Page 1 of 2

Weapons Research

PostPosted: Thu Dec 06, 2007 11:30 am
by Evolution Prime
I am looking for volunteers for weapon experiments. I would like a broad range of players for the, from the higher levels all the way down to the lower levels. This way we can test as many stat configurations. with these weapons. We can divide up the list of weapons based on the amount of stats each has. I am sure the Decepticons and others will be having their experiments going, and I beleive we should as well.

Here is my plan. We divide up the weapons based on the volunteers experience points. The higher your level, the more experience you can experiment with for configurations. The weapon effectiveness needs to be judged how it varies with the stats like strength and speed, along with the stats and armor of the enemy being hit.

Here is a list of things I want to be looked at:

Damage Range, How often it misses, Recharge time, How it is affected by speed, strength, firepower

If you can think of anything else that may be useful, add it in.

What I am looking for is just a brief write up on each weapon with the above list. I can then compile the list and make it one big write up and post it here on the boards as I get things in.

Here is what I can offer you. I have about 11.5 million energon in my account that I am doing nothing with. I will reimburse any volunteer 50% of the weapon cost and 50% of the armour that is put back on when reset, until either my bank is run dry or the reset happens. CR Chamber costs are on you to come up with the energon for that. Just send me a PM with the weapon being tested, the bot being tested on, and the armour cost.

PostPosted: Thu Dec 06, 2007 11:33 am
by Dragonslayer
I'm volunteering. Click my sig to see my team.

PostPosted: Thu Dec 06, 2007 11:37 am
by Evolution Prime
Dragonslayer wrote:I'm volunteering. Click my sig to see my team.


Here is what I would like you to start out on:

Metal-slashing Claw
Diffraction Sword
Vibro-Axe
Sonic Sword
Armor Axe

PostPosted: Thu Dec 06, 2007 11:38 am
by jazzrules

PostPosted: Thu Dec 06, 2007 11:46 am
by Evolution Prime
*EDIT* jazzrules has dropped out of weapons research due to real life issues. I thank him for volunteering.

PostPosted: Thu Dec 06, 2007 12:47 pm
by Scantron
Count me in, EP! Here's my squad:

http://www.seibertron.com/heavymetalwar ... hp?id=4086

If it helps, I currently have a Dual-Bladed Sword [0/8], Metallikato Blade (0/9) and Saber Blade (0/10) in my inventory; I've used the D-BS once (similar results to the depowered BBs). I also have multiple Gatling Guns (10/3), equipped across characters with a range of STR and SPD, and can access the Rocket-Propelled Mortar Shells (10/7) without having to reset a character. I'm generally going up against opponents with maxed-out armor and high or maxed-out endurance. I also have all useful stats (except SKL and CRG) maxed out on two of my characters and could test weapons at their maximum potential.

PostPosted: Thu Dec 06, 2007 1:32 pm
by Evolution Prime
Thank you Scantron. Since you are at a high level, I will assign the higher end weapons to you. I will let you start with the weapons that you have currently and then let you pick the rest. Unless another higher level comes along, I will let you work with whatever high level weapon you choose.

PostPosted: Thu Dec 06, 2007 3:07 pm
by Wingspan
I'm in. Team link in the sig.

I'll also volunteer to compile data; I'd want raw data though. This will let every volunteer focus on collection of data while I coordinate the appropriate analysis. Enough data will yield stastistics we can work with and new raw data can be appended and new results posted. For raw data, I would like the following format provided per attack per mission / arena bout.

Suggested Raw Data Format wrote:[Wep Name], [A-SPD], [A-STR], [Hit(1/0)], [Attack Time], [Avoided(1/0)], [Damage %], [D-SPD], [D-RNK], [D-END], [D-Armor]
.....


"A-" means attacker, "D-" means defender. The comma seperated values can easily be transfered to Excel and I'll build statistics for Min / Max / Average Damage given the Attacker and Defender stats. The 1/0 (true/false, for non-computer types) for Hits will let me calculate hit percentage for weapons given stats. Duplicating data (A/D-stats, weapon name) across lines will help me write more generic queries so I can do overall comparisons of hits, misses, armor effectiveness, STR effectiveness, etc.

I know collecting Defender stats will be more difficult, but it will be necessary to ensure the numbers are accurate and consider some normalizing of the data so we understand true raw damage and hit %.

One suggestion, we can also track other Faction teams / transformers that have weapons we're interested in by reviewing their histories or merely reviewing the battles in which we fight. Tracking enemies attacking us will also give us specific metrics for armor viability, so those with armor may (should?) be the ones to receive this task.

EP, provide your guidance on if the above is acceptable. I can also provide an XLS or some other template that Autbots can fill out and submit back for summaries. Given how long it took me to summarize one mission, that might be useful. For all, here's an example taken from the following mission given what I was discussing above.
http://seibertron.com/heavymetalwar/view_mission.php?mission_id=8734

Bombadeer using Electric Cannon wrote:Electric Cannon, 5, 0, 1, 00:51, 0, 25, 0, 0, 0, 0, None
Electric Cannon, 5, 0, 1, 03:48, 0, 30, 0, 4, 0, 0, Unknown
Electric Cannon, 5, 0, 1, 05:04, 0, Unknown, 0, 0, 0, 0, None
Electric Cannon, 5, 0, 1, 05:44, 0, Unknown, 0, 0, 0, 0, None
Electric Cannon, 5, 0, 1, 07:35, 1, 0, 0, 4, 1, 0, Unknown


Given the complexity of compiling the raw data, I would recommend a template be provided to simplify collection.

Advise, and I will respond appropriately.


Note: XLS is proprietary to Microsoft, but Open Office works wonders.

Note 2: I'm not interestd in CRG because it's too complex to try and count that in given tactic attacks, melee attacks, and everything else. Raw weapon data should suffice.

PostPosted: Thu Dec 06, 2007 3:37 pm
by ChocolateThunder
Wingspan, It may take me a while but do you want data PM'd, emailed or posted here?

PostPosted: Thu Dec 06, 2007 3:44 pm
by Wingspan
ChocolateThunder wrote:Wingspan, It may take me a while but do you want data PM'd, emailed or posted here?

Until EP says otherwise post here and put each Bot in each battle in a different quote block (like I did above with Bombadeer; if tracking more than one Bot in one mission, give each it's own quote block so I can do the math right for time elapsed between attack). It will keep things clean and make clear what's a data posting and what's conversation.

Edited to clarify quote block usage.

PostPosted: Thu Dec 06, 2007 5:12 pm
by Wingspan
Mkall wrote:Also those that rely on brawlers may notice a small change in damage output :WHISTLE:

I recommend we track brawlers as well. Perhaps brawlers can be a combination test for (1) STR damage output and (2) RNK/Armor damage reduction.

PostPosted: Thu Dec 06, 2007 6:35 pm
by Scantron
If EP gives the thumbs-up, I'd definitely like a copy of your XLS template, Wingspan. I've been trying to figure out how to store and compile data and that would be perfect. A couple of points I'd want to bring up, though:

- Should we count instances where a character uses punch/kick instead of their weapon (whether they connect, miss or have a punch/kick avoided)? For instance, let's say I have a Metallikato Blade equipped and use it at 00:05 and 02:08...but I punch/kick at 00:58 and 01:40. That might allow for some info about recharge time.
- For opponents' armor, should we just assume they have the best possible armor for their configuration? For instance, assuming a character with 10 STR/4 RNK will have 10 tons of Crystalline. I think this is generally a valid assumption to make, but wanted to bring it up and get opinions.
- This isn't an issue in the Arena, but iron-botting sometimes occurs in the Missions and could throw off results. It's pretty easy to detect, particularly with weapons that cause less damage, but I'm thinking that we should keep an eye out for iron-botting when recording data and note it in our results if we see it.

PostPosted: Thu Dec 06, 2007 10:32 pm
by Wingspan
Scantron wrote:... iron-botting ...

Say but? I genuinely do not unserstand the phrase (even though it was used in a complete sentence; blasted public school!). As for melee, that could be very beneficial. As for armor, I was going to handle both the known and unknown armor statistics. I'm thinking there are some armor nuiances that exist (or will be created) which we'll need to capture; that means we'll have to risk our own hides by trying some out and running the numbers. But let's not focus on that until we get the weapons rolling.

For the template once complete, I will provide. Is email on your profile?

I'm also presently reviewing the HTML quality of the webpages to see if I can do something a little slicker with C# / .NET. Will notify, but I *might* be able to have individuals deliver ZIP files of the HTML pages (for mission and for each participating transformer) and I'll gather data on every action. That's, naturally, a little more complex.

Why do I feel like Perceptor? I liked Blaster better...
:P

PostPosted: Thu Dec 06, 2007 10:45 pm
by Dragonslayer
Hey, listen. It's looking like I might not have my research findings done for a while. These next few days I'll be studying for my semester finals, and I'm trying hard not to get so distracted, so I'm laying off of HMW until about Wednesday. But rest assured, as soon as I get a chance, I shall return.

PostPosted: Thu Dec 06, 2007 11:17 pm
by Scantron
My e-mail address should be visible on my profile now; forgot I turned that off awhile back. Thanks for the template!

I've started compiling data on the Gatling Gun, using my own Excel sheet for now and just had a couple of other things come up:

- If you're going to be analyzing the data, Wingspan, approximately how many "data points" (weapon uses) would you be looking for with each weapon?
- Due to the difficulty of calculating overkill when stasis-locking an opponent, I've simply been noting a weapon's damage as "Unknown" when a shot puts the target into stasis lock, lest it throw off the calculations any.

Wingspan wrote:
Scantron wrote:... iron-botting ...

Say but? I genuinely do not unserstand the phrase (even though it was used in a complete sentence; blasted public school!).


Sorry 'bout that...

Iron-Botting (v) - Sending a character into a mission with pre-existing damage; sending a character into another mission without healing the damage said character obtained in their previous battle. IIRC, the term is derived from the Iron Man Triathalon, where atheletes compete in grueling competitions without much break. It is impossible to go iron-botting in the Arena, as the Arena blocks characters with <100% health from fighting. Iron-botting complicates weapons testing because if a character enters a mission at 80% health and you hit them with a weapon that does 5% damage, it'll look like they took a 25% hit right off the bat.

This used to be more of an issue before the Automatic CR option was added for the Missions; back in the day, Arena was Auto-CR only, Missions Manual-CR only. It's been a long time since I've heard much of it, but just wanted to put it out there as something to be aware of, in case someone notices a weird damage amount popping up for a particular weapon.

PostPosted: Fri Dec 07, 2007 12:17 am
by Wingspan
Outliers will be a concern. But hopefully with enough data we can entirely scrap the entire posibility of outliers by filtering them once a good average is known.

Data points is a good question. The ideal scenario is testing each weapon with each combination of FP / STR against opponents who have no ARMOR and who have END. Think the 'Cons will go for it? Anyway... Enough is certainly an interesting question. I'm unsure, but more is always better. In the long run, we'll have plenty to do necessary comparisons. For now, I'd want at least 20-50 attacks per weapon.

As another update, I'm having some success using regular expressions to parse mission / arena logs. What this could mean is rather than individuals manually parsing through the logs, they could save the HTML files and deliver to me (along with an HTML file per character that was participating in the battle). Once I have all the scenarios down (potential attacks, etc.), parsing out the character stats along with mission performance should be easy.

I'm going to focus on this, since the long run benefits are huge. Hopefully the template won't take too long and I can get that together tomorrow.

It will, at first, be just for data collection. Once data is a submitted, I'll start playing with Excel Pivot Tables which will really make things interesting.

PostPosted: Fri Dec 07, 2007 8:35 am
by Evolution Prime
I really like that idea Wingspan. I give it the big thumbs up. Go forward with your plan.

PostPosted: Fri Dec 07, 2007 10:29 am
by Wingspan
Evolution Prime wrote:I really like that idea Wingspan. I give it the big thumbs up. Go forward with your plan.

I've got almost all the regular expressions I need. They pull all a character's information (stats, armor, tactics, etc.) from their HTML page and all the actions from a mission (save handling misses and avoids). Haven't tested arena yet.

What this means is that each volunteer will need to take the 2 minutes a mission to save (1) the mission log as plain html and (2) the character pages of each character involved as plain html to then (3) ZIP and email to me. This will pull all the raw data needed and, for our Bots, it will even pull our armor and tactics for more detailed analysis over time.

Scantron, because of this success thus far I'm not going to be generating a template.

I'll provide details soon of how I would expect things delivered. For now to all, just save the plain HTML pages locally for the mission and characters grouped by mission (e.g. put them all in a folder labeled with the mission number).

PostPosted: Fri Dec 07, 2007 10:37 am
by Evolution Prime
Everything sounds good. I like the way you set it up Wingspan. It makes things a little easier to analyze. It will cutdown on the time it will take to put things together. Now we just need some more volunteers to get things going. I think we need another high level and three middle levels

High Levels (7-10) => Scantron
Middle levels (3-6) =>
Lower Levels (0-2) => Dragonslayer, Wingspan

PostPosted: Fri Dec 07, 2007 10:43 am
by Dragonslayer
You forgot jazzrules there. He's doing all the basic weapons.

PostPosted: Fri Dec 07, 2007 10:47 am
by Evolution Prime
Dragonslayer wrote:You forgot jazzrules there. He's doing all the basic weapons.


He dropped out due to personal reasons. I edited my post that I assigned him weapons. I don't know if Chocolate Thunder is volunteering or not.

If a level 0 or 1 'Bot wants to volunteer to just do the basic weapons, I am more than happy to pay for the all the costs for that volunteer.

PostPosted: Fri Dec 07, 2007 10:48 am
by Wingspan
Evolution Prime wrote:Everything sounds good. I like the way you set it up Wingspan. It makes things a little easier to analyze. It will cutdown on the time it will take to put things together. Now we just need some more volunteers to get things going. I think we need another high level and three middle levels

High Levels (7-10) => Scantron
Middle levels (3-6) =>
Lower Levels (0-2) => Dragonslayer, Wingspan

To get raw weapon data too, we can also scavenge from Decepticons, Maximals, and Predacons. Stalk their teams, download their result files, and submit as well. In general though, the benefits of known our own tactics and armor will be far more beneficial and scavenging should be a last resort, but if in a week we're still short it's something to consider.

PostPosted: Fri Dec 07, 2007 11:59 am
by Evolution Prime
Wingspan wrote:
Evolution Prime wrote:Everything sounds good. I like the way you set it up Wingspan. It makes things a little easier to analyze. It will cutdown on the time it will take to put things together. Now we just need some more volunteers to get things going. I think we need another high level and three middle levels

High Levels (7-10) => Scantron
Middle levels (3-6) =>
Lower Levels (0-2) => Dragonslayer, Wingspan

To get raw weapon data too, we can also scavenge from Decepticons, Maximals, and Predacons. Stalk their teams, download their result files, and submit as well. In general though, the benefits of known our own tactics and armor will be far more beneficial and scavenging should be a last resort, but if in a week we're still short it's something to consider.


That is also a possibility. If nothing else, it gives insight on their configurations and their weapons of choice. I think that maybe another thing we can look at.

PostPosted: Fri Dec 07, 2007 12:17 pm
by Jeysie
Evolution Prime wrote:If a level 0 or 1 'Bot wants to volunteer to just do the basic weapons, I am more than happy to pay for the all the costs for that volunteer.


I'm willing to do that, if you still need to, since I don't think my team has enough XP (or Energon, heh) yet for a lot of other configurations. Plus I already have an X-Ray laser anyway.

My team's profile.

Just tell me what sort of setup/data you need, specifically.

PostPosted: Fri Dec 07, 2007 12:27 pm
by Evolution Prime
Jeysie wrote:
Evolution Prime wrote:If a level 0 or 1 'Bot wants to volunteer to just do the basic weapons, I am more than happy to pay for the all the costs for that volunteer.


I'm willing to do that, if you still need to, since I don't think my team has enough XP (or Energon, heh) yet for a lot of other configurations. Plus I already have an X-Ray laser anyway.

My team's profile.

Just tell me what sort of setup/data you need, specifically.


Coordinate with Wingspan since he will be the one that is compiling data and has a good grasp on what to be looking for. We will just have you do the basic 0FP/0SKL weapons.