khaile
Joined: Nov 08, 2005
|
 Posted:
Dec 14, 2005 - 17:29 |
 
|
EDIT: Sorry, posted in the wrong forum. Should be under "features".
I think the Luck meter is nice, but the way it is calculated seems weird because of the nature of the game. The problem is that, most of the time, when you fail with a roll your turn ends. In other words, you have removed all your "chances" to fail with another roll. This is why a majority of the games have more than 50% luck for both teams.
Or this way: Consider an attempt to dodge and pick up the ball and a GFI with a Dwarf Longbeard (two 4+ rolls and a 2+ roll), and the Luck meter is zeroed:
- The dodge fails: You will have 0% luck.
- The dodge succeeds, but the pickup fails: You will have 50% luck.
- Both rolls succeeds, but the GFI fails: You will have 40% luck.
Why are you more lucky if you drop the ball on the first roll? That doesn't make sense. Consider what it would look like if you GFI'd first and picked up the ball after:
- Dodge succeeds, GFI fails: 25% luck
- Dodge succeeds, GFI succeeds, pickup fails: 61.54% luck
Oooh, I feel so lucky  Still a turn-over though. (Of course, in this example I'm really pushing it. The chance to succeed with these rolls is only 20.83%, but that's not really the point).
Another example. Let's say you are playing Elves against a bashy team. You decide to force your opponent to blitz his next turn, and then have to dodge away with four linemen. The chance of succeeding with four 2+ rolls is actually rather small (48.23%, a binomial distribution), but this is what it would look like on your luck meter:
- First fails: 0%
- Second fails: 16.67%
- Third fails: 28.57%
- Fourth fails: 37.5%
I don't have any real arguments, it just "feels" weird.
Ok, so to the point: In bloodbowl, luck is much about "avoiding turnovers". Perhaps there could be a counter, in addition to the luck meter, telling you your "Turn-over risk points", TORP (the acronym invents the meaning, hehe). It would be calculated somewhat like this:
Each time you succeed with a roll that involved the risk of getting a turn-over, you receive points depending on that risk. For example, dodging with an elf has a 16.6% risk of causing a turn-over, so if you succeed with your roll you gain 17 TORPs (a rounded percentage). In other words, succeeding with 4 dodges gives you 68 TORPs. If you fail with a roll (causing a turn-over), you just get 0 TORPs.
Note that only turn-overs are considered here, so no "neutral" outcomes or unrelated stuff such as armor or injury rolls. These are covered by the usual luck meter. (Ie, blocking with a Dwarf Longbeard without coming up with a skull is worth 17 TORPs).
Rolls should be grouped by the current action. For example, a block with several dice are calculated together (ie, only 2 TORPs for a player with block), and a pass should include both rolls for passing and rolls for catching in one sum (pPassFailure*pCatchFailure). All failures are still considered 0 TORPs, so re-rolls are automatically ignored (ie, failing a 3+ pass and succeding it on the re-roll is 0 + 33 TORPs, there's no "unluck" to have to use a Pass skill).
In the end, this is just another value, interpreted in a different way.
Ok, no more typing.  |
|
|
blizzt95
Joined: Apr 12, 2005
|
| Posted:
Dec 14, 2005 - 17:36 |
|
Good idea in theory. Doubt it will be implemented. |
|
|
shadow46x2
Joined: Nov 22, 2003
|
| Posted:
Dec 14, 2005 - 17:38 |
|
i hate you for bringing this up....:-p
simple solution.....
don't use the luck meter
--j |
_________________
| origami wrote: | | There is no god but Nuffle, and Shadow is his prophet. |
  |
|
f_alk
Joined: Sep 30, 2005
|
| Posted:
Dec 14, 2005 - 18:35 |
|
| khaile wrote: | | I think the Luck meter is nice, but the way it is calculated seems weird because of the nature of the game. The problem is that, most of the time, when you fail with a roll your turn ends. In other words, you have removed all your "chances" to fail with another roll. This is why a majority of the games have more than 50% luck for both teams. |
Agreed. Still, the luck meter is a very good indication of who has more luck than the other. It is not the absolute value that should be of concern for you, but the difference.
| Quote: | | Or this way: Consider an attempt to dodge and pick up the ball and a GFI with a Dwarf Longbeard (two 4+ rolls and a 2+ roll), and the Luck meter is zeroed: |
The luck meter is designed to work with 50% starting value. Setting it to "zero" massively falsifies the effect. For the following, i assume that with "zero" you mean 0/0... any 0/n would lead to different results depending on the n.
| Quote: | | - The dodge succeeds, but the pickup fails: You will have 50% luck. |
You are at 66.67% following your scenario with the "zeroed" Luckmeter.
If the luck meter starts with 1/2, you would have a rating of 72.72 percent.
If the luck meter starts at 50/100, you would have a rating of 50.7 percent.
(Just to show you the effect of the starting point).
The calculation of the luck meter is not such a problem. It indeed is very elegant. The problem is setting the starting point.
| Quote: | | Why are you more lucky if you drop the ball on the first roll? That doesn't make sense. |
No, you were just starting from a point that doesn't make sense.
| Quote: | Another example. Let's say you are playing Elves against a bashy team. You decide to force your opponent to blitz his next turn, and then have to dodge away with four linemen. The chance of succeeding with four 2+ rolls is actually rather small (48.23%, a binomial distribution), but this is what it would look like on your luck meter:
- First fails: 0%
- Second fails: 16.67%
- Third fails: 28.57%
- Fourth fails: 37.5%
I don't have any real arguments, it just "feels" weird.
|
So, what is wrong with this example? You have a chance slightly less than 50% to get all right ... and you approach this value. The point is that if you started with a value 0/0 ... which (i repeat myself) is not a number and has no meaning at all.
If you start with 1/2 you get:
- First fails: 12.5%
- Second fails: 23.9%
- Third fails: 32.7%
- Fourth fails: 0.45%
other starting values (even if they result in 50%) will lead to different results.
So, the only weakness the luck-meter has is its initialisation. An idea (and easy to implement) would be:
Find out the average number of dice rolls in a bloodbowl match (say N). The database is large enough to give a decent result. Initialise the luck meter with (N/2)/N. And no, do not truncate, as i have shown above how that influences the luck meter.
Anyway, the luck meter is good, and valid. Due to the initialisation, it can only be used as a relative tool, to compare who has more and who has less luck. |
|
|
khaile
Joined: Nov 08, 2005
|
| Posted:
Dec 15, 2005 - 11:38 |
|
Heh, heard of false modesty?
Of course I know the starting value matters, I was just making an example and used zero as the base. My whole point was that because of the game rules, you only get to fail once per turn. Again, I'm making assumptions and ignoring re-rolls etcetera, but let's look at this example:
You have played 15 turns. Your dice have been statistically correct. In other words, the nominator is exactly twice the size of the denominator, lets say,
10000000 / 20000000 (I'm just choosing something "near infinity").
Now, you want to dodge with all of your elves, ie attempting 11 2+ dice rolls. The thing here is that on the first roll that fails, your turn ends. That means the chance to get each result is
P(roll)=p^(rolls-1)*q
Plus the last option that all rolls succeed (p^11). Ie, the table becomes 1/6 (chance that the first roll fails), 5/6*1/6 (second roll fails), (5/6)^2*(1/6) (third roll fails) and so on.
For each possible result, you can calculate the ending "luck" factor with this formula:
H(roll) = (startingNominator + (roll-1)/p) / (startingDenominator + (roll-1)/p + 1/q)
(It's 1/q since you can only fail a single roll.)
Now, using the rule of complete possible outcome, you can calculate the estimated value with this formula:
E = SUM(H(roll)*P(roll)) + H(allSucceeds)*P(allSucceeds), roll = 1 through 11
If you calculate that you come up with the estimated luck value of 56,73%. It doesn't mean that you successfully dodged with all elves, it just means that, if you plan to dodge with 11 elves, the most probably luck value when your turn ends is 56%.
Ok, let's go back to the starting value. If you do these dodges in your first turn, the most probably luck value afterwards is 45% (given the starting value is 1/2). You may or may not have been lucky, you just have chosen to make a lot of dice rolls.
The luck formula IS correct considering the statistical outcomes of the dice, but it doesn't work when the game rules change how often you roll them.
Another problem with the luck meter is that it considers rolls that doesn't really affect the game much, as well as ignoring rolls that do. For example:
It counts armor rolls as luck. If you have a POW, luck goes up, but if the Goblin (or whatever) passes his armor roll, your luck goes down. The end result may not change your luck that much, but the fact that the goblin went down is a lot more important than he passed his armor roll.
My suggestion doesn't help much to sort out luck considering bouncing balls, crowd throw-ins, kickoff blitzes etc... But perhaps there could be an alternative luck meter that ignores "neutral" results and rolls that have no risk of causing a turn-over.
In the end, it was only a suggestion. I never planned to write such an essay about such a small thing (I don't look at the luck meter when I'm playing), but your post forced me to defend myself  |
|
|
Curro
Joined: Jun 07, 2005
|
| Posted:
Dec 15, 2005 - 11:43 |
|
I´m still thinking that you must stop caring about the luck-meter, please!! |
|
|
Istarios
Joined: Oct 09, 2004
|
| Posted:
Dec 15, 2005 - 11:49 |
|
|
f_alk
Joined: Sep 30, 2005
|
| Posted:
Dec 15, 2005 - 15:25 |
|
| khaile wrote: | | ... My whole point was that because of the game rules, you only get to fail once per turn. |
I agreed on that with you. I also agreed that this will usually lead to luck values higher than 50%. Of course, my whole point was that the luck meter can be used and does make sense for relative measurements
| Quote: | let's look at this example:...
10000000 / 20000000 ...Now, you want to dodge with all of your elves, ie attempting 11 2+ dice rolls....
If you calculate that you come up with the estimated luck value of 56,73%.
|
No, i come up with 50 - 2*10^(-6) %.
| Quote: | | If you do these dodges in your first turn, the most probably luck value afterwards is 45% (given the starting value is 1/2). |
I come up with 40.8 %.
But we agreed already, that this is not the point .
| Quote: | | Another problem with the luck meter is that it considers rolls that doesn't really affect the game much, as well as ignoring rolls that do. |
I think that's ok. What is more important, a GFI that gets you a Touchdown, or the injury roll after the failed GFI and broken armor that sends your player to death?
Is the injury roll for a legend 7 times as important as the injury roll for a lineman with 6 SP? Etc.
So much of that is personal preference (is it more important to kill the enemy or win the game?) ... i am glad it is not in there. Instead we have a measure for the dice statistics. It is a relevant number.
| Quote: | | (I don't look at the luck meter when I'm playing), but your post forced me to defend myself |
... no worries. At least we could scare off a few people.
I do look at the luck meter, but i am aware of what it actually measures.
I think the problem is not the luck meter (which has no preferences on which roll a player considers "important") but the players who think the luck meter somehow accounts for that. |
|
|
Macavity
Joined: Nov 23, 2004
|
| Posted:
Dec 15, 2005 - 15:34 |
|
MEZIR!!!!!!!!!!!!! Can I get your permission to Spam this to death, since it's been gone over 874 times? |
_________________
When I became a man I put away childish things, including the fear of childishness and the desire to be very grown up. -C.S. Lewis |
|
MrMojo
Joined: Apr 17, 2004
|
| Posted:
Dec 15, 2005 - 16:07 |
|
But you already started it! |
_________________
My post count
Jesus loves me this I know, 'cos my Bible tells me so. |
|
Firestata
Joined: Apr 20, 2004
|
| Posted:
Dec 15, 2005 - 16:30 |
|
Ok, noone is going to believe me but I have noticed two things:
1. The meter affect the dice rolls, so if a player has got more "unluck" according to the meter he will roll better
2. The meter believes all ones or skulls to be equally bad which of course is not the case.
This is according to me the real problems. |
|
|
Macavity
Joined: Nov 23, 2004
|
| Posted:
Dec 15, 2005 - 16:34 |
|
Hopefully someone will believe me, since I have noticed two things:
1. Everything that can be said about the Luck Meter has been.
2. Starting a topic on the Luck Meter, including the word, "ANOTHER" is really quite silly.
P.S. I understand that there are people who care about the silly thing. I don't. I don't mind that you do, but read over at least a couple of the last 874 posts on the topic and IF you have something NEW to bring up, why not post in one of those threads? |
_________________
When I became a man I put away childish things, including the fear of childishness and the desire to be very grown up. -C.S. Lewis |
|
Melmoth
Joined: May 05, 2004
|
| Posted:
Jan 01, 2006 - 16:40 |
|
|
|
|
| |
7s for fummbl?
Search
