Author Topic: Can anyone explain k/d +1 math? (Read 574 times)

fullback · « **on:** July 07, 2003, 05:09:18 PM »

I don't understand the scoring page math.

funkedup · « **Reply #1 on:** July 07, 2003, 05:16:09 PM »

it's k/(d+1). The 1 is to prevent division by zero.

Puck · « **Reply #2 on:** July 07, 2003, 05:17:41 PM »

If you have 1 kill and NO deaths, you end up with a k/d of 1/0, which really annoys most programmic math APIs. K/D+1 ensures you have a legal operation.

Of course, it makes your K/D calculation WRONG, but it is easy.

Personally I think there needs to be two scores; streak and K/D. When you have no deaths k/d is worthless, and if ($DEATH=0) {ignore}. Once you have died, then you can worry about K/D ratio. Streak is simply the number of kills since your last death/capture. A bail/landing wouldn't count against streak or k/d.

Just imagine, in WWII, George Preddy gets his first kill, but can't put the flag on his bird because his k/(d+1) is still 1...and after his first death, rather than a 1:1 k/d he's at 1:2 for some reason...

Anyway, put on your nomex underwear, because the score potato flames are about to start...

Just my 2 perks.

funkedup · « **Reply #3 on:** July 07, 2003, 05:19:36 PM »

It's
k/(d+1)

not
k/d+1

Puck · « **Reply #4 on:** July 07, 2003, 05:25:40 PM »

Quote

Originally posted by funkedup
It's
k/(d+1)

not
k/d+1

Precisely correct, and I was inconsistant with my formula. I should have just said fx=K/(D+1) where K = kills and D = deaths. Then used fx rather than the whole formula in further conversation. My aplogies.

What's surprising is the code segment didn't get trashed. One of the Rules of the Monistary: Thou shalt not post code fragments on pain of heavy flaming.

fullback · « **Reply #5 on:** July 07, 2003, 05:25:56 PM »

Of course I assumed it was k/(d+1) to avoid division by zero, but the *result* does not match the kill and death statistics.

Let me try to make this more clear: Does anyone know the formula *actually* being used to compute k/d ratio, since the ratio presented does not represent the kill/death data?

funkedup · « **Reply #6 on:** July 07, 2003, 05:41:24 PM »

IIRC they count bails and captures as partial deaths.

hitech · « **Reply #7 on:** July 07, 2003, 05:42:16 PM »

Per help system under scoring.

Air Categories
Kills per Death = TotalKills / ((Discos * 0.5) + Bails + Captures + Deaths + 1)

Arlo · « **Reply #8 on:** July 07, 2003, 05:44:20 PM »

Hot damn! Don't you go nowhere! There tons of people wanna ask you tons o' things one forum up!

Puck · « **Reply #9 on:** July 07, 2003, 05:46:27 PM »

Quote

Originally posted by hitech
Per help system under scoring.

Air Categories
Kills per Death = TotalKills / ((Discos * 0.5) + Bails + Captures + Deaths + 1)

What does the expensive bribe get me? Ki84 modeled? A quick change in the scoring tabulation so I can sponsor another contest?

wetrat · « **Reply #10 on:** July 07, 2003, 05:58:00 PM »

If you've ever done any sort of programming, then you'll know that division by 0 = bad.

ccvi · « **Reply #11 on:** July 07, 2003, 06:18:56 PM »

It isn't Kills per Deaths, it's Kills per Life, thus the +1

fullback · « **Reply #12 on:** July 07, 2003, 07:50:24 PM »

Thanks hitech.

k/d=kills/0.5*(discos + bails + captured) + (deaths +1)

funkedup · « **Reply #13 on:** July 08, 2003, 02:13:57 AM »

Quote

Originally posted by fullback
Thanks hitech.

k/d=kills/0.5*(discos + bails + captured) + (deaths +1)

Your equation has operator precedence issues.

fullback · « **Reply #14 on:** July 08, 2003, 06:53:33 AM »

Oooh you are a clever one. Remind me to *not* post in the early morning...

k/d = total kills/[(discos + bails + ditches)/2 + (deaths + 1)]

or

kd = total kills / [ 0.5*(discos + bails + ditches) + (deaths + 1)]