Scoring and scoresheet

[hammy]

After a fair degree of confusion and some odd results when using the far more precise calculations from the FoG scoring program that is currently under development it has become clear that there is an issue with the scoresheets.

In order to sort things out so that manual scoring and the computer system actually get the same results a new score calculation sheet has been produced.

Essentially the FoG scoring system is 0-10 points for the proportion of your army's breakpoint remaining plus 0-10 points for the proportion of your opponents breakpoint you have taken plus 5 points if you break the opposing army without yourself breaking.

The old scoresheet rounded the 0-10 score to an integer, when you add two rounded numbers you can end up with quite significant rounding errors. The software worked things exactly and it was not uncommon for a score calculated from the table on the sheet to be at least 1 point different from the score the program calculated. This combined with players not really understanding how to fill out the score sheet has meant that the score cacluation table has been removed from the scoresheet and all you need to put now is number of BGs and number of AP lost for each army. The score calculation table has been expanded and separated, it now works to 1 decimal place and the software has been modified to round to the same point. As a result manual scoring and computer scoring will produce the same result.

I used this table to run manual scoring at the Games Expo comp and it proved easy enough to use and from a player point of view just having to fill in AP lost made lives easier too.

The scoreslip and calculation table can be found at:
http://www.fieldofglory.com/competition/scoresheet.html

My understanding is that tournament organisers wishing to use the FoG draw software are welcome to do so but need to contact Slitherine to get a copy sent to them.

[dave_r]

Am I the only one who thinks this is somewhat more difficult than it should be?

Surely after your game it should be possible to actually work out who won and who lost? This could be potentially tricky when club games are thought about...

[nikgaukroger]

No Dave you are quite right.

The solution should have been to deal in whole integers and change the software to match that so that the manual sheet was easy and seen to be easy in calculating the score, putting in all the decimal places make it look more difficult (even though the math is simple) and messy.

This is what comes when you let science based people do things

[hammy]

Most people have had lots of problems with the old manual sheet which is the main reason behind pulling the how to calcuate your score bit off the bottom. If you want to know who won you can work it out.

The system could still round to integers but you do get issues where by rounding to integers if for example I lose 15% of my army break and you lose 14.9% of yours that then a game that really should be a draw is scored as 11-9. You can also get a game where one army loses 24.9% and the other loses 15% which ends up as a 10-10 draw.

Having an extra decimal place significantly reduces the odities in the scoring system.

[dave_r]

Most people have had lots of problems with the old manual sheet which is the main reason behind pulling the how to calcuate your score bit off the bottom. If you want to know who won you can work it out.

No you can't - unless you have a copy of that table thingy then it is not possible. To prevent people from getting an unfair advantage (presumably by actually being able to tell the score following a game) this was removed from the Challenge score sheets because people were getting it wrong because the computer model worked it out differently.

The system could still round to integers but you do get issues where by rounding to integers if for example I lose 15% of my army break and you lose 14.9% of yours that then a game that really should be a draw is scored as 11-9. You can also get a game where one army loses 24.9% and the other loses 15% which ends up as a 10-10 draw.

Perhaps I am missing the point here but is this not the tail wagging the dog? Surely it would have been easier to actually fix the computer model so that it actually mirrored the printed results? Even I could program that.

Having an extra decimal place significantly reduces the odities in the scoring system.

Does it? As far as I was aware there were no oddities in the scoring system and in fact the only oddities were in the computer model which attempted to mimic the written scoring system. The fact that this has meant a massive change to how we calculate scores manually is to me the completely illogical thing to do.

I had a game at the club on Monday and somebody asked me afterwards who won. I replied I didn't know... Sums it up really.

[hammy]

OK,

The old scoring sheet had some rather nasty rounding errors.

The scoring is:

You: Your %age left + Opponents %age lost
Opponent: essentially 20-your score.

I don't have a copy of the 25-0 score sheet to hand so can't point out the particular problems but if there is rounding in the scoresheet it needs to be rounding you can work out yourself if you need to.

If we play a game where I have say 11 BGs and you have 12 and we both lose 6 APs then you have edged a win as I lost a higher percentage of my AP than you. By the scoresheet this might be a 10-10 or an 11-9. If you plug the numbers into the new sheet, I have got 4.5 tenths of my break left and you have 5 tenths of yours left so you have won 10.5 to 9.5 which rounds to 11-9.

The army list spreadsheet currently has a space that it fills with points lost per AP. I am sure that changing it in line with this sheet would work perfectly well.

[dave_r]

Would multiplying by 10 achieve exactly the same results?

Tip: If you divide by 10 at the end then.....

[philqw78]

I never knew how the scores were calculated until now. I just filled in the sheet and handed it in for the marvellous mechanical mouse organ to calculate, or guessed. Now I can sit wondering what is the best number of BGs to have.

[dave_r]

The best number of Battle Groups is approximately 85.

That should ensure you win more than you lose

[hammy]

Would multiplying by 10 achieve exactly the same results?

Tip: If you divide by 10 at the end then.....

It has exactly the same outcome but we decided to remove the need to divide by 10 by just putting the numbers with the decimal in them.

Most people aren't that bothered about their exact score hence a scoreslip with just AP lost the exact numbers can be done by the scorer.

There were definitley areas in the old scoresheet where there were some really big rounding issues. This is a halfway house between the way the computer used to do it and the printed sheet. The computer software has been changed to work to 1 decimal and the score calcuator now includes the same detail. the two are the same so you can if you want work out your exact score which was the whole point of the exercise.

[dave_r]

Most people aren't that bothered about their exact score

Hmmm, that must be why I never see people clustered about the results trying to work out who they are going to play next in a comp?

[hammy]

Most people aren't that bothered about their exact score

Hmmm, that must be why I never see people clustered about the results trying to work out who they are going to play next in a comp?

If you want to work it out just carry a copy of the sheet with you or look at the wall next to the scoreboard where there will be copies of the sheet for your use.

[SirGarnet]

I had a game at the club on Monday and somebody asked me afterwards who won. I replied I didn't know... Sums it up really.

Sounds realistic - sometimes you don't know who won a battle until well afterward - sometimes the historians have to decide.

[peterrjohnston]

Am I the only one who thinks this is somewhat more difficult than it should be?

I think so too, as basically the score is:

100% - winner's percentage loss* + loser's percentage loss*

Divide by 10

Round to nearest integer.

Loser's score is always 20 - winners.

Add 5 if you've ground them into the dust like the dogs they are, or, to be precise, were.

*ie BG lost/BG total as a percentage


That table is, well, maybe best not said.

Rgds,
Peter

[hammy]

I think so too, as basically the score is:

100% - winner's percentage loss* + loser's percentage loss*

Divide by 10

Round to nearest integer.

Loser's score is always 20 - winners.

Add 5 if you've ground them into the dust like the dogs they are, or, to be precise, were.

*ie BG lost/BG total as a percentage

That table is, well, maybe best not said.

I am open to suggestions if you think there is a better way of doing it.

The old scorsheet was not right and there were a lot of situations where you could end up with significant errors against the correct score if you used the scoresheet. Yes the computer could be made to work exactly at the score sheet but then people would aim their armies to 13 or 14 or however many BGs got the best of the rounding error on the scoresheet.

Yes this sheet has a lot of numbers on it. Players can always work out their scores with a calculator or even in their heads if they are good enough at mental arithmetic. With the previous score sheet was almost impossible to work out an exact score, you could get an idea that it would be say 11 or 12 against 8 or 9 but without the table it was impossible to be sure. A significant minority of players really had no idea how to work out the score and errors were comonplace hence this atttempt to 'clarify' things.

[nikgaukroger]

I am open to suggestions if you think there is a better way of doing it.

Easy - make the score sheet only deal in whole integers, make sure the software works the same way. Much better perception from players/potential players and much cleaner presentation.

I really don't think players trying to get the better of the rounding will be a material issue. Anyway the route of mathematical "purity" just leads to gecko and accelerated pairings so lets not go down that roue

[hammy]

I am open to suggestions if you think there is a better way of doing it.

Easy - make the score sheet only deal in whole integers, make sure the software works the same way. Much better perception from players/potential players and much cleaner presentation.

I really don't think players trying to get the better of the rounding will be a material issue. Anyway the route of mathematical "purity" just leads to gecko and accelerated pairings so lets not go down that roue

Possibly.

I was going to reply but it is just going round in circles.

Rounding to full integers does introduce some reeally significant discrepancies. Yes the program could round to integers but you would still need a table the same size as the one on this scoresheet so what is the issue.

All a player needs to do is to record the number of BG lost. If they want to work out their score there is a chart that they can look things up on. I don't see how integers or one decimal point actually makes a difference.

[timmy1]

Hammy, I am with you but not sure how much that helps. I don't have any difficulty with the new chart (but then I did not with the old as losing all my BG gives the same result no matter how many I have...). I am all for mathmatical and intellectual purity but remember we live in an age where many people can't count beyond 2 without a calculator or spreadsheet despite have a Maths 'A' level.

Nik, I think Hammy is right. The really competitive types out there will work out the size that gives the best rounding benefit and we will have the forums dominated by the 'what is the best number of BG and best number of bases per BG for a competition win' rather than the FAR more important questions about how Sassanids should be graded and are Shooty Cav undercosted?

[peterrjohnston]

I am open to suggestions if you think there is a better way of doing it.

Yes, as Nik says, the old version with integers.

The old scorsheet was not right and there were a lot of situations where you could end up with significant errors against the correct score if you used the scoresheet. Yes the computer could be made to work exactly at the score sheet but then people would aim their armies to 13 or 14 or however many BGs got the best of the rounding error on the scoresheet.

So correct the old scoresheet. And the errors can be no more than 1 point per game, assuming, of course,
you hit a particular combination of loses. Not something I'd regard as significant in the context of 100 points.
And frankly, if players are "optimising" there army to take advantage of a kink in the table to possibly
gain 4 points in a 4 game tournament, they're almost certainly penalising themselves in army design.

With the previous score sheet was almost impossible to work out an exact score

You just cross-referenced as the new table, so I don't see it as any different. But adding decimal places for
many people will make it look like a maths problem. Or do you mean the spreadsheet(?) tournament program
couldn't work out the correct score? If it is a spreadsheet, then it's perfect for using a look-up table based on
integers.

It's worth bearing in mind at some tournaments score and draw is still done by hand. Adding decimal
places guarantees more errors, believe me.


I have no problem with either table from a calculation point of view. As Nik says, it's the
majority that matters.

[peterrjohnston]

Hammy, I am with you but not sure how much that helps. I don't have any difficulty with the new chart (but then I did not with the old as losing all my BG gives the same result no matter how many I have...). I am all for mathmatical and intellectual purity but remember we live in an age where many people can't count beyond 2 without a calculator or spreadsheet despite have a Maths 'A' level.

A bit of calculus might throw me on a groggy Sunday morning , but you're right, it's the average punter one needs to think of.

Nik, I think Hammy is right. The really competitive types out there will work out the size that gives the best rounding benefit and we will have the forums dominated by the 'what is the best number of BG and best number of bases per BG for a competition win' rather than the FAR more important questions about how Sassanids should be graded and are Shooty Cav undercosted?

So let them. There will always be this, although optimising to maybe achieve one extra point per game isn't really worth worrying about.

Rgds,
Peter