Twbd Mvp

[Adman]

doesn't happen very often tho

[lunch3]

Shark formula definitely needs to be fixed. tks should be slightly more lenient, same for reps left.

Also in rest of formulas, kills should be worth slightly more then deaths. I think I remember u gain/lose mvp rating points for kills/deaths in the same increments. But that encourages camping. It should be something like 3 points each kill and -2 points each death or something like that. The more deaths usually means the more u actually made ur presence mean something on the tactical level.

[Ewan]

I think flag claims should definatly be a big bonus.

I just got 2 terrkills, 4tks and 10 flag claims but had a -44k rating.

I think maybe deaths with a repel left should be a slight penalty and deaths with no reps should be a slight bonus?

Anyone who's sharked some knows rep & die, rep & die.


Also, if it's possible, it would be nice if you get an extra bonus for multiple kills, a 4fer from a shark will be much more valuable than 4 seperate kills. Maybe this can be done with kills per death? (This would also apply to javs, although for a jav kills per death wouldn't be applicable, as they have many shots per death).

[JacKieChan>]

shark mvp

so maybe the formula is just punched in wrong on the bot, I think the repels used is garbage, i think it needs to be removed from the equation... what purpose does it serve... axe that...

I shark a lot too.

I think reps should be axed from the formula as well. Some wacky games you attach die alot or get killed right away due to no fault of your own.

Terr kills, low tks, flag touches, and a good ratio w/l should get the shark mvp.

Prolly to get shark mvp though it should be hard, like high terr kills, and low tks.

[Ewan]

I don't agree that a high w/l ratio should help.

It's not necessarily good if you don't die a lot in a shark.

[Ekko]

Lets use Kims suggestion.. then.. then... cramming would die XDXD.

[ogron]

Bosshawk got MVP in shark tonight...

OMG WTF BBQ !!!!!

that shit stole my mvp

Um, I think the only 2 things you expect from a Shark (other than repping), are TeKs and lack of TK's. Your W:L ratio isn't overly important, since the game isnt about points, its about holding the damn flag

what are the current formula's for basing stats?

[Jeansi]

different for each ship.

[Jeansi]

Shark: 0.5points(RepelUsedPerDeath) * (.12terr + sum(.06allotherships) - 0.006deaths - (.07(allothershipstks) + .15terrtk + ..09sharkTK))

Terr: .8points * (.06wb + .06jav + .08spid + .12terr + .1x + .06lanc + .09shark - .12deaths)

Spider: .4points * (.06wb + .06jav + .04spid + .08terr + .05x + .05lanc + .09shark - .05deaths)

Jav: .5Points * (.05wb + .05jav + .06spid + 0.12terr + .07x + .05lanc + .09shark - .05deaths - (.07wbTK + .07javTK + .06spiderTK + .15terrTK + .06WeaselTK + .07LancTK + .09SharkTK))

Lanc: .6Points * (.07wb + .07jav + .05spid + 0.12terr + .05x + .06lanc + .08shark - .04deaths)

they aren't using any other ships in ?go base atm so I don't know how to check them

[ogron]

Shark: 0.5points(RepelUsedPerDeath) * (.12terr + sum(.06allotherships) - 0.006deaths - (.07(allothershipstks) + .15terrtk + ..09sharkTK))

My suggestion:

(Points / ((Total Team Deaths-Team Shark Deaths) / RepelUsedPerDeath)) * (Total Reps/Total Match Length) * 0.3(kills) + (1000*Teks + 250*FC - 500*TKs)


Example 1:

http://twd.trenchwars.org/body.php?a...nMatchID=63323

Lunch3:

(7540 / ((468-159) / 2.8)) * (212/22.47) * 0.3(29) + (1000*4 + 250*4 - 500*6) =

68 * 9.43 * 8.7 + 2000 =

7578pts, or 8k


Example 2:

http://twd.trenchwars.org/body.php?a...nMatchID=63320

Lunch3

(9593 / ((205-77) / 3.0)) * (111/16.5) * 0.3(33) + (1000*3 + 250*2 - 500*4) =

228 * 6.7 * 10 + 1500 =

16776, or 17k


Fictional example 1 (awesome game):

Match Win: 15:00-10:00

Lunch3 gets 40-70, a 2.9 rep ratio with 203 reps, 10 TeKs, 4 TK's, 8 FC's

His team has a total count of W:L of 400-250, with 130 shark deaths for his team, he plays an amazing game with a ton of Teks and FC's and his team keeps incredibly low death toll, should in theory be MVP by a mile - game of his life, and the top spider getting 23k with 90-45 and 2 teks and 16000pts or whatever.

(12000 / ((250-130) / 2.9)) * (203/25) * 0.3(40) + (1000*10 + 250*8 - 500*2) =

300 * 8 * 12 + 11000 =

39000 or 39k


Fictional example 2:

Lunch3 has the worst game of his life. HOLY NOOB!

team loses 350-500, 15:00-3:00, he gets 5-90, 4000pts, 12 TK's, 1 Teks, 4 FC's, a rep ratio of 1.8, 160 total reps, boo-urns.

(4000 / ((500-170) / 1.8)) * (160/18) * 0.3(5) + (1000*1 + 250*4 - 500*12) =

22 * 8.8 * 1.5 + -5000

-4710 or -5k


Obviously almost every aspect of this formula needs to be tweaked more specifically (39k is too high and -5k is too low), but the overall jist of it promotes good sharking and places value where it needs to be, and allows for sharks to MVP without having to get 50 kills vs. noobs.

Sharking is all about keeping your team alive, making the most of your reps, and getting TeKs when and if possible and minimizing TK's.

Let's get some formula ideas out there and come up with something that's fair and equitable for sharks, since they do all the unglorified dirty work

[Ewan]

I don't think spids should get more credit for killing a shark than killing a wb or jav, sharks often want to die.

I don't get what the 0.07x in the formulas mean.

[Abatis]

Points?

This is a newb question by a newb, so I've already flamed myself, but why use points in the ratios? You get more points when your team already has the flag, but usually isn't the stuff you do to get the flag that makes you mvp? The one exception is the terr usually has a higher bounty but teks already have a weighted advantadge.
Anyways, here's my attempt at formulas. A simple formula better than the current one would be just (total reps used * 100). If a shark died 100 times and repped 3 times each death which is pretty good would get 30k rating and probably mvp.
More complex formula (250 * (deaths * (avg reps per death / 2.25) - 100 * (tks - kills) + (500 * teks))
This formula basically rewards death as long you use your reps and punishes tks, and it obviously needs tweaking.
Here's an example with the complex formula if a shark got 80 deaths, 2.5 avg reps per death, 5 tks, 15 kills and 3 teks.
(250 * (80 * (2.5/2.25)-100*(5-15)+(500*3)=25k rating

[Kings]

This is a newb question by a newb, so I've already flamed myself, but why use points in the ratios? You get more points when your team already has the flag, but usually isn't the stuff you do to get the flag that makes you mvp? The one exception is the terr usually has a higher bounty but teks already have a weighted advantadge.
Anyways, here's my attempt at formulas. A simple formula better than the current one would be just (total reps used * 100). If a shark died 100 times and repped 3 times each death which is pretty good would get 30k rating and probably mvp.
More complex formula (250 * (deaths * (avg reps per death / 2.25) - 100 * (tks - kills) + (500 * teks))
This formula basically rewards death as long you use your reps and punishes tks, and it obviously needs tweaking.
Here's an example with the complex formula if a shark got 80 deaths, 2.5 avg reps per death, 5 tks, 15 kills and 3 teks.
(250 * (80 * (2.5/2.25)-100*(5-15)+(500*3)=25k rating

BECAUSE YOU ENED SOMETHING TO CALCULATE MVP!
what would you calculate mvp then? with ship speed?

OMFG, I USED THRUST ALL GAME LONG, MY SHIP SPEED WAS 187263721637 MILES AN HOUR, IM MVP!

[lunch3]

I like your formula clausehouse, it takes into account the number of team deaths which is a good comparision to how well a shark performed. Maybe factor that with time then it might work slightly better. From what you were saying, and others, these formulas have too much of a range. A unit step function would work by just cutting off were you need the limits but it seems a bit crude.

A sine function seems good, use your best possible game scenario to determine best rating. But determine this best rating for a long game where 2 sharks are needed and both teams are cramming. Say you figure out that 30k is best possible rating (if you're not talking about a game where enemy just rams mines pumping up shark's rating exponentially) then u make rating=absolute value(30k*sin(absolute value(formula*90 degrees/ best real rating determined by ur last formula))). If a shark gets really close to original 30k rating then it should equate to same thing...the farther away from that past rating they are the lower they stay....more in a comfined range that u wanted. The absolute value takes care of ratings below 0k and above 30k. The only problem is that if a shark is able to get...(with the old formula) a rating of 60k or negative 60k, then they get maximum rating value of 30k. So just be sure that whatever ur coefficient is (whatever number you determine to be best real rating) that 3/4th times this number is a rating that a shark is not really able to get. a sinc function would take care of that decrepency but don't know if u can use that function.

Talk to Force of Nature, he's the one who had code and worked on last formulas I believe. Also I'll post a few more RB games where the sharks did very well imo. If you can bear to do more work you can use those to get that maximal prefereed rating, or if u decide not to take my advice at least look at games to get better idea for what are good shark games.

http://twd.trenchwars.org/body.php?a...nMatchID=62821
http://twd.trenchwars.org/body.php?a...nMatchID=63236
http://twd.trenchwars.org/body.php?a...nMatchID=63230

i noticed there are a decent amount of tks, but the way sharks played in those game allowed our team a lot of time on clock and gave our spiders very good recs.

[Abatis]

BECAUSE YOU ENED SOMETHING TO CALCULATE MVP!
what would you calculate mvp then? with ship speed?

OMFG, I USED THRUST ALL GAME LONG, MY SHIP SPEED WAS 187263721637 MILES AN HOUR, IM MVP!

How about kills?