[Antrax]

I've been reading Klinger's essays on losing trick count. As is common in such essays, the method works perfectly for all hands there. I kibitzed some vugraph and tried applying LTC to the hands there and correlate them to the player's decisions, and it seemed somewhat hit and miss. I was wondering if it's possible to characterize the misses - what's inherently flawed about the LTC? Does NLTC fix it?
Also, are bidding conventions expected to change to accommodate incorporation of LTC, or is it strictly another hand evaluation tool?

[helene_t]

Basic LTC means:
- one trick for a working top honour (AKQ)
- three tricks for a side void
- two tricks for a side singleton
- one trick for a side doubleton

A worthless hand is 12 losers so you subtract the tricks from 12 to get the LTC. It is easier, though, to count tricks rather than losers. Say each player has 5 tricks (or 7 LTC if you like). This is 14 losers so you make 24-14=10 tricks. But 5+5=10 is simpler. For some reason, all textbooks go the de-tour via losers instead.

The most obvious flaw with this is that a queen carries the same weight as an ace.

Modified LTC is the same except that:
- a working queen is half a trick
- an ace is one and a half trick
- a sec ace would be three and a half trick (two tricks for the singleton, one and a half for the ace. This is too much. Count only three tricks.
- Some players count a quarter of a trick for working jacks and for queens in doubletons.

Even after adjustments it is still flawed, in that:
- the number of trumps are not included.
- shortness is given the same weight regardless of whether it comes with the long or the short trumps.
- shortness is overvalued. 4-2 in two side suits is of course better than 3-3 but it is not a king better, especially if you have only 8 trumps together.

I think it is useful when deciding whether to bid game over partner's preempt. Agree that a 3M opening shows give or take 6 1/2 modified LTC (or some such). As responder you try to guess how many of those losers you can cover.

It can also be useful when making general game tries after 1M-2M. Agree that
1M-2M
3M*
shows give or take 5 modified LTC.

If you use unmodified LTC it works reasonably in auctions where both partners use them since overvalued queens in one hand are often balanced by an undervalued queenless hand opposite. However, in auctions like
1NT-2cl
2M-4M?
you obviously can't use unmodified LTC since opener has promised a certain number of HCPs, not unmodifed LTCs. Unmodified LTCs don't correlated strongly enough with HCPs to be useful in such auctions.

Lawrence/Wirgren proposed short suit totals which is a similar philosofy excpet that they devalue duplicated shortness. A doubleton is not an asset if partner has shortness in the same suit.

In contested auctions I don't use mLTC. Two reasons:
- Often it is a LOTT decision so mLTC are irrelevant
- You will have additional information from the auction such as risks of enemy ruffs, likely bad trump split, likely winning fineses, likely fitting honours. So it all gets too complicated for formalizing. I just try to figure out how many tricks we will lose. If I can;'t figure it out myself I try to show a feature so maybe partner will be able to figure it out.

[Zelandakh]

Modern LTC is functionally the same as A = 4.5, K = 3, Q = 1.5 together with 3/6/9 for shortages. Compared with something like 4.5/3/1.5/0.75 and 5/3/1 it is clear that LTC values shortages more and does not count jacks, even if supported by a higher honour. If you use 3/2/1 for shortages in the hand with long trumps then the difference is even greater. All I am really doing here is agreeing with helene's post but trying to provide a mathematical reasoning which makes the reasons she gave more obvious.

[Siegmund]

LTC is not perfect, and you wouldn't expect such a simple method to be perfect. Its biggest value IMO is that it turns your attention to counting tricks rather than counting points. When a point-count method predicts the wrong number of tricks, all you can do is either shrug your shoulders, or turn to ever-more-complicated refinements that still don't work. When a trick-counting method predicts the wrong number of tricks, you can look at the hand afterward, ask yourself which card did I count as a winner, but fail to win a trick with? or which card did I count as a loser, but won a trick with anyway? and see why.
This leads to common-sense adjustments like counting Kxx and Qxx as more than two losers when the person on your left has opened the bidding.

I will also suggest you concentrate more on Klinger's "losers minus cover cards" formulation, where one partner counts losers and the other partner counts winners, than on the 24-my losers-your losers formulation.

I am in a minority on this forum, for liking LTC at all in the first place, and for disliking the formulaic adjusted-LTC approach. As previously noted, "adjusted LTC" is essentially a 3-2-1 point count method, and ceases to be about counting actual tricks at all.

Re the 2nd question - LTC is, on the surface, just a hand evaluation tool, that sharpens up your judgment about e.g. limit vs game-going raises, rather than something that automatically changes your system. But you certainly can -- and, if you like LTC, should -- re-orient parts of your system around LTC if you and your partner both are using it. An obvious example is your choice of game try after 1M-2M: if it goes 1H-2H-3C, partner has sent you a clear message: "count your king and queen of hearts and clubs as full cover cards. Do NOT count your queen of spades as a cover card. Decide between 3H and 4H accordingly. If you counter-try with 3D, that must mean 'I have something in diamonds that I'm not sure whether to count as a cover card'."

[nigel_k]

If your goal is to be really good at this game, any formulaic approach to deciding what to bid will not work.

Reading books by Klinger may not work too well anyway, but that is a separate issue.

[Antrax]

My goal is to have fun. Getting better is means to an end. I expect the intuition part of bidding will come after a lot of practice based on rules.

Thanks everyone.

[P_Marlowe]

Hi,

The LTC works reasonably well, but you need judgement to decide in which case you rely
more on the LTC and in which cases not, and in which cases to disregard him completely.

The LTC is a statistical method, it assumes, that you have at most 3 loosers in a suit.
If you have a 4-4 fit, this basically implies for the trump suit, that the remaining
5 cards split 3-2, which happens roughly 66% of the time, i.e. the prediction is true 2
out of 3, i.e. you will hit the target twice and miss it once.
As far as I know 68% is the real number.
Applying the LTC in the presence of a 44 is still considered reasonable, but it becomes
borderline.

Assume you have a 7411 shape, the 4 card suit being 5432, the LTC will say, that the 2 is
not to be counted as a looser, since you can ruff it.
Now if we assume your 7 card suit will become trump, how likely is it, that you will be
able to ruff the 2? Not likely, if your p has 1 or 2 cards in your long suit, they will
prevent you from ruffing the 2 with one of those cards.
Coming up with a hit and miss statistics is a bit more complicate, but you may get it anyway.

Those are just easy example, that showes, that you need to keep the limitations of the LTC
in mind, when using the LTC, I like the LTC very much, but the LTC serves only as a tool that
helps to decide borderline cases.
But if you do this, than the LTC wont be a hit and miss thing.

With kind regards
Marlowe

PS: You find the above mentioned examples in Klingers book, but in the appendix, where he is talking
about adjustments, but those are just examples, in the end think about the preconditions the LTC is
assuming to be true and you will find most of the examples to be obvious.

[blackshoe]

LTC works best when you have a nine card fit. It works less often when you have an eight card fit. When you have no fit, it sucks.

LTC plus cover cards is generally more accurate (given the required fit) than raw LTC. I should point out, though, that the concept was invented by George Rosenkranz, not by Klinger.

[Antrax]

I didn't mean to attribute it to Klinger; I merely said that's whose essays I've read.

[ArtK78]

If you are interested in a system that incorporates MLTC into hand evaluation, check out Romex. I have not played Romex in many years, and I learned it from the original text (back in the mid 1970s). Romex has been updated a number of times since then, but I understand that its major suit raise structure and other sequences is still based on MLTC.

The presentation of MLTC by George Rosenkranz is different than the way it is often presented (particularly the way it is often presented in these fora). For example, opener's bids are based on the number of losers in his hand (based on MLTC). Responder, however, considers potential and actual cover cards (cards that cover losers in opener's hand), not the number of losers in responder's hand. So, for example, opener opens with one of a major. Typically, his loser count ranges from 8 (subminimum) to 5 (maximum) [with less than 5 losers, opener's hand may qualify for a Dynamic 1NT opening or a Strong 2♣ opening]. Responder's bids are based on the number of cover cards and potential cover cards he holds. For example, responder knows that the A, K and Q of opener's major are absolute cover cards, since the absence of those cards from opener's hand means opener is counting a loser for each of them. Side suit Aces, Kings and Queens are potential cover cards (Aces more so than Kings, and Kings more so than Queens) because they are only covering a loser in opener's hand if opener has length in that side suit. Similarly, shortness with adequate trump support can be one or more potential cover cards. The object of the bidding is to determine how well the cover cards in responder's hand cover the losers in opener's hand.

Various game tries are used to show length and shortness so that responder can guage how well his potential cover cards are working.

Assuming an adequate trump fit, if all but 3 of the losers shown by opener's bidding are covered by responder, a major suit game should be reached. And if responder can cover all but one of opener's losers, a slam can be reached.

This is not nearly as mystefying in real life as it may appear from the explanation above. By the way, point count is included as a guide to all of the bidding as well.

[Apologies to Blackshoe, who mentioned some of these points in his post above and also attributed them to George Rosenkranz]

[blackshoe]

The latest books on Romex have a lot of good stuff on hand evaluation. If you're interested, check out the first chapters of Godfrey's Bridge Challenge and Bid to Win, Play for Pleasure. The latter book is the latest full description of Romex, with all the bells and whistles. The former is a simpler introduction to the system, which should be usable by anyone familiar with 2/1.

Actually, the latest wrinkle is a "two card system" - Romex Forcing Club (a Precision variant, basically) when not vulnerable at MPs or when favorable at IMPs, Romex otherwise. The difference, basically, is the use of 1♣ in the Forcing Club variant to replace both 1NT and 2♣, which are two of the forcing openings in Romex. The third forcing opening is 2♦. Prior to the "two card" system, Romex had a fourth forcing opening: 2NT (natural, 25-26 HCP, 9 controls, balanced, forcing to at least 3NT).