F2P Mage and range underpowered

[The_Maddest]

The "Snare" or "Entangle" spell.

 

 

 

You want to add a 10-15 second holding spell after people hardly stand a chance against the 5 second bind?

 

Try to think about all forms for F2P PvP before making such suggestions.

[compfreak847]

 

Well, in my own defense, I did explicitly state that it was pure speculation.

 

I also gave other reasons why the combat triangle isn't balanced, including damage per unit time.

 

 

 

When I have free time, I will set up some controlled tests, and post my results, something I haven't seen you do here.

 

Figure that 1k-10k hits with various attack, defense, range, and magic combinations should shed some light on this idea. It should give a small enough standard error for a t-test or a chi-squared test.

 

No, you made up some random numbers about DPT on the spot. I've done a ton of tests like that. Hint: AT LEAST 20 hours of steady combat is required in a specific outfit with exactly the same gear for accurate results with PVP equipment. So your going to have to get higher stats to avoid leveling mid-test.

[sees_all1]

No, you made up some random numbers about DPT on the spot.

 

So what are your correct numbers then, compfreak? None of the numbers I posted were random, and they were all theoretically correct.

 

 

 

I've done a ton of tests like that.

 

Cool beans. Can I see your data and analyze it for myself?

 

Hint: AT LEAST 20 hours of steady combat is required in a specific outfit with exactly the same gear for accurate results with PVP equipment.

 

Don't tell me how to get data. I know what has to be done, how it has to be done, and why it has to be done. I doubt that you've controlled as many variables as I plan on controlling i.e. the opponent.

 

 

 

If melee is as Jagex says it is, then someone with a max hit of 19, an attack level of 99 should hit on someone with 1 defense about 95% of the time. They should also average around 9 or 10 per hit. The more times I hit on someone else, the lower my standard error is, the closer to the true mean.

 

 

 

So your going to have to get higher stats to avoid leveling mid-test.

 

This is where your problem is. Looks like all the tests you've performed were on one account. I don't plan on using my main account for this.

[compfreak847]

 

So what are your correct numbers then, compfreak? None of the numbers I posted were random, and they were all theoretically correct.

 

Exactly that. "Theoretically". Basically you made up some accuracy rates and quoted them as 'statistics'.

 

 

 

 

Cool beans. Can I see your data and analyze it for myself?

 

Nope, lost most of it when I reformatted about a month ago. I'm currently building it back up, but I currently only have the graphs for claws.

 

 

 

 

Don't tell me how to get data. I know what has to be done, how it has to be done, and why it has to be done. I doubt that you've controlled as many variables as I plan on controlling i.e. the opponent.

 

Really. Let me explain my testing scenario, then:

 

 

 

I get a friend with specific stats, in this case testing veracs. I ensure that we both pot at the same time after the 1 minute refresh, then begin running from opposite ends so hits begin at the same time. Each hit is screenshotted to ensure data is correct and for verification and easier counting. Food is eaten at specified HP intervals, and every minute fighting is paused for re-potting should the pots offset the hit rate. Armor and equipment is ALWAYS standardized, that's an obvious given.

 

 

 

 

If melee is as Jagex says it is, then someone with a max hit of 19, an attack level of 99 should hit on someone with 1 defense about 95% of the time. They should also average around 9 or 10 per hit. The more times I hit on someone else, the lower my standard error is, the closer to the true mean.

 

Why are you continuing to make up random statistics? Where did this magical '95%' number come from? Could you point me to the KB page with the hit formulas, or will I be forced to assume that this is yet another one of your trumped up numbers :roll:

 

 

 

 

This is where your problem is. Looks like all the tests you've performed were on one account. I don't plan on using my main account for this.

 

Really. I'm curious as to just what accounts you are using, as a single equipment setup will obtain an average of 800k-1m XP, and leveling throughout the course of the equipment test would completely ruin the numbers.

[sees_all1]

 

 

If melee is as Jagex says it is, then someone with a max hit of 19, an attack level of 99 should hit on someone with 1 defense about 95% of the time. They should also average around 9 or 10 per hit. The more times I hit on someone else, the lower my standard error is, the closer to the true mean.

 

Why are you continuing to make up random statistics? Where did this magical '95%' number come from? Could you point me to the KB page with the hit formulas, or will I be forced to assume that this is yet another one of your trumped up numbers :roll:

 

 

 

 

[hide=from postbag 38 - chicken, jad?]The Maths

 

 

 

First, some interesting numbers:

 

 

 

A rooster has 7 Hitpoints. They have a 0.77% chance of hitting TzTok-Jad, and then only a 50% chance of inflicting a measly 1 Hitpoint in damage. Against an Yt-HurKot they are much more successful, and have a 5.48% chance of hitting, followed by the same 50% chance of inflicting 1 Hitpoint of pain.

 

 

 

TzTok-Jad has 250 Hitpoints, and will ordinarily regain 1 lost Hitpoint every minute. When he has 150 Hitpoints left, four Yt-HurKot will spawn to heal him. As a godly creature of magma-fuelled destruction, TzTok-Jad has an average chance (across all attack methods) to hit a rooster of 99.78%, and has an average maximum hit of 96.5 damage.

 

 

 

Yt-HurKot have 60 Hitpoints, and will heal 1 Hitpoint per minute. They have a 98.85% chance of hitting a rooster, and can inflict up to 11 Hitpoints damage.

 

 

 

Right, on with the amusing theories

 

 

 

1. Each rooster has a 0.385% chance of wounding TzTok-Jad, so you need 259.74 roosters to inflict one Hitpoint of damage. As TzTok-Jad has 250 Hitpoints, it will therefore require 64,935 roosters to defeat TzTok-Jad in a single instant of frenzied pecking.

 

 

 

2. Because we can still assume that 64,935 pecks are required, and that 24 occur each attack cycle (which, for a rooster, is approximately every 2.4 seconds), we can know that it will take 6,493.5 seconds, or 1 hour 48 minutes 13.5 seconds. Not bad, for a pack of feathered wimps, Im sure youll agree! The roosters, however, are constantly being attacked by TzTok-Jad, who has a 99.78% chance of hitting a rooster in melee, and a 92.75% chance of killing them outright with each hit, so he has a 92.55% chance of killing a rooster with every attack. In the time he survives, TzTok-Jad would be able to slay 2,509 roosters, so you would then only need 2,533 roosters (because you still need 24 roosters around Jad at the time of his demise).

 

 

 

3. Roosters are foolish creatures and unlikely to think of attacking the Yt-HurKot as they arrive to heal their master, so it would then become effectively impossible.

 

 

 

If, however, the roosters were elevated intellectually to the level at which this would occur to them, they could easily continue to attack TzTok-Jad while the remaining thousands of feathered beasts deal with the irritating little blighters. This will, of course, take much longer and require far more birds.

 

 

 

TzTok-Jad heals himself at a rate of 1 every minute, while the 24 roosters are inflicting 1 point of damage every 25.97 seconds on average, so the roosters instead inflict 1.31 Hitpoints of damage every minute (TzTok-Jad having healed one Hitpoint during that time). This means that it will take them 1 hour 16 minutes 20.4 seconds to get TzTok-Jad to 150 health, thus spawning the wretched Yt-HurKot, and a further 1 hour 54 minutes 30 seconds to finish him off if the other roosters are capable of occupying them. TzTok-Jad, therefore, will be able to slay 4,593.65 roosters in the 3 hours 10 minutes 50.4 seconds the battle lasts.

 

 

 

The Yt-HurKot, meanwhile, will no doubt be finishing a few off themselves. The chickens have a 2.74% chance of doing 1 point of damage to an Yt-HurKot every 2.4 seconds, but 8 can attack each of them, so will do 0.21 Hitpoints damage every time they attack. At this rate, it will take them 12 minutes 45.6 seconds to kill an Yt-HurKot (including the time calculated for the Yt-HurKots normal regeneration). In that time, however, their prey will finish off 172.43 roosters. As Yt-HurKot only respawn regularly if TzTok-Jad gets back to full health, the birds will only have to face 4 of them and suffer a loss of 689.72 of their number.

 

 

 

In total, this means that you will need 5,308 roosters to make it through the previous 63 waves in order to have good odds of winning a fire cape. Of course, roosters have no need of capes, so theyll probably just ask for some TokKul so they can save up for an onyx to craft into a ring of stone, in the hope that at least one of their hens will survive the brutal culling they face every day at the hands of eager fletchers and adventurers completing Cooks Assistant.[/hide]

 

From postbag 38.

 

There is a number generated to determine if there is a hit or not, and there is a number generated to determine how high the hit is.

 

Assume the person will hit 100% of the time. Then, if the numbers are evenly distributed, and a max hit of 19, 19/20 of the hits will hit 1-19, and 1/20 will hit a 0. I'm not pulling numbers out of my behind, and right now I'm starting to doubt your credibility.

 

 

 

 

This is where your problem is. Looks like all the tests you've performed were on one account. I don't plan on using my main account for this.

 

Really. I'm curious as to just what accounts you are using, as a single equipment setup will obtain an average of 800k-1m XP, and leveling throughout the course of the equipment test would completely ruin the numbers.

 

 

 

I'll post my experiments and my findings when I'm completely through with them, thank you very much. Just note that most statistical analysis don't require tens of thousands of data points, 2000 should be more than sufficient for each set up. Also, Chi-squared tests require about 10 data points for each bin, and if there's only 10 bins (0-9), 100 data points would suffice. But if you don't know statistical analysis, I'm not going to explain the tests here.

 

 

 

Now, there are questions that I'm still looking to answer, and you haven't provided them.

 

-How is hits/splashes/damage determined for range, and magic?

 

-What is the correct ratio for defense level and defense bonus?

 

-Does magic level affect magic defense?

 

 

 

I'm going to be extremely meticulous in my approach, to not mess anything up, and I'll make sure to post all my data and my conclusions. The only thing you've done so far is to post your conclusions. Next time, back it up (your drive, literally. those thousands of hours lost must've stunk).

[compfreak847]

I usually do back things up, but this was my own fault. Was doing my usual reformatting, and I forgot to back up my RS folder. Hundreds of hours and thousands of screenshots gone, including all my old videos and screenshots from 4+ years back with everything from hilts to level ups to long-gone friends. It stunk, badly.

 

 

 

Answers to your questions:

 

 

 

-Basically, it works like this. With a certain armor and weapon combonation, each melee\range\mage attack has a certan chance of hitting. A random number is generated to determine this - if the number indicates that the hit fails, melee\range hits zero and magic splashes. If it is determined that the attack hits, a number is generated between 0 and your max hit (inclusive) for you to hit

 

 

 

-There is no apparent ratio. It has far too many factors and an element of semi-randomness to identify any variable relating to bonuses

 

 

 

-Yes, this is answered on the KB page

 

 

 

Like I said, once you achieve 10+ hours of testing you'll realize what I'm talking about - there are undeniable patterns and 'hit streaks' that make testing require far more time - in the end, the results are similar to your average hit, but the hit spread will look significantly different.

 

 

 

I still don't understand how your going to do any testing with less then maxed stats. At maxed stats I simply sara\zammy brew down to the stats I want to test, but at your level your going to be leveling throughout the tests, leading to inaccurate results.

[sees_all1]

Like I said, once you achieve 10+ hours of testing you'll realize what I'm talking about - there are undeniable patterns and 'hit streaks' that make testing require far more time - in the end, the results are similar to your average hit, but the hit spread will look significantly different.

 

 

 

I still don't understand how your going to do any testing with less then maxed stats. At maxed stats I simply sara\zammy brew down to the stats I want to test, but at your level your going to be leveling throughout the tests, leading to inaccurate results.

 

 

 

I'm going to identify one thing to test at a time, and I'm going to grind it out. I'm going to create a bunch of throwaway accounts to do this, so leveling isn't a problem. With the types of analysis I plan on doing, 1000's of hits provide .1% standard error, 100's of hits provide .5% SE. I don't need to test an account with 80 attack, 90 attack, 99 attack, if I can find the trend at 10, 20, 30...

 

 

 

Don't worry about how I plan on doing this, because I probably won't finish for another 6-10 months. It will be fairly scientific.

[compfreak847]

Trends will be MUCH easier to identify at high levels, and indeed your going to require high hits for many of them - with a hit range of, say, 1-9 your going to notice far less then 1-30 or 1-80. Your missing the SE part - the hits have a non-random component to them, and they aren't simply standard linear numbers generated randomly. They have a very odd cluster towards the ends and center, and avoidance of certain numbers that I have yet to completely unravel. You can't just assume a few thousand hits will give you highly accurate results.

 

 

 

And how you plan on doing it is also important - if you charge into it without realizing all the factors involved, you'll end up with very inaccurate results that will ruin your data. Your correct in that it will take many months; I spent almost a year on mine at close to 2 hours per day.

[sees_all1]

Trends will be MUCH easier to identify at high levels, and indeed your going to require high hits for many of them - with a hit range of, say, 1-9 your going to notice far less then 1-30 or 1-80. Your missing the SE part - the hits have a non-random component to them, and they aren't simply standard linear numbers generated randomly. They have a very odd cluster towards the ends and center, and avoidance of certain numbers that I have yet to completely unravel. You can't just assume a few thousand hits will give you highly accurate results.

 

 

 

And how you plan on doing it is also important - if you charge into it without realizing all the factors involved, you'll end up with very inaccurate results that will ruin your data. Your correct in that it will take many months; I spent almost a year on mine at close to 2 hours per day.

 

 

 

A few thousand hits is all I need to test something with stats, and if my hit range is from 0-9, I'll need fewer about 10x fewer hits than if its 0-99.

 

Once I'm done, I'll post my method, my data, my analysis, and my conclusions. Then you can scrutinize it all you want. Until then, I've got calc III homework due at midnight.

[compfreak847]

You aren't getting my point. I'm saying there's subtle patterns that your never going to recognize without a very large sample size, and the higher your hits the finer resolution you'll get on a graph. Besides, if your going to do it at a low level, you'll literally get in 100 hits and have to start training another account to the exact same stats, all the while having a friend keep up with you.

[sees_all1]

You aren't getting my point. I'm saying there's subtle patterns that your never going to recognize without a very large sample size, and the higher your hits the finer resolution you'll get on a graph. Besides, if your going to do it at a low level, you'll literally get in 100 hits and have to start training another account to the exact same stats, all the while having a friend keep up with you.

 

 

 

With what Jagex posted in that postbag, and how I believe hits are calculated, its very simple to set up an experiment, and be 99.7% sure of your results. If you don't know what you're doing, you're not going to prove anything to yourself, nor anyone else. If your method is repeatable, the data set large enough, and you use the right tests, no one can argue with the conclusions, espescially if you're 99.7% certain from the tests.

 

 

 

Read up about the Chi-squared Goodness of Fit tests, and you can understand how I'm going to test each one.

 

 

 

For example:

 

Hypothesis: Someone with maxed attack and a max hit of 10 should hit an even spread.

 

Make assumptions: Someone with 99 attack, +51 in attack bonuses should hit 149 times to every time they splash once, when hitting on someone with 1 defense.

 

Of the 149 times they "hit", they should hit from 0-max, and those hits should be distributed evenly, with 0 hitting slightly more because of the 1/150 thing.

 

Now, say their max is 9. They could hit 0,1,2,3,4,5,6,7,8,9. If you hit 1000 times, you'd expect about 100 hits in each, with a couple more in the zero column.

 

Then you'd do the Goodness of Fit test, to see if the actual data conforms to what you expected. If it doesn't conform, then you could repeat, and gather new data. Or, you could try another hypothesis, that there's "trends" or something along that nature.

 

 

 

There are many variables to play with, some easier than others. With many simple hypotheses, and many different tests, an idea behind the actual random numbers driving the combat triangle can be determined, and then accurate statements can be made. Whatever is going on inside the runescape servers isn't magic - there are methods constructed from more methods from the Java API. There are numbers that Jagex displays. None of us on the outside know exactly how they are used, but we can figure it out.

[compfreak847]

Your not getting what I'm saying. I'm not suggesting that hits have more then a simple random calculator behind them - I'm telling you. For example, certain hits have a statistically significant lower hit rate then others - 69 and 67 are examples of that, and 66 is one of the more common hits. Your going to need much more then the numbers equivalent to a simple random generator for accurate results.

[sees_all1]

Your not getting what I'm saying. I'm not suggesting that hits have more then a simple random calculator behind them - I'm telling you. For example, certain hits have a statistically significant lower hit rate then others - 69 and 67 are examples of that, and 66 is one of the more common hits. Your going to need much more then the numbers equivalent to a simple random generator for accurate results.

 

I hear what you're saying, but remember these numbers are pseudo-random, and for all intensive purposes it is random.

 

If what you're saying is correct, then some number, like 8, would hit more often than 9.

 

 

 

I can test that. I can get another set of data, and test it again. I can repeat this 100 times. If its statistically significant, and it remains statistically significant for 4 sets of data, then the probability that there is a fluke is very small. You don't have to explain that to me, I know stats.

 

 

 

I don't know what setups you were using, or your opponent's setups. I can, however, setup my own people to get my own data, explain to you exactly what I did. If you feel so inclined, you can replicate it.

 

 

 

As of right now, I can't replicate what you're telling me, so to me, its void. What I want to do is to do my own analysis of this, so next time I argue with you, I have statistics and data to back it up, and I can show you exactly how I got that data - and you can get your own data using the same method, and you can test your data to mine, so you can tell I'm not pulling numbers out of my behind.

 

 

 

Also, it serves a purpose to the RS community as a whole. Next time they think, gee, should I raise my Strength level, or my defense level, they can look at the data, and the analysis, and judge for themselves.

[compfreak847]

I'm saying that the effect is subtle enough you won't notice it in 100 or even 1,000 samples. I'm making sure you don't just do 1,000 hits and then call it good. Data collected is ONLY good for the levels\equipment\situations where it is tested, so this isn't exactly gods gift to RS - it might be handy for the people who are 46 strength wearing full rune with a square shield, strength ammy, and fighter boots praying +10% strength vs. an opponent with full d hide, 21 defence, 34 HP, praying +10% defence and casting water blast.

 

 

 

Exaggeration, but you get my ponit. You can't pull out a few numbers from various equipments and stats and extrapolate that data to cover the entire range of armors and stats. The number of variables, and indeed, the entire random number generation system is far more complex then you realize.

 

 

 

Also, your signature is wrong... Melee maxes 10, and mage maxes 16. Your overdoing both of their hits :P

[sees_all1]

I'm saying that the effect is subtle enough you won't notice it in 100 or even 1,000 samples. I'm making sure you don't just do 1,000 hits and then call it good. Data collected is ONLY good for the levels\equipment\situations where it is tested, so this isn't exactly gods gift to RS - it might be handy for the people who are 46 strength wearing full rune with a square shield, strength ammy, and fighter boots praying +10% strength vs. an opponent with full d hide, 21 defence, 34 HP, praying +10% defence and casting water blast.

 

How many methods do you think Jagex programmed into Runescape to determine whether there is a hit or not, and how hard should it hit?

 

I'm guessing about 3, maybe 4 for standard weapons (I'm not counting specials because they do some crazy things).

 

Tip.it's combat calculator is (from my point of view) correct. I haven't had an instance where it's prediction was inaccurate.

 

How did Tip.it get its calculator? Tons of data. Tons and tons of data.

 

 

 

Now, If I carefully go about this the correct way, and I figure out what their method for determining whether there is a hit or not, and the amount that is hit, then that probability formula could be applied to all sorts of situations, including estimated number of hits per level, estimated victor in a 1v1 battle, etc. All these things would need basic assumptions, and it is those assumptions that need to be checked.

 

 

 

 

 

Exaggeration, but you get my ponit. You can't pull out a few numbers from various equipments and stats and extrapolate that data to cover the entire range of armors and stats. The number of variables, and indeed, the entire random number generation system is far more complex then you realize.

 

I've told you now quite a few times: I'm familiar with statistics. I'm not going to make some outrageous claims based on the data I collect, because I know better than that. And once I post my procedures, my data and my findings, you can scrutinize it all you want.

 

 

 

Also, your signature is wrong... Melee maxes 10, and mage maxes 16. Your overdoing both of their hits :P

 

Lol. Those are two level 90 F2Pers going at it. And if you've read the suggestion, you'd understand whats happening. Funny, I haven't seen a response from you yet about it. I figured you would be all over my suggestion, how out of whack the spells are, etc. :P

[compfreak847]

 

How many methods do you think Jagex programmed into Runescape to determine whether there is a hit or not, and how hard should it hit?

 

I'm guessing about 3, maybe 4 for standard weapons (I'm not counting specials because they do some crazy things).

 

Tip.it's combat calculator is (from my point of view) correct. I haven't had an instance where it's prediction was inaccurate.

 

How did Tip.it get its calculator? Tons of data. Tons and tons of data.

 

What on EARTH are you talking about? I see nothing even remotely related to hit accuracy or hit spreads. The closest thing I see is a max hit calculator, which has nothing whatsoever to do with the hit spread and is far, FAR simpler to test.

 

 

 

 

Now, If I carefully go about this the correct way, and I figure out what their method for determining whether there is a hit or not, and the amount that is hit, then that probability formula could be applied to all sorts of situations, including estimated number of hits per level, estimated victor in a 1v1 battle, etc. All these things would need basic assumptions, and it is those assumptions that need to be checked.

 

There's just plain too many variables. I can name literally dozens. To put the number of combonations into perspective, just try 99^3 - combinations for melee skills alone (970,299 different stat arrangements). Now factor in the literally hundreds of armor, weapon, potion, food, and prayer costs, (99^10) and square that number to factor in your opponent. There's no cut and dried formula like your insinuating.

 

 

 

 

I've told you now quite a few times: I'm familiar with statistics. I'm not going to make some outrageous claims based on the data I collect, because I know better than that. And once I post my procedures, my data and my findings, you can scrutinize it all you want.

 

Look, you can drop the whole statistics class thing. I wasn't too shabby in math back in my collage days, and I picked up a thing or two here and there. This isn't complex statistics we're arguing. If your going to do such overly simplistic "testing" as your suggesting, your only going to end up with completely falsified numbers that do nothing but muddy the waters and confuse the issue with incorrect data.

 

 

 

 

Lol. Those are two level 90 F2Pers going at it. And if you've read the suggestion, you'd understand whats happening. Funny, I haven't seen a response from you yet about it. I figured you would be all over my suggestion, how out of whack the spells are, etc. :P

 

I had thought your mage suggestion thread was the only one. Sorry, but your new spell book thing is a joke - I'm just curious how you justify FTP getting a spellbook far more powerful then member's current, massively overpowered spellbook? I'll put it this way:

 

 

 

Currently, members magic is insanely overpowered and able to crush any combat style at any level

 

Your proposed "fix" for FTP mage: Give them a spell book that makes member's current book look like a joke

 

 

 

Yeah, I don't really have much to say about that besides 'try members mage' :|

[sees_all1]

 

How many methods do you think Jagex programmed into Runescape to determine whether there is a hit or not, and how hard should it hit?

 

I'm guessing about 3, maybe 4 for standard weapons (I'm not counting specials because they do some crazy things).

 

Tip.it's combat calculator is (from my point of view) correct. I haven't had an instance where it's prediction was inaccurate.

 

How did Tip.it get its calculator? Tons of data. Tons and tons of data.

 

What on EARTH are you talking about? I see nothing even remotely related to hit accuracy or hit spreads. The closest thing I see is a max hit calculator, which has nothing whatsoever to do with the hit spread and is far, FAR simpler to test.

 

 

 

Runescape is programmed in Java. Runescape, at its very core is composed of objects and methods. There are a few ways to generate a random number in Java, using the Math class, the Random class, etc. There might also be a couple ways to generate random numbers that aren't in a even distribution.

 

 

 

The purpose of programming in objects is to reuse code - there are 3 different classes. I doubt Jagex overloaded the calculate_hit method (or whatever they called it) more than 4 times. Whether you hit or not isn't magic - it has something to do with levels, bonuses, and your opponent's defense and bonuses.

 

 

 

If the numbers generated in Runescape are random (which for this they are), then with enough simple analysis you'll be able to determine the underlying method for determining a hit, and how much of a hit.

 

 

 

But I feel that explaining all this to you is a waste of time because you know it already. What you tested was "In this armour setup, against this armour setup, what was my average damage per hit ratio" What I want to know is "With X attack, Y strength versus Z defense and n bonuses, what is the probability of a hit, and therefore my average damage per hit"

[compfreak847]

I'm telling you it isn't a simple random number generator, and different variables have different effects. I'll show you one result I have from my hit spreads, Dragon Claws:

 

 

 

D claws have a very odd hit spread. They appear to accentuate the 'hit one hit often, hit another hit very rarely' theory. For examples, 34-17-8-9 is one of the most common of all hits, while 36-18-8-9 is many times rarer. A similar effect is observed with 0-22-0-21. Now, the interesting fact is that these (undeniably) not random hits are affected in dramatically different ways by different variables. For example, if I drop my strength level by 10, my 34-17-8-9 drops, becoming 64 or 63. 22-21 becomes 20-19 etc. etc. Makes sense, right? How about this: When I turn prayer off, I STILL get the same hit patterns - I'm just as likely to hit 22-21 with piety on as off, even though if my level goes down my hits drop accordingly. To compound the problem is armor bonuses - they seem to have next to no set pattern, and, like piety, little effect on the final hit outcome.

 

 

 

I haven't finished testing on claws (mostly thanks to the 4 different types of hits, it's like testing 4 individual weapons randomly chosen :lol: ) but when I do I should be able to delve into percentages. For now, however, it is an incontrovertible pattern.

 

 

 

I'm just using this as an example for you. Your still under the impression that hits are simply a random number, and that is not the case. Therefore, if you do a thousand tests you might come up with mildly accurate results for the average hits - but without deep analysis of hit patterns, it's completely worthless for all other setups.

[sees_all1]

You're testing the DClaw special then?

 

Special Attacks for weapons will use an overwritten method to calculate a hit, pure and simple.

 

I won't be testing special attacks, so I won't have that as a confounding variable.

 

 

 

You can test special attacks until you're blue in the face, but unless you try to understand whats going on underneath it all, you won't understand the mechanics of the game, and therefore will draw wrong conclusions.

[compfreak847]

That's not at ALL what I meant. I'm saying that hits aren't the simple randomly generated numbers you think they are - they have exhibit patterns not related to a random data set. I'm afraid that your the one without a clue of what your talking about; you haven't tested a single weapon yet, and I was using dragon claws as an example that I currently have more data on. Normal AGS hits exhibit the same characteristics. Until you see the numbers and realize what I'm talking about, you'll end up with a laughably simplistic and inaccurate representation of hit generation. I suggest you stop trying to play it off as regular hits until you actually do some real testing, not numbers you've made up on the spot :roll:

[sees_all1]

[compfreak847]

After all your talk about knowing statistics, you should know enough to realize that it's obvious when a number is being influenced by an outside variable with a large sample size. By your logic, one could also say levels and equipment bonuses are fake and don't do anything to your hit - your just getting unlucky and hitting low when you don't wield a weapon.

[John Halo Chief]

viewtopic.php?f=37&t=784491

[Sir_Kurity]

Well, i have 57 magic, and i cant be bothered training it any higher, so i guess someone can verse me now, im level 34 mage.

 

 

 

Add sir kurity if you wan't to duel me.

[Veritas94]

"We recognize that mage may be underpowered and there are people with rather large cerebellums on the problem." - Mod MMG

 

 

 

Runescape forums, quick find code: 13-14-613-58401007

 

 

 

HELL YES!!!!!!!!!!!!!!!!!!!