[jogs]

Short Suit Totals(SST)

There seems to be a misconception of what SST means. There is no definition
provided by Lawrence on his website about fighting the LAW.
Here's my definition:
SST is the sum of the shorter holding of the two short suits within the partnership.
SST is the number of possible quick losers in those two short suits.

Examples
4=5=3=1 /////////// 5=4=3=1

The shorter diamond holding is 3. The shorter club holding is 1.
SST = 3 + 1 = 4
--
4=5=3=1 /////////// 5=4=1=3

The shorter diamond holding is 1. The shorter club holding is also 1.
SST = 1 + 1= 2
--
4=5=3=1 ////////// 5=4=2=2

The shorter diamond holding is 2. The shorter club holding is 1.
SST = 2 + 1 = 3
--
4=3=3=3 ////////// 5=4=2=2

The shorter diamond holding is 2. The shorter club holding is 2.
SST = 2 + 2 = 4
-------------------------
Lawrence claims tricks have nothing to do with trumps. That statement is
false. Studies using least square estimates confirm that tricks are correlated
to trumps. SST is also a valid parameter for estimating tricks.
The truth is the true random variable for estimating tricks is the joint 2X4
suit pattern of the partnership hands. All 26(2X13) cards of our partnership
affects the number of tricks we can make. SST does not replace trumps.
Trumps is one end of this variable. SST is the other end. Together trumps
and SST estimate tricks better than either on a stand alone basis.
---------------------------
Analogy: Using American football. What's the most important attribute for
a wide receiver? Speed, size, or good hands. Wrong question. They are all
important.
Bridge: HCP, trumps, and SST. The partnerships who know all three
parameters are better positioned for making correct bidding decisions than
partnerships utilizing only two of the parameters.
---------------------------
Lawrence's SST may be used to gain insight for uncovering value due to suit
pattern(or shape if you prefer). Hand evaluation should evolve from counting
and adjusting one's points to estimating tricks for the partnership. The more
parameters partnerships use to evaluate their hands the better their
estimates of their tricks.

jogs

[navahak]

jogs, on 2014-December-27, 08:52, said:

Lawrence claims tricks have nothing to do with trumps. That statement is
false. Studies using least square estimates confirm that tricks are correlated
to trumps.

This is a bit more complex interaction with two parameters that you claim here are not related to each other. Our total trump length has direct influence to the probability distribution of opponents SST. That makes trump total correlate with number of tricks available in board. That is reasoning behind Lawrence's claim that trumps has nothing to do with tricks. It is a bit extreme claim when clearly thinking process from trump length to SST makes them clearly related and it can be also tough to give equation why LoTT works.

SST+WP is way to estimate losers in your contract. But losers aren't only thing that mandates how many tricks can be taken in a given contract. You need also enough winners to score the tricks that you think you will make. That causes variance both ways from simple SST+WP loser evaluation.
One might quickly think that is proof trumps influence the equation but that isn't complete truth. There is also possibility that tricks are provided by a long side suit which makes trump length only one of possible source of tricks to archive enough to match loser based hand evaluation.

[mike777]

In his book starting on page 127 short suit total is taking your sides shortest suit and add it to your sides second shortest suit.

You must have a fit to use SST.

If both sides have shortness in the same suit count it only once.

If you have 3 short suits, stiffs, voids and doubletons, is a special case.
1) add the two shortest suits
2) now check the third suit
if doubleton subtract one
if stiff subtract two
if void subtract three

Please note I may have skipped over a few other pointers spread out in his book

[rhm]

jogs, on 2014-December-27, 08:52, said:

Short Suit Totals(SST)

---------------------------
Lawrence's SST may be used to gain insight for uncovering value due to suit
pattern(or shape if you prefer). Hand evaluation should evolve from counting
and adjusting one's points to estimating tricks for the partnership. The more
parameters partnerships use to evaluate their hands the better their
estimates of their tricks.

jogs

Your insights are not wrong, but the question is whether they are practical.
There is little point in finding parameters of two hands, which will predict tricks more precisely, if these parameters depend on the combined holding of two hands but are frequently not revealed in the bidding.
During the bidding I may not locate partner's shortages nor whether there is duplication opposite my shortages or not.
Exchanging this information may make your trick estimation better but it would also help opponents to contest the bidding and how to defend eventually.
Remember also that your objective in the bidding may not be to contract for the number of tricks you can actually take.
For example if a major suit opening or preempt is raised to game I do not know until dummy comes down whether this raise was made with the intention of making or as an advanced sacrifice.
Often dummy does not know or care either.
Since one objective in the bidding is to select a "suitable" trump suit, information about total trumps tends to be exchanged during the bidding.
That is not true about other parameters like side suit shortages.

Rainer Herrmann

[SteveMoe]

I've found Lawrence and Wirgren's arguments thought provoking but with a poor foundation as to why SST translates to tricks. They were irritated by the correlational nature and poor accuracy of the LoTT. Their solution is a bit too complex for execution at the table, and lack a compelling causal argument. Working Points can be hard to estimate.

Steve Bloom has presented a thoughtful analysis and models tricks based on two suit fit at http://bridgewinners...-application-2/
This is the first of seven linked articles.

Steve's approach is compelling because it's simple and usable at the table, with few inputs that are accurately known. He also offers a plausible cause-effect framework.

Food for thought.

[mike777]

The argument is it is your short suits that will determine what your trumps are worth. He argues that without adding working points you wont get far with sst.

However it seems after ten plus years the FTL approach has not caught on, perhaps as posters point out it is too complex to estimate?

[MrAce]

Rainer nailed it imo.

[navahak]

I agree that SST+WP information is rarely available at the table. When the information is available visualizing partner's hand is more accurate method to estimate tricks. But SST+WP is still a tool for post-mortems when trying to figure why we should have bid more or less and what suggested the better biding choice.

[jogs]

Short Suit Totals(SST)

Since To Bid or Not to Bid Cohen has convinced the bridge masses to think
in terms of total tricks. We often know our combined trumps. We can only
guess at their combined trumps. Also today we know that tricks=trumps
only about 40% of the time.
These examples will be estimating our tricks.
----
We have two 9 card fits. 5=4 and 4=5.
To simplify the examples all hands are normalized to we have all the honors
in the majors and they have all the honors in the minors.

west		east	tricks
5=4=4=0 	4=5=0=4	 13	
5=4=3=1		4=5=1=3	 11
5=4=2=2		4=5=2=2	 _9	
5=4=1=3		4=5=3=1	 11
5=4=0=4		4=5=4=0	 13

We probably can't make all 13 tricks in lines 1 & 5. The potential is there.
Lawrence would argue it is only the SST that matters. Our combined length in
the two long suits is 18. We only have 18+ cards in our two long suits 3 1/2%
of the time. In this special case SST gives us our immediate losers. It is
relatively easy to exchange info on our trump length during the auction. It is
extremely difficult to decipher our SST during the auction.

west		east	tricks
5=4=4=0 	5=4=0=4	 13	
5=4=3=1		5=4=1=3	 11
5=4=2=2		5=4=2=2	 _9	
5=4=1=3		5=4=3=1	 11
5=4=0=4		5=4=4=0	 13

Trumps do matter. In this special case, an additional trump is not worth a
full additional trick. Still the extra trump is worth fractionally more tricks.
If the QJT of trumps is exchanged from the EW hands for small trumps in
the NS hands in each of the two sets of examples, the extra trump is worth
even more fractional tricks.
Trumps and SST are subvariables. The partnership 2X4 suit pattern is
the independent random variable for estimating tricks.

jogs

[benlessard]

For me there was a point where my hand evaluations went up a level or two without using any tools, as I look back reaching that point was way more important than anything else.

I look at LTC,KR,Zar and SST and find that at one point these would have been worthwhile for me, but one or 2 years later they would be pointless.

I know without any doubt that I bid better without them than with them.


1- Visualize partner hands, start with 3 sample hand and look how well do you fare. If the 3 hands fall on the same side and you tought it was a close decision you are unlikely to go wrong. If its 2/1 pick 3 new hands if you have time.

2- If your not sure post them or ask other players afterward.

3- Remember the hand where your bidding judgement was off and try to find out why you misevaluated.

4- build a collection of hand where you judgement is different than the majority. Revisit them once in a while.

5- If you have offbeat views about some situations keep a steady record of them.

I think overcalling 1NT should be 16-19. I also think that its often better to raise with 2 cards than to pass.

1S--(2H)--??

Qx
Axxx
xx
Axxxx

I know very well that the majority doesnt see it that way. So I keep tracks of similar hand to make sure my judgement isnt off.

6- play it safe; if your not sure aim for the middle. You are used to need Y to overcall in a particular spot but see good agressive players do it with Y-2, try Y-1. You might be wrong but you will be off by just a little while staying at Y or trying Y-2 is risking to leak imps. I see this as a form of trust of the law of large numbers by using other players empirical results.

[jogs]

It requires a long bidding sequence to exchange info on WP.
WP is a side distraction. With a long bidding sequence why not
just attempt to count our tricks?
This thread is devoted to SST. Future post will suggest ways
to use SST to gain insight on tricks due to suit patterns.

[jogs]

benlessard, on 2014-December-28, 17:15, said:

I also think that its often better to raise with 2 cards than to pass.

1S--(2H)--??

Qx
Axxx
xx
Axxxx

My thinking is way out of the box. I think negative doubles only
apply to 1m - (1♠) - X

With 1♠ - 2♥ - X

No one cares about the minors. This double shows two spades.
The 2♠ shows 3 spades. This is sort of like support
doubles. If a raise shows x cards, the double shows (x-1)
cards.

[jogs]

SST

In the general case for every one unit lower in SST one can expect about
0.25 to 0.4 extra tricks. For any specific board one can only gain or lose
tricks in integer increment amounts . On a specific board one can not gain
a fractional trick. It is either a full trick or no tricks.

The 8 card trump fit

The 8 card trump fit is studied. 3 trump fits are in this study. The 4-4 fit,
5-3 fit, and 6-2 fit are parts of the study. High card points is normalized,
meaning each side has half the deck. In theory we expect to make as many
tricks as we have trumps.
It is not easy to design an unbiased study comparing the expected tricks
of the three 8-card trump fits. It IS easy the calculate the expected short suit
totals. Ran one million iterations of each of the three 8-card trump fits. Threw
out any cases that contained a longer fit.
This is not a proof. It is a naive approach to present evidence of the relative
values of various 8 card fits.

These were the results.

4-4 fit: 3.664 SST
5-3 fit: 3.564
6-2 fit: 3.255

These results suggest that the 6-2 > 5-3 > 4-4. The 6-2 fit plays better than
either the 5-3 fit or 4-4 fit. By plays better, the expected tricks are greater
with the same HCP strength. The 5-3 fit plays marginally better than the
4-4 fit. Smaller SST suggests more expected tricks.
Assume no ruffing for extra tricks and solid trumps. The 4-4 fit produces
only 4 trump tricks. The 6-2 fit produces 6 trump tricks. This would also
suggests that 6-2 fit produces more tricks than the 4-4 fit.

4=3=3=3 //// 4=3=3=3
Mirror patterns produce fewer tricks. SST=6 is only possible when the
trump fit is 4-4. This is why in general the 5-3 fit is better than the 4-4 fit.

jogs

[mike777]

jogs, on 2014-December-28, 18:37, said:

It requires a long bidding sequence to exchange info on WP.
WP is a side distraction. With a long bidding sequence why not
just attempt to count our tricks?
This thread is devoted to SST. Future post will suggest ways
to use SST to gain insight on tricks due to suit patterns.

the authors argue that sst alone is not a sound foundation but adding WP and a fit creates a sound foundation.
the authors do give many examples of a short sequence where WP info is exchanged.
However the authors do state in their opinion the biggest negative to FTL is when sst is low and wp are few, in this case you need extra trumps often to set up the side suit.

with all of the above it should be noted that "FTL" is rarely quoted in the literature or here in the forums over the past ten years

[helene_t]

Can someone generate 10000 dd analyzed hands and calculate the variance in tt explained by total trumps and sst? Or I can do it myself if someone can recommend a dd solver for linux.

[jogs]

Power and Pattern(P&P)

There are two independent random variables to measure tricks. They are power and pattern.
Power is the set of all parameters measuring tricks by using honors and high spot cards.
Pattern is the set of all parameters measuring tricks by using the length and shortness of suits.
It is the 2X4 suit pattern of the partnership. All 2X13 cards of the partnership is part of
pattern.

L/W calls power, Winning Points. HCP is a component of power which is proportional
to tricks. Trumps is the component of pattern which is proportional to tricks. Both
trumps and SST are subgroups of pattern. Trumps is one end and SST is the another
end

Our tricks
E(tricks) = trumps + (HCP-20)/3

By the third or fourth bid of a bidding sequence, we should know our combined trumps.
We should also know our HCP within one point. We don't know their trumps.

Our tricks
E(tricks) = trumps + (HCP-20)/3 + SSTadj

With luck we can sometimes deduce our SST. Use the SST adjustment to improve
our estimates.