 3.89%) at matchpoints. Both these numbers are not significantly different from “average”. So with 5530 and 5440 and 11 hcp, other factors beside hcp should be used in determining you opening bid (rebids? Quality of suits?).
Now lets turn our attention to hands with 13 hcp and a singleton ♠ (14 pearson points). I divided this search up into 1444 and 1(543) hand patterns. Passing with both of these 14 pearson point hands were statistically worse than taking action. With 1444 pass was worth on average -1.46 (+/- 0.38) and 28.19 (+/- 2.85). With 5431 pattern, the results were even more in the favor of bidding. The passers earned (on average) -1.94 (+/- 0.19) and 29.28% (+/- 1.14).
So with 13 points or more, ignoring pearson points seems to be clearly the winning strategy. What happens as we weaken the hand further with regards to hcp? How about 12 hcp and 2 spades (along with no six card suit)? Passing the hand out earned -1.05 imps (+/- 0.03) and 42.95 (+/- 0.32). This 12 hcp and 2 spades totaled 14 Pearson, so how about 11 hcp and 3 spades? Here we divided the hands into unbalanced (5431) and balanced (4333 and 4432) with 3 spades in each case. With unbalanced, passing again was bad, earning -0.98 imps (+/- 0.08) and 44.53% (+/- 0.68). So far the data has been consistently that opening with 14 Pearson points is better than passing. But with 4333/4432 we got a split decision. Passing at imps continued to give (on average) a poor result of -1.09 imps (+/- 0.01), but gave a better matchpoint result 52.74 (+/- 0.04).
So with the possible exception of 11 hcp and 3 spades, bidding in fourth seat with “only” 14 pearson points and “typical” distributions (0 to 3 spades, no six card suit) appears to be statistically the best strategy.
What about with 13 pearson points? I showed above that with a void, bidding with 13 and even 12 hcp was a winner. What about with other hand patterns and fewer hcp than studied above?
With a singleton spade and 12 hcp (13 pearson points), passing with 5431 patterns earned -0.94 imps ((+/- 0.13) and 38.88% (+/- 0.33); while it was a little safer to pass with 4441 pattern, getting only -0.18 (+/- 0.19) at imps, but 43.62 (+/- 1.92). Note, this matches Frances feeling that at matchpoints you want to bid with these non-descript hands, but here at imps, the number is not statistically significant from average. Since the 12 hcp hand with 5431 (12 Pearson points) was so overwhelmingly in favor of bidding over passing, I looked at 11 hcp hands with same pattern. Here again, the passers did worse than “average”, earning -0.55 imps (+/- 0.07) and 46.9% (+/- 0.82).
So there is nothing magical about 15 Pearson points, and in fact, with 14 and normal hands, it was better to bid than to pass, and indeed, opening with even 13 and 12 pearson points worked out (generally) for the best. Thus, perhaps the role for Pearson points is if you should open light with 10 or 11 hcp in 4th seat. We have the data above for opening hands with 5440 short in spades and 11 hcp was essential average. What if the short suit was not spades? Would that change the results? While passing with 5440 or 5530 short in spades and 11 hcp earned essentially “average” (not significantly different from average), passing with same hcp and distribution but holding five spades was horrible .. -1.57 imps (+/- 0.35) and 38.93 (+/- 2.11).
So if holding five spades and 11 hcp it is better to reopen while holding a void in spades and same distribution in other suits it is roughly a toss up, what does that say about the decision to reopen when holding four spades and some average distribution. This would lend itself to an extensive study (the value of a third, fourth, fifth spade) in pass out position but keeping the hand “pattern” the same. If such a study found that extra spades increases the likelihood they opponents have to outbid you by going to the next level, such a finding might support a real significant meaning to pearson points, After all, reopening with 11 hcp and 5 spades is better than reopening with 11 hcp and a spade void (as shown above), so there is a grain of truth to pearson points somewhere.
This data probably may reveal the “truth” behind pearson points. If you have an opening hand, even a marginal opening hand, the number of spades is just one indicator on whether you should open it in fourth seat. Good distribution counts more, and the data suggest that all 12 hcp, regardless of spades, should be opened. Distributional hands with 11 hcp should be opened. Perhaps if you are trying to decide to open with 10 or 11 hcp (and with 11 hcp, with balanced distribution), looking towards spades might should factor into the final decision, but clearly it should not be the over-arching factor.