ACBL strats

[Elianna]

I've noticed with the recent change in score printouts, the ACBL games are now stratified, but I was wondering how the strats were decided. For example, I find it very puzzling that Justin's (Jlall) pair was in Flight C (in the 8pm pairs game today), considering that he has at least 1,000 points (probably more like 2500).

[bendare]

Good question. It is quite amazing that players who are in the 95 PLUS % ranking by ACBL masterpoint holdings (percentages can be found at ACBL's home page under "membership by master point holdings") are regularly showing up under B and C "flights" (???) in the BBO ACBL games.

 

I have heard a dozen speculations on what the A, B and C are about, none of which makes any sense. Please enlighten us, Fred.

[keylime]

Justin's I think GLM -

 

I think it's probably a glitch and it'll get fixed.

[hrothgar]

It might be useful to review the difference between Stratified and Flighted events. Both Stratified and Flighted events divide the pools of players into subsets, however, the motivation is very different.

 

"Flighted" events are segregated based on ability. This ensure that pairs in the lower flights are not competing against more silled players. It can also be argued that the players in the upper flights benefit by reducing the variance.

 

"Stratified" events used some other criteria to segement the player pool. There are a number of different reasonas to stratify an event. Events are typically stratified to increase the statistical accuracy of the contest by creating "fully-meshed" movements.

 

In short, there is a difference between "Strat C" and "Flight C"

[Elianna]

I know what flights and strats are, thank you. I was in a discussion about NAPs and flights were on my mind.

 

Also, Justin may be a GLM now, but at the beginning of the year, he was an SLM, and I didn't feel like claiming honors for him that he may or may not have.

 

But this still doesn't answer the question of what he was doing in the C strat.

[hrothgar]

I know what flights and strats are, thank you.  I was in a discussion about NAPs and flights were on my mind.

 

But this still doesn't answer the question of what he was doing in the C strat.

Potentially my posting was less clear than I thought...

 

If we are actually talking about a stratified event, than there is no deterministic relationship between a players level and the section that they play in. Most Stratfieid events seed players randomly. On occasion, there are attempts to deliberately seed players to achieve an un-biased distribution.

[Elianna]

Look, in the new way of listing scores by section, the ACBL games online list of winners by section, they now list people's stratification IN ADDITION to sections. Please take a look at this before you reply again.

[hrothgar]

Look, in the new way of listing scores by section, the ACBL games online list of winners by section, they now list people's stratification IN ADDITION to sections.  Please take a look at this before you reply again.

 

can you provide a URL that I can't use to look at this...

 

I normally don't pay much attention to the ACBL events and there aren't any events with multiple sections running at the moment.

[Elianna]

I don't think that it NEEDS multiple sections to display strats, it's just when you go past the first page. I don't know where a url is that displays this, just wait for the current pairs to finish (prolly about five-ten minutes from now, and take a look at more than the first few pages of the score listing).

[awm]

There seems to be some confusion here between section and stratification/flight.

 

Players are divided into stratifications/flights based on their skill level. In acbl-land this is usually done by masterpoint holding, which is a poor substitute for skill level, but that's a different argument.

 

In a flighted event, players compete only against those in their same "skill level."

 

In a stratified event, players compete against everyone, but their scores are separately compared against those in their skill level. This way, even though you may have experts and beginners in the same field, the beginners can still obtain recognition (masterpoint awards) by doing better than the others at their level. A game score of 50% (or even lower!) can be quite an accomplishment for a beginner playing in a field filled with experts.

 

The division into sections exists because there are often too many players to sensibly run a game (you won't get to play against a substantial proportion of the pairs sitting the other direction). Essentially, the division into sections splits a single game into many separate games, although additional masterpoints are sometimes awarded for the best results overall (between sections). The idea here is, it's not as fair to compare my score against the score of people who played my cards against totally different opponents. Section divisions are (usually) done at random in a stratified event, or (occasionally) attempts are made to balance out the strength of the field in different sections.

 

Anyways, I think Elianna's complaint was that Justin Lall (BBO star, professional player, and holder of around 2500 acbl masterpoints) was categorized as being in the lowest skill level in a recent acbl tourney. This seems a bit off, and she was wondering how the stratifications were determined.

[uday]

Let me summarize.

 

1. I have temporarily disabled the stratification while I sort this all out

 

2. Strat C is not Flight C. We can theoretically break the field into any number of strata (lets say 3). We'd like the strata to be of equal size. This is not a requirement.

 

3. All non-acbl-members must be put into the highest stratum (this is the ACBL rule I was working with)

 

4. So, tourneys with a lot of non-members seem to force the non-members into the top stratum.

 

I understand why the acbl wants non-members in the 1st stratum. I don't understand how I will be able to do that while simultaneously making sure Justin is also put into that stratum while the strata are of roughly equal sizes. I suspect it cannot be done.

 

One suggestion on the table is to assign equivalency points based on external factors. This is how the acbl sometimes deals w/this in real life (foreign players). But this can't currently be automated, and is easy to abuse. Maybe BBO Points should be what the equivalencies are based on.

 

Anyway, nothing is clear to me at this point. I'm thinking this over and discussing w/the ACBL and a few helpful members, trying to come up with something that allows ACBL members to benefit from stratification in an environment full of nonmembers.

 

 

ui

[tooncestdc]

I'm not sure if the strata are working as designed. I am Toonces on BBO and play with tsukata. I have 320 MP and she has 40 MP. We are usually being placed in flight A for tournaments or occasionally Flight B. See the 6/12/04 8:00 PM Pairs for an example.

 

If jlall is being placed in C and I am placed in A than either our tournaments had vast differences in non-members in this example or there may be bug in the program.

 

Also, I noticed that with the new system, we are a lot less likely to pay out the maximum number of MPs anymore. For example, in the 6/12 8:00 PM game, we had 68 pairs. We were divided into sections of 12, 11, and 11 tables, and the top payout was 0.96 or 0.88. If we were divided up into 2 17-table sections, the top payout would have been 1.20 and just about as many people would have placed. Is there a reason for keeping the sections and the awards smaller? Does this way actually pay out more overall MPs?

[JSilver]

Is this trip really necessary?

 

It seems to me there are plenty of winners in these tournaments without stratification to provide smaller awards to more people. If you play consistently, you are sure to scratch eventually, regardless of skill level. Meanwhile, players who want to be rewarded for beating a limited field can choose the recently added 299er game.

 

 

As someone with a tad over 1000 masterpoints (which, IMO, are a poor measure of skill), I've found myself in any stratum from A to C from one tourney to the next. It seems to me the only rational way to stratify, assuming it's really necessary, is to set fixed boundaries and forget equalizing the numbers in each, just as is done at face-to-face tournaments.

[Guest Jlall]

lol..just saw this thread, nice flight C baby. I didn't even know about this ;)

[tooncestdc]

The biggest problem with putting non-members into the top strat is that it pushes top players into the B and C strats, making it harder for true flight B & C players to scratch.

 

Here's a possible solution:

 

Place members and non-members into different sections. I see no reason that requires sections to have the same number of tables. In ACBL-ese, they are just two seperate events run with the same boards. In the "member" section, strats are done with MPs. In the "non-member" section, strats are done with BBO points or not at all. When necessary, we can always move a non-member into the member section as a Strat A pair.

 

Strats can either be done with fixed or varied sizes. But if we vary the sizes based on who has entered, instead of choosing equal-sized strats, it may be in the players' best interests to make the strat sizes so that the most people can scratch. For example:

 

Pairs Place % scratch

2 1 50%

3 1 33%

4 2 50%

5 2 40%

6 2 33%

7 3 42%

8 3 37%

9 4 44%

10 4 40%

11 4 36%

12 5 42%

13 5 38%

14 6 43%

15 6 40%

16 6 38%

 

Therefore, 4, 7, 9, 12, and 14 offer the most MPs/pair while 6, 8, and 11 offer the fewest MPs/pair. 6 is the worst option at 33%, but often is a default strat size when trying to balance the sizes.

[JSilver]

The biggest problem with putting non-members into the top strat is that it pushes top players into the B and C strats, making it harder for true flight B & C players to scratch.

 

Here's a possible solution:

 

Place members and non-members into different sections.  I see no reason that requires sections to have the same number of tables.  In ACBL-ese, they are just two seperate events run with the same boards.  In the "member" section, strats are done with MPs.  In the "non-member" section, strats are done with BBO points or not at all.  When necessary, we can always move a non-member into the member section as a Strat A pair.

 

Good suggestion.

[andych]

The biggest problem with putting non-members into the top strat is that it pushes top players into the B and C strats, making it harder for true flight B & C players to scratch.

 

Here's a possible solution:

 

Place members and non-members into different sections.  I see no reason that requires sections to have the same number of tables.  In ACBL-ese, they are just two seperate events run with the same boards.  In the "member" section, strats are done with MPs.  In the "non-member" section, strats are done with BBO points or not at all.  When necessary, we can always move a non-member into the member section as a Strat A pair.

 

Good suggestion.

I dislike the idea of segregating ACBL and non ACBL members.

I would rather prefer having no strats if no satifying solution available.

 

PLEASE DONT!

 

:) :lol:

[keylime]

Here's what I would recommend.

 

Since non-members are automatically placed into the highest flight, and since we are offering three flights, use some metric like "30-30-40" or "20-40-40" where by all the acbl members would be placed into the appropriate percentage?

 

Here on the Island, we have A LOT of under 200 players. Our breaks are 0-200/200-750/unlimited. As strange as the numbers appear, they do work for the most part.

[McBruce]

A simple solution which works well in my Wednesday night club game is this:

 

--Any pair with one or two non-members is in strat A

(actually, this is not normally done in club bridge, but I can see why it would be necessary online)

 

--Any pair of two ACBL member Life Masters is also in strat A.

 

--Any pair containing two ACBL members, one of whom is a Life Master, is in strat B.

 

--Any pair containing two ACBL members, neither of whom is a Life Master, is in strat C.

 

Simple--the software counts letters in the first character of player numbers.

 

There is absolutely no need to have any preset minimum sizes for a strat. If there are not enough pairs in a strat, there are simply no awards for that strat. But everyone is always eligible for the top strat.

 

If you have 16 pairs, 2 in A, 14 in B and 2 in C, your actual strat sizes are 16/14/2. You may not be able to give out awards in C but the C players are eligible for B and A. Note though that strat A, despite only 2 A pairs, has 16 pairs eligible.

 

Those who think that strats should be set higher are mistaken. Online points are virtually worthless to LMs. The purpose of the ACBL games should be to encourage non-LMs to play and earn points, and encourage non-members to join. What better (and simpler) way to do this than to stratify by number of LMs in a pair?

[tooncestdc]

McBruce, I think your idea is a good and simple one. But I was curious about:

 

Online points are virtually worthless to LMs.

 

What about online points makes you say that? Is it the limitation on overall points or what? As far as I can tell, the "online point" requirement are tougher on NLMs who can't move up unless 2/3 of the points at any level are "colored". As far as being past LM, a point is a point, essentially. There certainly doesn't seem to be much of an advantage for a LM to play at a club game over an ACBL online tournament, unless they are trying for a masterpoint race.

[McBruce]

McBruce, I think your idea is a good and simple one.  But I was curious about:

 

Online points are virtually worthless to LMs.

 

What about online points makes you say that? Is it the limitation on overall points or what? As far as I can tell, the "online point" requirement are tougher on NLMs who can't move up unless 2/3 of the points at any level are "colored". As far as being past LM, a point is a point, essentially. There certainly doesn't seem to be much of an advantage for a LM to play at a club game over an ACBL online tournament, unless they are trying for a masterpoint race.

I don't know the current rules for online masterpoints very well, but the ACBL is never going to give out points online in the same proportions that it gives out points at Regionals and NABCs, where winning a mid-level KO (4 wins) will get you several dozen gold points for four sessions of bridge against people who are about at your masterpoint level. How many online tournament wins must one get to score 24 masterpoints?

[Elianna]

How many online tournament wins must one get to score 24 masterpoints?

Usually 24 tournament wins.

 

I think that the reason they may not be useful for LMs is that they don't count for MM races like club points do, which also seem to acrue at the same level (one club win is usually about 1 point).

 

They are useful in that they count for total points, so you can move up levels (and the more points you have, the more seeding points for various stuff, but as was pointed out, they don't come in great enough quantities to be significant in this regard for many people).

[keylime]

I don't mind online games, but I think awarding 80% club rated for 12 boards, when the average clubs plays MORE than twice that number of boards, is excessive.

 

Now, if it was 65%, then I might have a rethink.

 

If online points don't matter, just ask my pard - she made LM on the ability to get online points once her other colors were finished. They are necessary.

[McBruce]

A club game can be worth as much as 1.50, more if it is a club championship or other special game. Most club TDs do not realize that playing a Howell or an arrow-switch movement increases the masterpoints for the winner, while eliminating the biggest problem with Mitchell movements: 3rd place and masterpoints in one direction may be a worse game than 5th in the other direction. If you have eight tables and play a Mitchell each direction winner gets 0.80 (0.10 per pair in each direction), but if you instead play an arrow switch in the last round and make it a one-winner game, there are sixteen pairs in the field and the winner gets 1.50.

[Elianna]

A club game can be worth as much as 1.50, more if it is a club championship or other special game. Most club TDs do not realize that playing a Howell or an arrow-switch movement increases the masterpoints for the winner, while eliminating the biggest problem with Mitchell movements: 3rd place and masterpoints in one direction may be a worse game than 5th in the other direction. If you have eight tables and play a Mitchell each direction winner gets 0.80 (0.10 per pair in each direction), but if you instead play an arrow switch in the last round and make it a one-winner game, there are sixteen pairs in the field and the winner gets 1.50.

It also decreases the total points awarded, increases confusion (especially at tables where one pair stays at a table, and has to keep switching directions. they inevitably forget at least once), and increases randomness.