What the heck is NABLA?

[Michael Dorosh]

Sorry for not keeping up more with the tourney threads, I am not all that competitive when it comes to CM.

But I notice the term NABLA scoring being bandied about the Boots and Tracks tourney thread, and since I

am a competitor in that one, I must say I am interested.

Can someone tell me what NABLA stands for/refers to, and who invented it/where it comes from?

If I am understanding what I am reading correctly, it seems quite clever. Correct me if I'm wrong, but it means

that a group of players play identical scenarios, and then their scores are compared to each other - all the Allied

players against the Allied players, and the Germans against the Germans, with scoring being done by comparing

them along those lines rather than in Allied vs. German terms, as a true test of their skill at that particular

scenario.

I read the recent thread with the the twist being added of taking the top score on each side and making it 100,

with the others being a percentage of that.

So again, correct me if I am wrong, but given a sampling of x number of games, if all the games ended in a

60-40 win for the Allies, all the competitors would be ranked at 100 percent?

[JasonC]

I don't know where the abbreviation got started, but I saw that scoring system proposed for one tournament. So I assume it just refers to that tournament, generalized to the scoring method to apply it elsewhere.

As for the consequences of it, yes, if everyone does exactly as well on both sides, all players would be scored as 100. The score measures what percentage of the best performance for one's side, each player achieves. So if they are all exactly equal, each has done just as well as the best on their side, and so would merit 100.

If on the other hand, one player consistently won 80-20 with the Germans, then other Germans would be "held" to that "standard". And e.g. one who also won every game, but only 60-40 on average, would only score 75%. The "normalized" side based scoring system is meant to measure relative skill of the players on the same side, not anything between sides, and is meant to "factor out" side-based advantages.

That is my understanding of it, anyway. I didn't come up with the system.

[The Tanklord]

Well, firstly nabla is a math term. (might be a name for something else too)

One user on these board has it as username and I think he came up with the idea the scoring system so it was named after his username.

I might be wrong, it has happened before

[Enoch]

Michael,

As others have pointed out the NABLA scoring system was designed to score recent CM tourneys. It was first used with the original Rumblings of War tourney.

In short the system compares the score of the players playing each side in a scenerio. Each player is scored against the median score for that side. So if you score above the median you get more "points", if you score below the median you get less "points." This design allows for the use of unbalanced scenerios in a tourney since each player is only being compared to other players who played the same side of that scenerio. It also minimizes the effect of outlying scores. So, if someone does really well in a scenerio as compared to everyone else playing that side, the system somewhat lowers their score in relation to the other players than if you were simply using the raw raw CM scores.

Here is the thread that gets into the dirty details of how the system works.

This thread has the final results from the first RoW tourney and you can see hoe the system works.

[ June 11, 2002, 02:41 PM: Message edited by: Enoch ]

[jgdpzr]

Whew, for a second there I thought MD was interested in finding the nearest NAMBLA chapter.

[Sitzkrieg]

Originally posted by jgdpzr:

Whew, for a second there I thought MD was interested in finding the nearest NAMBLA chapter.

He could have just asked a Catholic priest.

[John D Salt]

Originally posted by The Tanklord:

Well, firstly nabla is a math term. (might be a name for something else too)

One user on these board has it as username and I think he came up with the idea the scoring system so it was named after his username.

I might be wrong, it has happened before

http://www.lacim.uqam.ca/~zabrocki/nabla/

for people who enjoy terrifying themselves with incomprehensible maths.

All the best,

John.

[MrSpkr]

Originally posted by sitzkrieg:

</font><blockquote>quote:</font><hr />Originally posted by jgdpzr:

Whew, for a second there I thought MD was interested in finding the nearest NAMBLA chapter.

He could have just asked a Catholic priest.</font>

[Treeburst155]

M. Dorosh,

There are basically two types of the Nabla Scoring System developed by Jarmo Hurri (Nabla) of Finland. He came up with them so unbalanced scenarios could be used in competitions.

The main Nabla system assigns a score for a player based on the median score achieved by all tourney participants who played that same side of a particular scenario. This score is arrived at through a rather complex formula created by Nabla and then packaged into a DOS program created by Nabla for me.

The second Nabla system is for smaller competitions where there are not enough samples to determine an accurate median. In this system there is no complex formula. If you score the best of four players who played the same side of the same scenario, you get three points. The person who scored the worst from that side gets 0 points for that scenario. Second highest gets 2 points, third gets one point for that scenario.

"Broken!" has recently proposed "normalizing the scores to 100" in the second system. This simply awards points in a more proportional manner based on relative performance. If four Allied scores for a scenario are 80,50,40, and 20 the original Integer Nabla would assign scores as follows:

80=3, 50=2, 40=1, 20=0

Note that the final points assigned are not proportional to the raw CM points.

"Broken!'s" normalized scores would look like this:

80=100, 50=62.5, 40=50, 20=25

The highest score is considered perfect, and all other scores are adjusted to the same percentage of the high score they originally had. This method maintains the actual relationship of players' raw scores.

Treeburst155 out.

[ June 11, 2002, 04:40 PM: Message edited by: Treeburst155 ]

[Michael Dorosh]

Thanks to all for the replies.

So what advantage do the normalized scores have over a normalized "raw" score?

[Treeburst155]

The highest score from each side will always have equal value (100). In an unbalanced scenario, 30 points may be the high score for a side. This score should be equal in value to the high score from the strong side. Normalizing to 100 allows this.

Treeburst155 out.

[CombinedArms]

TB,

Which of the two Nabla scoring systems are we using in ROW II, the B&T Brawl. I'm guessing it's the more complex, first-mentioned system, based on Nabla's DOS program?

[Treeburst155]

Yes, we are using the main Nabla system since every scenario is being played 36 times. We will get an excellent median from that many scores.

Treeburst155 out.

[Nabla]

Originally posted by Michael Dorosh:

Sorry for not keeping up more with the tourney threads, I am not all that competitive when it comes to CM. But I notice the term NABLA scoring being bandied about the Boots and Tracks tourney thread, and since I

am a competitor in that one, I must say I am interested.

Can someone tell me what NABLA stands for/refers to, and who invented it/where it comes from?

That would be me.

Sorry for the late response, but I've been on holiday for the past week. Treeburst has already given a good description of the system above. If you want to know the details and the evolution of the system see the following long thread

http://www.battlefront.com/cgi-bin/bbs/ultimatebb.cgi?ubb=get_topic;f=1;t=021512

Another person has requested the source of the program set from me while I was on holiday. I will make it available for download this evening here (you can already find the DOS binaries there). The program code is standard C++, so you should be able to compile it under a variety of different operating systems. I have compiled it under DOS and Linux.

If you have any other questions, feel free to ask.

[Nabla]

Program source is ready for download now.

[acrashb]

Gents, I'm digging through the NABLA system with an eye to setting up an online tournament management system that uses NABLA to score.

No garantees, but if it turns out to be not too much work (i.e., if it's a good self-learning project that doesn't turn into a lifestyle), you'll have access to it.

I would expect some progres in a week or two or three.

[Treeburst155]

Hey, that sounds great, Aaron! I hope it works out. Thanks for taking a crack at a project like that.

Treeburst155 out.

[WineCape]

Hope to hear from you soon with this aaronb! It might just save Treeburst's marriage as he is spending way to much time in from of his 'puter of late managing all and sunder.

Sincerely,

Charl Theron

Chess … is the most beautiful and reasonable of all games – Madame de Sėvignė

Chess … is the game which reflects most honour on human wit -- Voltaire

Chess … is a sad waste of brains -- Sir Walter Scott

Chess … is a beautiful mistress -- Bent Larsen

Chess … is a cure for headaches – John Maynard Keynes

Chess … is vanity – Alexander Alekhine

Chess … is a fight – Emanuel Lasker

Chess … is a test of wills – Paul Keres

Chess … is the art of analysis – Mikhail Botvinnik

Chess … is the art of battle for the victorious battle of art – Tartakower

Chess … is imagination – David Bronstein

Chess … is the struggle against error – Johannes Zuckertort

Chess … is art, science and sport – Anatoly Karpov

Chess … is too difficult to be a game, and not serious enough to be science or art – Napoleon

Chess … is the touchstone of the intellect – Siegbert Tarrasch

Chess … is me – Salvador Dali

Chess … is not merely an idle amusement; life is a kind of chess – Benjamin Franklin

Chess … is beautiful enough to waste your life for – Hans Ree

Chess … is most certainly not my life – Tony Miles

Chess … is life in miniature – Gary Kasparov

Chess … is like life – Boris Spassky

Chess … is my life – Viktor Korchnoi

Chess … is life! – Bobby Fischer

[acrashb]

Originally posted by WineCape:

Hope to hear from you soon with this aaronb! It might just save Treeburst's marriage as he is spending way to much time in from of his 'puter of late managing all and sunder.

Sincerely,

Charl Theron

Good to hear there's some interest - especially from a tournament 'sponsor' - and if I can save TB's marriage, well, then he'll owe me one

I think the data structures are in good shape, now for the event analysis (process flow charting).

The analysis is not too bad, although I'm rusty. What will take a bit of time is self-learning a number of technologies that I haven't used before, to do web delivery.

The initial degree of automation will be modest. Minimally, players can enter and review their own scores, and the TourneyMaster will be able to do mass emails for communicating. Multi-platform, as best as I can manage.

[tar]

I suppose one other scoring variant would be to use something like the SAT scores from the US College Entrance Exams. This would be used only as a method of determining points for the players on each side.

Briefly, the method is as follows.

Scores are assumed to be normally distributed. A Mean Value and Standard Deviation are computed. Any person with a score equal to the mean value gets a score of 500. Each standard deviation (up to a maximum of 3) is worth plus or minus 100 points. This is scaled linearly.

Example. Mean score 35, standard deviation of 10. A player with a score of 50 is 15 points or 1.5 standard deviations above the mean. That would give them 500 plus 1.5 times 100 = 650 points.

If you want to keep the scores lower, then one could use a 20 to 80 or a 2 to 8 scale instead.