Welcome to the Forum Archive!

Years of conversation fill a ton of digital pages, and we've kept all of it accessible to browse or copy over. Whether you're looking for reveal articles for older champions, or the first time that Rammus rolled into an "OK" thread, or anything in between, you can find it here. When you're finished, check out the boards to join in the latest League of Legends discussions.

GO TO BOARDS


Real Average Game Time, and Average IP / Game!

123
Comment below rating threshold, click here to show it.

Spellsy

Recruiter

03-31-2011

Attached is an Excel chart which computes and averages the game time of 188 games, and then computes the IP gain that the old system would have given, and compares it all.

*note, this does not include any IP gain from the first win of the day*

Here are the values:
Average Game time: 34.44 minutes*

Average IP Gain per game with the new system: 97.7 for win, 65.1 for loss
Average IP Gain for these same games with the old system: 115.1 for win, 68 for loss

Average IP gain per minute for the New System: 2.9 (win), 1.91 (loss)
Average IP gain per minute for the Old System: 3.57 (win), 2.08 (loss)

So, the new system is 25 IP less per hour (assuming 50 % w/l). (170 vs 145)

Also in the excel file, is a calculator in which you can compute all of these statistics using just the IP values in your match history of your last 10 games (remember to subtract the 150 from your first win of the day).



How was this done? and how accurate is this analysis?
@ how.
What I did in order to derive these numbers was I collected a bunch of IP values from people's match history via this thread: http://www.leagueoflegends.com/board/showthread.php?t=558007 . From there I created a data sheet with all of the values there and then needed to figure out a function which converts IP gain, given win/loss, into Elapsed game time. To do this, I used the inverse of the equations in this thread: http://www.leagueoflegends.com/board/showthread.php?t=620994 which approximates the IP gain of a match for the old and new ip systems, given the game time elapsed. While I double checked the New IP system equations against my data points to make sure they made sense, I had no way to check the Old IP system equation, so there is a possible source of error (if this guy's old equation is flawed).

Once I had the data and the equations, everything was essentially done, and I just did some descriptive statistics analysis (like computing averages, sums, etc.) on it, and formated it etc.

@ accuracy.
Well, I believe these statistics are pretty accurate on how the IP system change effected the average forum goer. However, There are naturally some sources of error.
One source of error was in the data collection stage. The sample was not a random sample, instead it was a voluntary forum sample, causing an innately biased data result. One could say that people who use the forums are naturally more experienced in the game, which means this data is only from one subset of the population rather than a sample of the whole population. However, while this bias is definitely there, I believe it is only minor/trivial.
Another source of error is that the function that Riot uses to determine IP rewards for each game is naturally a discrete, peacewise function. This means that there is no fully accurate inverse to this function, because, for example, the IP value 86 is rewarded for a range of game times (roughly 29:00-29:45 or so). However if I input 86 into the inverse equation (because of its linear properties), I come out with 29.5, which is naturally an explicit value, which is only an approximation/assumption of the true time. The range of error of this is the 20-25 minute mark, where the IP gain remains constant for that 5 minutes (76 for win 51 for loss). So, when I input "76" into my inverse equation, it outputs 23*, when in reality the game could have been anywhere from 20-25 minutes long.
A last possible source of error in this analysis is the possibility that the referenced equations were wrong. While I double checked the New IP system equation with my reference source of data points, the referenced equations are only approximations and do not give 100% accurate outputs. And, the Old IP system equation could be flat wrong and I wouldnt know it, because I can not cross check it with anything, as I have no old IP data references.

However, Overall, I believe that these sources of error are only minor, and this data is overall accurate! check it out yourself!

*I computed the average game time using an assumption that 76/51 IP games are 23 minutes long, when they could have been anywhere from 20 to 25 minutes. If I change assumption value (which I used 23) to a value anywhere from 20-25 it directly changes the average game time. However, even with a "safe" assumption of 25 minutes, the average game length is ~34.8 minutes, still significantly below where riot currently says, unfortunately .

So in the end, I'm just hoping this will bring true light to the IP changes, and hopefully Riot can tweak the system alittle in order to create an IP system which gives equivalent average IP gains as the last system, because I know riot is being honest and they truly didn't want to nerf IP.


Comment below rating threshold, click here to show it.

Irken Warrior

Member

03-31-2011

Hopefully this puts an end to everyone being all mad and stuffs

(EhehehEhEhweheheheHEHEh)


Comment below rating threshold, click here to show it.

Cithsirion

Senior Member

03-31-2011

It's significantly worse for Twisted Treeline players.


Comment below rating threshold, click here to show it.

Spellsy

Recruiter

03-31-2011

Quote:
Cithsirion:
It's significantly worse for Twisted Treeline players.


hmmm i wonder


Comment below rating threshold, click here to show it.

bloboblob

Recruiter

03-31-2011

Beautiful


Comment below rating threshold, click here to show it.

Uccisore

Senior Member

03-31-2011

One of the things the old system did was give longer games. Taking 188 new, shorter games, and imprinting the old IP system doesn't tell us anything useful. 188 games in the old system would have taken longer to play, and thus the IP/minute would be lower than you're representing here.


Comment below rating threshold, click here to show it.

SilentlyAlive

Senior Member

03-31-2011

I only got 80 IP for a half hour game I played today and won zzzzzz


Comment below rating threshold, click here to show it.

Devil Man Type X

Senior Member

03-31-2011

Proves exactly what we've all known. Also shows what *******s Riot are when compared to their lies about IP.


Comment below rating threshold, click here to show it.

Spellsy

Recruiter

03-31-2011

Quote:
Uccisore:
One of the things the old system did was give longer games. Taking 188 new, shorter games, and imprinting the old IP system doesn't tell us anything useful. 188 games in the old system would have taken longer to play, and thus the IP/minute would be lower than you're representing here.


no, it is valid to compare how the IP change effected our IP rewards in real time. the IP system change doesn't change game times directly (or at least, there is a very low probability that people have changed the way they play in order to compensate for the new IP system.), so it is completely fine to compare the differences between IP gain for the games we JUST played, as riot is saying they are trying to keep the IP system now as it would have been.

while these IP systems may have gained the same IP/min 1 month ago, we dont play 1 month ago, we play now and the IP system should reflect current values, not old values


Comment below rating threshold, click here to show it.

Rezyk

Senior Member

03-31-2011

If you calculate the old gains more correctly, you get ~136 per hour. So it looks like bumping it up to 145 was already a buff, nice.

This also corroborates with Riot's results quite nicely.

If you want 170 per hour, that's just asking for another huge buff on top, way past the original average.


123