Thursday November 26 2009
Counter-Strike
Official Design Partner
Story Header

Counter-Strike: GameSensed: The Importance of the First Frag

By: Mark Cheben - Published July 04, 2005 at 3:04 AM EDT - Writer Archive
How important is it to get the first frag in a round? Important enough to revolve your entire strategy around striking first? Absolutely, says GameSense lead developer Jason "Warthog" Roman.


By: Jason "Warthog" Roman

This is the first of many articles (once every week or two) which will use GameSense as a statistical basis for revealing important and interesting information about competitive Counter-Strike. The visible part of GameSense on the GotFrag website is just a glimpse of the system's capabilities, and you will see this in the upcoming months. In the meantime, these articles should wet your appetite.

Today I am analyzing how often a team wins when they get the first frag in a round. I asked a few people around the community how often a team wins when they score the first frag. The responses I received were anywhere from 50-80% of the time, and judging by the number of 50-60% guesses, many players thought it wasn't very important at all.

Wrong.

The team that gets the first frag in a round wins 75% of those rounds. Seventy-five percent! Simply put, if your team is better at striking first, then odds are, you're going to win the match. Period. We've got 15,000 Rounds of CS in GameSense to back this up. It is also staggering how consistent that 75% mark is. I took several random samples of 10, 50, and 100 rounds at a time, and that percentage never deviated by more than a couple percent. Now, first frag statistic is alarming by itself, but there is a lot more we can reveal.

How strong is your position to win the round if you get the first 2, 3, or 4 frags? It's almost a guarantee. Get the first 2 frags and your chance of winning the round jumps drastically to 90%. Even more telling is teams that get the first 3 frags win 98% of the those rounds. Incredible. The other team might as well type kill in console and save everyone time. Given these staggering numbers, you may ask, "Has any team ever won a round after the other team got the first 4 frags?" Yes, but it's happened literally just a handful of times - 7 out of 2839 to be exact - a 99.8% success rate for the team getting the first four. That's almost as good as a pregnancy test (not that I have any experience with that). If there is enough interest I will do a follow-up article on those seven of the greatest clutch (or choke) rounds of all time. It's technically possible to win the round if all 5 of your players were fragged without once fragging the other team, but it's never happened. Let's summarize our results with a chart.


% of Times Winning the Round When Getting the First X Frags

First X Frags Round Win %
174.8 %
290.7 %
397.8 %
499.8 %
5100.0 %


I affectionately refer to this as the '75/90 Rule' or the '75/90/98 Rule.' Now we can break this data down into several more interesting points, so let's start with how the First Frag data is affected by round time. The results may surprise you.

While gathering information for this article I was asked a relevant question: how does your data change when comparing the old 3-minute round timer to the current 1:45? The general consensus was that the first frag would have more of an impact with the shortened round time in determining which team won the round, since the game is (theoretically) played at a faster pace. So I checked the numbers and not surprisingly, the general consensus was correct...sort of. The difference was less than one percent, both around the 75% range - ultimately insignificant. The same insignificance held true for teams getting the first 2 or 3 frags - a 1-2% change at most. I'm busting out full chart mode today so here's the actual numbers.


By Round Time, % of Times Winning the Round When Getting the First X Frags

First X Frags 1:45 Round Win % 3:00 Round Win %
175.0 %74.4 %
291.3 %90.0 %
398.2 %96.8 %


The conclusion here is that the shortened round time had absolutely no significant impact on the importance of striking first.
Continued (1/3) »
Page:

User Comments

1
18% Ranking 18% Ranking 18% Ranking 18% Ranking
0 Frags
+   -

good read
3
17% Ranking 17% Ranking 17% Ranking 17% Ranking
0 Frags
+   -

nice analysis
4
21% Ranking 21% Ranking 21% Ranking 21% Ranking
0 Frags
+   -

nice calculations there and nice read
5
50% Ranking 50% Ranking 50% Ranking 50% Ranking
0 Frags
+   -

great piece
6
9% Ranking 9% Ranking 9% Ranking 9% Ranking
0 Frags
+   -

nice article but experience clans vs medium clans this statistic not real.

and 3:00 for 1:45 = more action in the game !

3:00 = hold ZzZzz..zZz.. hold hold...
7
16% Ranking 16% Ranking 16% Ranking 16% Ranking
0 Frags
+   -

infomrtaive and useful
"So he speaks Spanish? Is that similar to Danish/Swedish/Norwegian?" * pA 4 ever!
8
39% Ranking 39% Ranking 39% Ranking 39% Ranking
0 Frags
+   -

Interesting.

wow i loved this article... please write more statistical analysis stuff, great read!
11
36% Ranking 36% Ranking 36% Ranking 36% Ranking
0 Frags
+   -

yeah, I can tell when maps are even like mill and d2 and maps where the round is decided by the initial confrontation like inferno or cbble. I'm surprised that nuke is the most even, though. It seems like it would be more one sided, but I dunno. Interesting.
The only thing I cheat on is girls.
13
8% Ranking 8% Ranking 8% Ranking 8% Ranking
0 Frags
+   -

noice
14
40% Ranking 40% Ranking 40% Ranking 40% Ranking
0 Frags
+   -

Such a nice kid #12 LOL
15
5% Ranking 5% Ranking 5% Ranking 5% Ranking
0 Frags
+   -

best_article_ever
silly cbble
16
36% Ranking 36% Ranking 36% Ranking 36% Ranking
0 Frags
+   -

pretty interesting but It would be wrong to modify setups because of this if you rely on taking back the bombsite then making strats that will probably give you the first frag wont make you win more rounds
17
108% Ranking 108% Ranking 108% Ranking 108% Ranking
0 Frags
+   -

interesting article and stats, more like these would be cool
18
60% Ranking 60% Ranking 60% Ranking 60% Ranking
0 Frags
+   -

Great article Wart, i love you
#GotFrag - #eMg - RIP Marcus "COLE-" Cole
19
33% Ranking 33% Ranking 33% Ranking 33% Ranking
0 Frags
+   -

thats pretty good never thought about that id like to see something on rushes vs. taking it slow
Have some fun and win.
20
80% Ranking 80% Ranking 80% Ranking 80% Ranking
0 Frags
+   -

cool article.
:x
21
7% Ranking 7% Ranking 7% Ranking 7% Ranking
0 Frags
+   -

#12 ur math is sure very bad....

U can just take an average for pistol rounds... 2 every match.
i.e. 2 every 20-25 rounds(avg rounds per match)

so 15000/25 *2 (no. of pistol rounds in the data set) = 600 x 2 = 1200.

so to see the ratio we can just find out the %

1200/15000 = x / % given in article

so

x = 150/12 * (% given in article)

Just replace the % in artcle to get x which is the respective % for pistol round.

See such stupid posts like #12 makes mathematicians out of gamers :)))
22
26% Ranking 26% Ranking 26% Ranking 26% Ranking
0 Frags
+   -

That is really surprising that 75% of the time a first frag means you win the round. I don't see that in source as much, at least with my team.
#unforgiven
23
98% Ranking 98% Ranking 98% Ranking 98% Ranking
0 Frags
+   -

interesting... Most of it is common sense, but to put some statistics on it is nice... Could convince teams to play a little more agressively on CT...
24
14% Ranking 14% Ranking 14% Ranking 14% Ranking
0 Frags
+   -

uhm where did you get the number from winning with the entrance kill in a pistol round?

Oh you made it up.. thanks.

Considering he presents this article with alot of real data and expect us to use an average for a pistol round only proves my point of bad journalism.

Did he mention save rounds (where they useually try to get the first kill and save the gun)?

Or do you have all these answers in ur ass and you'd expect me to swallow it?

Nice try though.

This comment was edited at 07/04/2005 3:47 AM
25
9% Ranking 9% Ranking 9% Ranking 9% Ranking
0 Frags
+   -

I don't think you can look at it from a purely statistical view. I think it matters how the first frag was done and what the situation is in the round. I think first frags are far more important when the teams are on even terms (i.e. pistol round) and when teams are picked off in the middle of their strat. If they are doing a someone slow and organized strat and you are able to get one pick you force them to adjust and compensate which might make them play in an uncomfortable situation. On the otherside, if the first frag was made when a team is actually hitting the site and after that the player is run over I think the first frag doesn't matter as much because they have accomplished their objective of taking an area over and play defensively from then on.
26
38% Ranking 38% Ranking 38% Ranking 38% Ranking
0 Frags
+   -

good job.
27
101% Ranking 101% Ranking 101% Ranking 101% Ranking
GotFrag Prime!
0 Frags
+   -

#24 you cant tell a save round from a log unless you sit through the demos.
28
17% Ranking 17% Ranking 17% Ranking 17% Ranking
0 Frags
+   -

good read.
29
14% Ranking 14% Ranking 14% Ranking 14% Ranking
0 Frags
+   -

#27 exactly thats why these numbers are... rubbish.

Bad sampling means ur numbers are all garbage. If you took a statistic class you'd know that.

OH WAIT BUT YOU COULD DIVIDE!
30
142% Ranking 142% Ranking 142% Ranking 142% Ranking
0 Frags
+   -

Uhh at the end of page 2, did u say that nuke was an even sided map?
I fart when I pee, how bout you?
31
16% Ranking 16% Ranking 16% Ranking 16% Ranking
0 Frags
+   -

guess that all of us love counter-strike for watching the eco rounds that doesn't belong to the 78,89% of the "x" round :p that the team eco-ing won! ;). Good read but if we had more statistical numbers would be better, I mean we don't even know which demos are we talking about, cpl winter? cpl summer? what?
Advancedshooting
32
24% Ranking 24% Ranking 24% Ranking 24% Ranking
0 Frags
+   -

#22, GameSense now supports CS:Source (we used it for CAL Source playoffs this season, check out GameSense), so we'll be able to check that.

#25, of course statistics never tell the whole story, but they can spot trends and patterns very well.

#24, I find it amusing that because one piece of data is missing, you dismiss the entire article as garbage. We could go on all day about different scenarios with this statistic - checking the 2nd half vs. the 1st half, etc.

However I will concede the point you are making, that it would be useful to look at this stat for pistol rounds. Therefore I quickly checked, and there is indeed a difference. The team getting the first frag in pistol round wins 68% of the time vs. 75%. Also I checked rounds 2 and 3, since those are the only ones we can assume to be save rounds. The number was 80%. The 80% number is more logic, but it also shows that the first frag does not matter as much on the pistol round. I thank you for bringing that point to my attention, but would appreciate if you made constructive suggestions in the future.
GotFrag [b]GameSense[/b] Stats | Lead Developer
33
14% Ranking 14% Ranking 14% Ranking 14% Ranking
0 Frags
+   -

lol woah sorry about being an ass =x

EDIT: lol i posted this before reading 32.

anyways more constructive criticism skipping/messing up "one" part of a problem will result in guess what... the wrong answer.

What can we really use these number for? Pistol rounds? Nope no mentionioning it. Save rounds? Nope. How about gun rounds? Nope because these numbers arent accurate considering you didnt use only gun rounds or save rounds when you punched these number up in your calculator.

This comment was edited at 07/04/2005 4:00 AM
34
20% Ranking 20% Ranking 20% Ranking 20% Ranking
0 Frags
+   -

nice article! keep it up! and yes, make an article about those 7 1 vs 5 situation taken.
35
24% Ranking 24% Ranking 24% Ranking 24% Ranking
0 Frags
+   -

#30, yes I did.

#33, please explain how this is 'bad sampling.' I've got 15,000 rounds of data and sampled both randomly, and using all of the rounds.
GotFrag [b]GameSense[/b] Stats | Lead Developer
36
14% Ranking 14% Ranking 14% Ranking 14% Ranking
0 Frags
+   -

Why is this bad sampling? Because these numbers arent useful for anything. To say you have a 75% percent chance at winning any round w/ the entry kill is wrong. Regardless of the money situation.
37
22% Ranking 22% Ranking 22% Ranking 22% Ranking
0 Frags
+   -

nice read in my opinion. xlent brings up a good point about save rounds and pistol rounds but #27 brings up another good point, its kinda hard to sit through all those demos. I liked the article, because I think it gives some sort of a base to how important entry kills are for CTs. Just because it doesn't account for save rounds specifically doesn't mean the numbers are garbage. Gun rounds happen more than save and eco rounds, so it gives you an idea how it matters on gun rounds, which is more significant in the entirity of the match. But I will give you this, its a good point, but you don't have to flame the guy for it. If everyone flames authors of these articles, then we won't have them anymore, and ya never know, one might help. Thank you for the article.
38
98% Ranking 98% Ranking 98% Ranking 98% Ranking
0 Frags
+   -

over ALL it's true #36 ... wow, of course there's scenarios it doesn't hold true. But from an OVERALL VIEW, it's right. You can't ignore statistics...
39
24% Ranking 24% Ranking 24% Ranking 24% Ranking
0 Frags
+   -

#36, that is just plain incorrect. I'm not sure you even know what the term 'sampling' means the way you are speaking. Perhaps you could elaborate further on the point you are trying to make.

Again #37 I just calculated the pistol round data (and the following 2nd and 3rd rounds), and found that the number in Round 1s changes to 68% instead of 74%, making the entry kill less important. I will add that in my next article as a follow-up point to this one in case people don't read the comments.
GotFrag [b]GameSense[/b] Stats | Lead Developer
40
14% Ranking 14% Ranking 14% Ranking 14% Ranking
0 Frags
+   -

#37 lol okay i guess everything is just dandy...

Thats why they have this comment section at the end of the page to leave comments. Do I question his ability to write and present data? Yes otherwise they'ed be throwing who knows what kind of garbage down our throats.

Im not trying to flame and i hope the author doesnt take any of this personal but as a reader isnt it my responsibility to let other people know when something is wrong?

EDIT:
#39 arent you sampling for the percentage of times winning after getting the entry kill?

I guess if I'm saving and stack my team long A and someone walks into my stack then i have a 78.9% chance of winning ct side?

If you say so.

I dont know about you but everything is getting a bit vague to be an article about statistics.

I must admit I shouldnt criticise the sampling more the use of the sampling.

This comment was edited at 07/04/2005 4:20 AM
41
22% Ranking 22% Ranking 22% Ranking 22% Ranking
0 Frags
+   -

yeah, I am tired and I wrote my post kind of slow, didn't get a chance to read yours about the pistol stats. I appreciate the time and effort that you put forth for the article. Keep up the good work, I love opening up my browser and seeing a new cs article about things like this. Thank you again.
42
22% Ranking 22% Ranking 22% Ranking 22% Ranking
0 Frags
+   -

#40, didn't say you can't tell him what you think, just sayin you could be a little more constructive about it instead of saying "you are wrong."
43
18% Ranking 18% Ranking 18% Ranking 18% Ranking
0 Frags
+   -

ct inferno first kill , winning rate 86% gasp!
[sGcstrike] www.sGcstrike.com .
44
15% Ranking 15% Ranking 15% Ranking 15% Ranking
0 Frags
+   -

One of the top research articles written on this site, very nice.
#Team-Apathy | CAL-m / CEVO-a
46
88% Ranking 88% Ranking 88% Ranking 88% Ranking
0 Frags
+   -

great article, deserves prime. Makes you wonder how often you'll send your 5th to be bait from now on.
Mike "Australia" Antebi ° Former DPAD Editor in Chief °
48
8% Ranking 8% Ranking 8% Ranking 8% Ranking
0 Frags
+   -

didnt gamesense come out after the round time was changed to 1:45 anyway? how do you have data for 3 min rounds? made up? i think so
49
24% Ranking 24% Ranking 24% Ranking 24% Ranking
0 Frags
+   -

The 3-minute rounds were taken from CPL Winter 2003. I was at the event and collected all of the logs because I was already developing the stats system back then and knew it would be useful data.
GotFrag [b]GameSense[/b] Stats | Lead Developer
50
25% Ranking 25% Ranking 25% Ranking 25% Ranking
0 Frags
+   -

finally a good article, i commend u

More Pages

Submit Comments

Registered Users Only

In order to post comments, you must be a registered member. If you have not registered, it's free and easy!

Latest Poll