Interesting stats on number of rounds in a battle (in General)


AdminJonathan October 17 2007 6:51 PM EDT

for the last 24h or so, this is how many finished on or before the given round (i.e. 10194 finished in round 1 + 24461 finished in round 2 = 34655 on or before round 2):
 n_rounds | count
----------+--------
        1 |  10194
        2 |  34655
        3 |  64877
        4 |  94111
        5 | 114010
        6 | 129613
        7 | 141487
        8 | 150724
        9 | 158021
       10 | 163318
       11 | 167761
       12 | 171676
       13 | 174883
       14 | 177152
       15 | 178954
       16 | 180240
       17 | 181547
       18 | 182461
       19 | 183246
       20 | 184148
       21 | 184813
       22 | 185396
       23 | 185831
       24 | 186048
       25 | 187946
out of 191925 total (the rest stalemated).

Anyone care to make a nice graph? :)

48xVanished48 October 17 2007 7:14 PM EDT

Very interesting. Nice post Jonathan.

Admindudemus [jabberwocky] October 17 2007 7:21 PM EDT

so somewhere between 33 and 50 percent of non-stalemated battles end in ranged? wow!

[P]Mitt October 17 2007 7:30 PM EDT

What's interesting to me is that the number of battles that actually go to 25 rounds jumps from the regression line from 24 rounds.

Godpanda October 17 2007 7:40 PM EDT

Where does HoC factor in?

Adminedyit October 17 2007 7:43 PM EDT

rounds 1 - 4 I'd guess :)

Godpanda October 17 2007 7:47 PM EDT

I mean if two teams don't use HoC, then round 4 is melee.
If two teams DO use HoC, then round 4 in still ranged. Or if even one team does.


So, how does that number factor in?

[P]Mitt October 17 2007 7:55 PM EDT

hmm, I think that they are combined, as in HOC would be rounds 1-4 and without one, round 4 would be melee.

This could possibly account for the jump at the end? Because the people with HOC would have an extra round and the round 25/26(if HOC) would combine too.

Maybe.

Adminedyit October 17 2007 7:57 PM EDT

there is no round 26 as far as i know

smallpau1 - Go Blues [Lower My Fees] October 17 2007 8:14 PM EDT

HoC is round 1, round 2 is round 1 for non HoC, correct?

AdminJonathan October 17 2007 10:53 PM EDT

there is no way to separate out hoc rounds from this data. this is just "number of rounds something happened."

[P]Mitt October 18 2007 1:13 AM EDT

Oh so HoC doesn't add an extra round in the beginning, it just shifts the switch from range to melee?

so 1-3 ranged (3 rounds) and 4-25 melee (22 rounds) would become 1-4 ranged (4 rounds) and 5-25 melee (21 rounds)?

Iluvatar[NK] October 18 2007 1:42 AM EDT

Correct. Stalemates still occur after 25 rounds with HoC [it is not extended to 26 rounds]

[P]Mitt October 18 2007 1:59 AM EDT

So if someone with HoC fights someone without HoC, that someone without the HoC only technically has 24 rounds to win?

Iluvatar[NK] October 18 2007 2:07 AM EDT

To the best of my knowledge, yes. I don't have any 26 round stalemates :P

SNK3R October 18 2007 2:33 AM EDT

Did you want something like this?

Sacredpeanut October 18 2007 2:42 AM EDT

Like Mitt said, round 25 is the interesting one, far more battles seem to end in exactly 25 rounds than expected.

AdminQBnovice [Cult of the Valaraukar] October 18 2007 2:45 AM EDT

1900 out of nearly 200000?

Iluvatar[NK] October 18 2007 2:57 AM EDT

Eh, the lnreg of those points doesn't actually put it that far off, 'cause early on there are some off points as well. Interestingly, four round ends are pretty much exactly how the regression line predicts.

SNK3R October 18 2007 3:02 AM EDT

Bart suggested I do another graph as it might be of better interest, so this is a graph of the percentage of total versus that particular round:

Sacredpeanut October 18 2007 3:29 AM EDT

Deviations from the regression in the earlier rounds can be explained but it's quite difficult to explain why there would be more battle finishing in round 25 than any round from 15 - 24. Less battles ending in the early rounds than the regression predicts can be explained because of the difficulties in beating a 4 minion team in less than 4 rounds for example, what would explain the relatively large number of battles finishing in the 25th?

Hyrule Castle October 18 2007 3:34 AM EDT

i'm happy to say that i finish all my battles (i fight) within 3 rounds... about 2.3 round avg ^.^

Lochnivar October 18 2007 4:05 AM EDT

Wouldn't the number of battles ending in the 25th round be skewed by stalemates thus the difference is to be expected?

Incidentally my average battle on attack is 4.3 rounds and on defense is 7.9

It's an interesting insight considering I run a CoC team and I would assume that I'd prefer longer battles.

Fun numbers though!

Sacredpeanut October 18 2007 4:17 AM EDT

Stalemates aren't included

Last Gasp October 18 2007 12:08 PM EDT

Something else to keep in mind (though I've not a clue how it factors in) is the "self selecting" mechanism involved in this process. Each player has a different perspective on "acceptable opponent." For me, 10% and up is an acceptable bonus whereas somebody else might fight down to 1% bonus or below. That would change the rate at which fights end to a certain extent. Those I'm getting 1% bonuses on (though I don't fight them very often) I tend to beat quicker than those I'm getting 30% bonuses on.

AdminJonathan October 18 2007 12:16 PM EDT

thanks SNK :)

Tezmac October 18 2007 12:20 PM EDT

So 50-60% of the battles end in ranged, depending on how you look at the numbering of rounds 1-5 with HoC usage. Shocker.

AdminShade October 18 2007 12:28 PM EDT

What's the function of the initial curve?

QBsutekh137 October 18 2007 1:11 PM EDT

Any possibility of showing battles where both people are under 500K MPR, and those where at least one person is above 500K MPR? Same thing for 1 million MPR?

Just to see if the trends are the same at various levels of play...

SNK3R October 18 2007 1:46 PM EDT

Shade:

Assuming a logarithmic function:

For the Counts vs. Rounds graph: y = 58443Ln(x) + 15329

For the Percentage of Total vs. Rounds graph: y = 30.451Ln(x) + 7.9871

where x = round and y = whatever the y-axis was for that particular graph.

QBSefton [Black Cheetah Bazaar] October 18 2007 1:49 PM EDT

If it matters and it may not, the strategy on Sefton is specifically designed to take advantage of the long combat rounds, and I never finish before round 7, or I should say rarely, I am around 775K MPR something like that

QBSefton [Black Cheetah Bazaar] October 18 2007 1:50 PM EDT

sorry I meant around 725K MPR

Godpanda October 18 2007 3:14 PM EDT

Hmm... Does this confirm to anyone else that ranged damage is too high?

The whole rationale behind the huge ranged damage was that melee gets more rounds. Well, this graph proves differently. Roughly half the fight taking place are ending before melee. But, eh. Too many variables here to know.

QBRanger October 18 2007 3:30 PM EDT

Again, Ranged damage is not too high. I believe all my fights should end in 4 rounds or less. Anything to the contrary is uncivilized.

Nerevas October 18 2007 3:31 PM EDT

*Flexes his melee muscles*

deifeln October 18 2007 6:17 PM EDT

What if there was a skill that took away ranged rounds (sneak attack?)....

Iluvatar[NK] October 18 2007 6:36 PM EDT

A simple reduction to archer damage [beleg's, high elbs, archery, toA, HoC, EBs altogether is way overdone] should do it if a reduction is to be made. I'm not sure if Jon most battles to all end in 4 or 5 rounds, though. If he does, then no change needs to be made.

Personally, I think the melee component should be more than the ranged component, but what do I know :P

NSFY October 19 2007 1:32 PM EDT

I made this most excellent chart and sent it to Jonathan but he seems to have some sort of reluctance to open my Hotmail messages that have attachments. What if this had been important Ombudsman stuff?!?! I realize that SNK already provided some of this but mine has some cool looking mountain shaped data stuff and therefore a much more interesting function for Shade to worry about.

AdminJonathan October 19 2007 2:07 PM EDT

Glad you figured that out, NSFY. It really is a nice graph. :)
This thread is closed to new posts. However, you are welcome to reference it from a new thread; link this with the html <a href="/bboard/q-and-a-fetch-msg.tcl?msg_id=002FXm">Interesting stats on number of rounds in a battle</a>