[tournament-org] WhereIsTheBar

Tue Oct 2 22:57:27 BST 2018

[[inline]]

Geoff

33 Ashbury Close, Cambridge CB1 3RW 01223 710582

On 02/10/2018 21:49, Alison Bexfield via tournament-org wrote:
> I think we are overcomplicating this.  Go is a game of skill with no
> element of luck. A tournament is where players come together to see how
> they perform against others.
> Tournaments developed to find the strongest player.
> 
> The idea is that participants are treated equally to give an open
> competition from which a winner emerges.  I believe (from personal
> conversation but not scientifically proved) that most dan players are
> happy to play even games even when the opponent on paper is stronger by
> a number of games.  I have already stated I enjoy those games - I start
> on an equal footing and with a chance of winning, even if this is only
> an outside chance based on probability. Even if I lose I have a game to
> analyse and from which to learn.  (At the top end of the tournament we
> are not talking about sandbaggers as most players have established
> grades.)  These days, it is entirely possible that someone with a grade
> of one dan has done some serious study on the internet which has upped
> their playing strength and they want to test that out in real life and
> will play at a higher level than 1 dan when they do so.
> 
> The McMahon system evolved because it was realised that once grades
> become too far apart games become extremely one sided and no fun to the
> players.  For most people without a realistic chance of winning the
> tournament the McMahon system works to give evenly matched games. This
> is far superior to the swiss system used in a chess tournament where
> often the first two rounds are foregone conclusions and a waste of time
> for all.  But the bar is needed in the McMahon system to create that
> open competition at the top if the purpose is to find the strongest
> player on the day.
> 
> If we are advocating a different system where grade bands are used
> through out with no bar, we are running a handicap tournament and a
> tournament to find the strongest on the day player and should state
> this. 
[[
A. All I am suggesting is that where the top grades have gaps that are 
very large you run a strict McMahon with the bar set at 1 grade above 
the highest. Then the usual rule of hcap  = MMS diff - 1 applies and 
where there are no bad grade gaps players will get normal games.
So it is not a handicap tournament - it is McMahon.

B. A handicap tournament is often drawn by hand.  But you can run a 
Swiss Handicap where you use Swiss rules. This pairs players on the same 
score at random. Apply a handicap where appropriate. The same happens in 
McMahon where we are still trying to pair players on the same MMS.  Only 
at the very top in the example being discussed is this impossible.

C. I of course agree that the system should be clearly stated. All you 
would need to say is "We are running a McMahon tournament and the bar 
may be set at above the highest grade if there are large gaps at the top.
]]

  My feeling is that handicap tournaments have their place but the
> top players may choose to avoid them unless part of a longer event.
> 
> ---
> 
> Alison Bexfield
> On 02/10/2018 13:10, Neil Sandford via tournament-org wrote:
> 
>> What this means is that there are actually two competitions being fought at the same time. One will be won (almost certainly) by the 5D. The other is for recognition of the highest placed person who was not entered into the above-bar competition.  If you have a large number of equally ranked strong-ish (say 2D-1K) entrants, some will not be able to play their stronger peers unless there are exactly (rounds + 1) players above the bar - four in a typical one-day comp, seven for a weekend.   Can the split be done by GoR?
>>
>> Neil
>>
>> FROM: TobyManning via tournament-org <tournament-org at lists.britgo.org>
>> SENT: 02 October 2018 12:12
>> TO: tournament-org at lists.britgo.org
>> SUBJECT: Re: [tournament-org] WhereIsTheBar
>>
>> Neil:
>>
>> The purpose of the "bar" is predicated on the premise that the strongest players would prefer to play even games not handicap games, even if their proability of winning was less than 50%.
>>
>> Once you accept this hypothesis (which has been well tested), the bar follows almost automatically.
>>
>> In Geoff's example, if the bar is at 5 dan (but all games at the top are to be even) then the 5 dan will win the tournament even if he (she) loses all 3 games.
>>
>> To (mis)quote the Organiser's Handbook, without the bar the strongest players get an unreasonable (and uneccessary) advantage.
>>
>> In a tournament with a more normal entry at the top, say 2d/2d/1d/1d/1d/1k/1k/1k, with the bar at 1 kyu gives all 8 people an equal chance of winning the tournament - all they have to do is to win 3 games. If there was no bar then it would be practically impossible for a 1 kyu to win the tournament, and the 1 dans would be at a disadvantage compared with the 2 dans despite having to play them.
>>
>> Toby
>>
>> On 02/10/2018 11:51, Neil Sandford via tournament-org wrote:
>>
>> The bar has always mystified me. Why should the strongest player(s) be given a less challenging draw than the rest of the field? And why the next-to-strongest be penalised by not being given a handicap when playing another (stronger) player from above the bar?
>>
>> The purpose of the bar and the mechanism of handicap are two separate discussions. Perhaps we can learn from the golf approach to handicapping?
>>
>> Players are given an initial handicap (ranking) from assessment based on performance in three games. They can only play in a competition without a handicap by playing off zero (i.e. by being given the same rating as the strongest player in the competition - in Geoff's case, 5D).
>>
>> Given that that is unlikely to be very pleasant for either player, there is a provision for assessing three results from games at your own club __before__ the tournament. This is most frequently done through a discussion like "You are playing to 8kyu at the club but getting stronger. I suggest you enter as 7kyu".
>>
>> Then there is also https://senseis.xmp.net/?RankWorldwideComparison
>>
>> This attempts to normalise the various rating systems, including the main online servers. I used it in an inter-club competition on Saturday and was at best apprehensive. An AGA 1D was entered at 3kyu, for example. At the end of the day, everyone agreed that the match had been very close (8 games, typically within 10 points). The AGA 1D won both his games. But someone who regularly plays club games at his EGD rating (6kyu) lost both games. Geoff processed the results and only came up with one recommendation (for an unrated player entered as 4kyu):
>>
>> Beats NG(10.2k -5st) loses RP(5.8k - 2st) So 4k OK but 5k probably more realistic.
>>
>> Neil
>>
>> FROM: Gerry Gavigan via tournament-org <tournament-org at lists.britgo.org>
>> SENT: 01 October 2018 16:35
>> TO: ptm at tobymanning.co.uk
>> CC: TobyManning via tournament-org <tournament-org at lists.britgo.org>
>> SUBJECT: Re: [tournament-org] WhereIsTheBar
>>
>> That's an extreme interpretation of my comment. I'm not sure why you are inferring "no bar ever"
>>
>> In this example, as with Cork, there is a sparse entry.
>>
>> With no handicap the 5 dan will be shooting fish in a barrel and everyone else will be a fish.
>>
>> My understanding of the McMahon system is that it is designed to give evenly ranked games where possible. In a sparse entry that is not possible and to play without handicaps seems to negate the rating system.
>>
>> One lemma and two asides
>>
>> The lemma: if I cannot choose to enter at a grade above my current strength there seems to be a conjugate that I should not be required to play above my current strength (implicit in having a bar in a sparse tournament)
>>
>> Aside 1: At the EGC I was badly sandbagged by a Chinese player entering his first European tournament way below his actual strength (as he confessed to me after) It's no fun and if one took a simplistic calculation of total cost of attendance divided by number of games inmai tournament, expensive too.
>>
>> On 01 October 2018 at 15:57 ptm at tobymanning.co.uk wrote:
>>
>> Gerry:
>>
>> Are you seriously saying there should NEVER be a bar, or are you saying that in this extreme case, the bar should be at 5 dan?
>>
>> Toby
>>
>> On 2018-10-01 14:40, Gerry Gavigan wrote:
>>
>> As set it sounds to me, on behalf of all of those several stones weaker than the 5 dan, that there should not be a bar
>>
>> By any measure the 5 dan has already won.
>>
>> As a 1 kyu (I wish) without a handicap I am going to have a rubbish time.
>>
>> The purity of some tournament protocol should be secondary
>>
>> If the tournament police are going to insist on a bar the TD should consider bribing the 5 dan not to enter for the greater good and encouraging the potential lambs-to-the-slaughter to turn up next year.
>>
>> On 01 October 2018 at 12:20 TobyManning via tournament-org < tournament-org at lists.britgo.org> wrote:
>>
>> I set it at 1 kyu, but if an organiser set it at  2kyu I would not complain.
>>
>> Setting it at 3 kyu would be wrong.
>>
>> Toby
>>
>> On 01/10/2018 11:45, Geoff Kaniuk via tournament-org wrote:
>>
>> You have just registered your last player who happens to be 1d and
>> people are waiting impatiently,  wanting to get on with your 3 round
>> McMahon tournament.
>>
>> You are now faced with with the problem of where to set the the bar.
>> Suppose in this tournament the top players are:
>>
>> 5d 1d 1d 1k 1k 2k 2k 3k 3k 4k 4k 5k .....
>>
>> Assuming all players enter at a realistic grade, where would you set
>> the bar?
>>
>> It would be interesting to see your instinctive answer, rather than
>> consulting the Tournament Organiser's Handbook.
>>
>> _______________________________________________
>> tournament-org mailing list
>> tournament-org at lists.britgo.org
>> http://lists.britgo.org/cgi-bin/mailman/listinfo/tournament-org
> 
> _______________________________________________
> 
> tournament-org mailing list
> 
> tournament-org at lists.britgo.org
> 
> http://lists.britgo.org/cgi-bin/mailman/listinfo/tournament-org
> 
> _______________________________________________
> tournament-org mailing list
> tournament-org at lists.britgo.org
> http://lists.britgo.org/cgi-bin/mailman/listinfo/tournament-org
> 
> 
> _______________________________________________
> tournament-org mailing list
> tournament-org at lists.britgo.org
> http://lists.britgo.org/cgi-bin/mailman/listinfo/tournament-org
> 