BBO Discussion Forums: What's the rule of 17 for supporting partner's preempts? - BBO Discussion Forums

pilowsky, on 2020-January-23, 06:19, said:

The main reasons that TT raised are included in Mike's post directly below yours so perhaps you can find something there. I think it is fairly clear that debating the merits of this rule is fairly pointless though, so instead I will provide you with an opportunity to prove your case to us. Let us designate the number of trumps in the hand as T and the hcp as H. Then your rule can be stated as:-

A. If T+H>=18, bid 4M.

So now I would like you to compare this with some alternative simple rules:-

B. If T>=4, bid 4M
C. If T>=4, bid 4M unless specifically 4333
D. If 2T+H>=22, bid 4M
E. If 3T+H>=26, bid 4M
F. If 4T+H>=30, bid 4M

I would like you to analyse these, ideally for both majors but let's start with spades as that is simplest. Assume Dealer holds 5-9hcp and a good 6 card spade suit and that second seat holds either 0-14hcp and 3+ spades or 0-10hcp and 0-2 spades (we can refine this if the results seem close). Then compare results for each of the 6 rules above against the par score for the deal. I am confident that even within this limited analysis it will be clear that Ro17 is sub-optimal. But perhaps you will find otherwise. So please go away and run some analysis and then come back with your results. If you do it properly, maybe you will even surprise yourself...

Dear Mark, Thank you for your ideas, but that is not how research works in real life. I knew a theoretical physicist who once approached a senior Professor of Biology and told him that he had a Theory about how the brain worked and asked him to conduct a specific experiment to test the idea: that did not end well. I suggest that you do your work, and I'll do mine. Yes, I am still working on this problem. It is a large and interesting one. If it piques your interest, then it is you that should also do some research. This habit that some people have of providing instruction to others along the lines of you must do this, or that is not conducive to progress. Quite the opposite; I am sure that you, as a mathematician, are more qualified than me to test this hypothesis for the benefit of all. The answers may be of interest beyond Bridge: who knows, that's why it's called research.

pilowsky, on 2020-January-23, 14:02, said:

To be honest I was working under the assumption that your previous post was accurate.

pilowsky, on 2020-January-17, 01:25, said:

If you are serious about promoting this rule and its "obvious advantages" then you need to be able to present your real research together with your understanding of maths and probability. My suggestion is a way that you might achieve that end comparing against some alternative simple rules for which the advantages are not clear. My assumption was that you will already have a substantial amount of research behind you to make the claim but that it is perhaps not in a form that lends itself to such mathematical presentation. The above test would provide you with the statistics to back up your argument or, alternatively, to improve your rule.

The problem that you have here is that one of those simple alternatives, the LoTT, already has a body of statistics broadly supporting it. At the moment there is no such similar body of evidence for the Rule of 17, So from a personal point of view I will for the time being stick with the formula that the maths backs up as a broad basis for decisions and adjust according to experience and judgement. Anyone that can gather evidence for an improved baseline rule would be more than welcome. This has already happened to me a couple of times, not only the LoTT but also in terms of adjusting the values of aces and queens in the Milton Work count when strong mathematical evidence came to light that earlier analyses were not completely accurate.

It might also interest you to understand why I chose the specific formulae. I am giving you the benefit of the doubt here that your existing research and analysis has found an optimal solution (14hcp) for 4 card trump support, which I noted was the case for all of your examples. Then I constructed formulae that use the identical values for 4 card support but offer a gradually increasing weight to distribution over hcp. It seems to me completely obvious that this process is essential for any construction of an optimal formula balancing trumps and hcp, so if you have not already done work in this area then it seriously calls into question any assertion that the rule has been properly formulated.


In any case, you asked at the start of this thread for the input of a mathematician. You have my input now; it is up to you what you choose do with it.

pilowsky, on 2020-January-16, 18:32, said:

I am not a mathematician. However, I can read. My reading of "the rule" differs.

The above rule is not the "Rule of 17." The error you've made is that the formula should read: ∑HCP+TRUMPS and the result is >= 17 then bid 4 of that suit.

I hesitate to cite my sources, since that might be construed as an Appeal to Authority. You are free to do your own research.

Yes, the correction is right, it should be ≧.

pilowsky, on 2020-January-24, 02:35, said:

FYP

pilowsky, on 2020-January-24, 02:35, said:

I am at a loss to see the point of posting this hand unless the purpose was to highlight bad bidding.
responds 2♠ to a takeout double with an 11 count (♥Q may be wasted, but heart singleton should be very useful) and a decent 5 card spade suit. 2♠ could have been a 3/4 card suit and 0 count.

East make a "penalty double"??? of 2♥ and then raises 2♠ to game with only 3 card trump support and nothing close to having a 4♠ bid.

North opens a weak 2 with a side 4 card major which some players refuse to do, and makes a very speculative double that gets a little lucky to find partner with 2 tricks and East having very poor trumps for the bidding.. Of course, double dummy, 3NT is unbeatable.