Jump to content
[DE]Rebecca

Riven Trading & Toolbuilders: Phase 1

Recommended Posts

Posted (edited)

Here are some more thoughts on the matter. As previously shown, it's easy to infer some values that you devs probably don't want to expose. Because it's easier to show the following with lower numbers, I'm going to use the veiled Riven stats for Nintendo Switch again (w3), specifically Veiled Rifle Rivens.

Data (important values bold):

Spoiler

    {
        "itemType" : "Rifle Riven Mod",
        "compatibility" : null,
        "rerolled" : false,
        "avg" : 52.38,
        "stddev" : 51.21,
        "min" : 10,
        "max" : 652,
        "pop" : 15.4639,
        "median" : 50
    },

Inferred: Number of trades total: 195

 

A clear indicator that something is wrong with this information is the relation of mean ("avg") and standard deviation ("stddev"). For this kind of data, even at a quick glance you'd expect the standard deviation to not be higher than about 1/3 to 1/2 of the mean, as explained previously (more precisely, it would be mean - min). But here we have stdev/mean = 0.98. Surprisingly however, the median = 50 is not much lower than the mean = 52.38. So I'm not entirely sure if the median is a good (i.e. non-biased) indicator here, either.

Just to get an idea, I've generated a random normally distributed sample of 195 values with the given mean and standard deviation, and from that I created a histogram with 20 bins. This is meant to show what actual data could look like with these indicators (mean = 52.38 / stdev = 51.21):

Spoiler

A0AMFQ3.png

A histogram with 20 bins of randomly generated numbers that are normally distributed. The key point here is that these random numbers have the same mean and standard deviation as the provided Riven data. But if we pretend that these are Riven trade values, it's clear that this does not look realistic at all.

We know that the actual data has a min = 10 and a max = 652. Having a look at the graph it's clear that this definitely isn't a good way to summarise the trade data -- everything below 10 is essentially garbage, and there's a lot of that in this example.

Pistol Rivens have an even bigger problem with mean = 60.92, stdev = 314.95 and max = 2500. But in this case, the median is clearly a much more robust indicator with a value of median = 20.

The other categories have more sane values:

  • Shotgun:
    • min: 20
    • mean: 53.71
    • stdev: 17.29
    • median: 58
  • Kitgun:
    • min: 5
    • mean: 25.85
    • stdev: 8.72
    • median: 25
  • Zaw:
    • min: 12
    • mean: 30.5
    • stdev: 8.64
    • median: 30
  • Melee:
    • min: 5
    • mean: 32.55
    • stdev: 8.14
    • median: 30

 

Conclusion:

Aggregating and presenting the data through standard deviation and mean -- implying a normal distribution -- is usually a good idea because a lot of things can be modelled that way.

I've had a look at veiled mods where you'd think the prices would be distributed roughly that way because there aren't any variable stats. However, as shown above that's not necessarily the case and conclusions drawn from that might be misleading.

 

So here are some things to consider:

  • Process the data in a way that gets rid of these outliers. They're either clear attempts at manipulation or other "dubious" transactions (high end: plat transfer? / low end: special rebates between friends?). They're definitely not representative of the overall market.
     
  • Present the aggregated data in another way. If we had the information to create Box Plots instead of trying to shoehorn this data into normal distributions, that'd probably be a better way overall.
Edited by Kontrollo
embetterment
  • Like 1
  • Upvote 3

Share this post


Link to post
Share on other sites

At the risk of just talking to myself here, and I guess I will shut up for the time being after this post:

Here's a heads-up for those who want to turn population percentages ("pop" field) into actual number of trades. When I tried to infer the actual trade numbers I've found that in the first two weeks those pop. numbers are truncated, while in the third and presumably future weeks they are rounded to the nearest decimal.

Made up example: let's say the precise pop. percentage of some Riven category is 0.580271343. In the data they provide us there are always at most 4 decimals after the dot for the pop. field.

  • In week 1 and 2 that would result in 0.5802, simply leaving off the rest.
  • In week 3 (and later?) we'd instead get 0.5803, because the next decimal 7 is rounded up.

 

And here are the totals I've been able to infer:

  Week 1 Week 2 Week 3
PC 33177 36336 59117
PS4 11407 11062 12455
XB1 5886 6853 6872
Switch 1239 704 1261

 

Getting individual totals (i.e. by category) should be easy with this information.

  • Upvote 1

Share this post


Link to post
Share on other sites

It's a good start, but still abusable for the mafia.

I might not be a mathematical expert, but i can recommend some additional data fields that might be helpful:

  • How many times that riven has been traded between players?
  • Does both players were friends (existed in friend list), or clan members? << This might be abusable by Remove Friend , Trade , then Add Friend
  • What riven bonus/negatives are most common on high priced ones? By ignoring the bonus values, how much does each bonus type worth? (i.e. if Crit Chance exists in it, it'll worth 500p more, something like that)
  • A graph of price median every day for one month.
  • Riven disposition might affect the price, as popularity might not be enough.
  • Riven current rank. Don't forget about that one too. It might affect a bit, but still...

There are more that i missed, other players surely have their opinion on this. As far as i love riven mods, the only way for me to get them cheap is buying veiled ones. I believe DE can make Riven Trading much more bearable, especially for newer players who are naive on the trading chat.

Share this post


Link to post
Share on other sites

Suggestion:Stats combination rating like +CC +MS +DMG -Harmless 10 stars or tiers  , +Status duration +damage to infested +impact -CC 1 star or tier or anything that will simply identify the quality of the roll.Most new players dont roll rivens cause they cant value properly the rolls and many scammers exploit this ignorance.Though the ideal option would be an auction house even with trade slots restrictions and some basic data like average price just to get an idea like rubico unrolled 1.5k tatsu 1k.As it is now there always be a window for riven mafia or likeminded people to abuse and scam new players and its really sad considering that Warframe has the reputation of a great and supportive community.This small toxic part of the playbase must be disarmed atleast in my humble opinion.

PS.Sorry for the syntax english is not my main language.

Share this post


Link to post
Share on other sites

@[DE]Rebecca and all.

One thing that needs to be asserted is that rivens are not meant to be an investment instrument. Anyone who treats rivens as an investment instrument can sooner or later get motivated to manipulate prices whether in an organized way or through their own creative methods. The focus for DE should be to ensure that people buy rivens for their own personal use and not as a method to boost their platinum income like with most investment instruments in real life.

The second concern is about deliberately manipulating riven trade prices. It's hard to judge if certain trades were "deliberately" placed to skew statistics. But it is possible to discourage players from doing so reasonably. Allow me digress from the topic to cite an example from real life. Everyone wants the best security systems in their home to keep their assets safe while they are away. But the more sophisticated the system is the higher is its operational cost. So there's a trade off between the value of assets one wants to protect and the amount of investment that would be considered reasonable to protect that asset. The most "practical" solution is to have a security system that requires more effort and resources for a malicious user to break than the actual value of assets it is protecting. The underlying principle is to discourage malicious intent. Now how is this scenario relevant in case of rivens?

We don't need the most sophisticated analytical tools to police the riven trades if we could only make it less worthwhile for those who are seeking it as an investment instrument.

PROPOSED SOLUTION:

Have a cool-down period for individual rivens in trades before they can be traded again. I know at least one popular gaming platform has implemented this for items that are traded between players and have the potential for their prices to be manipulated. The cool-down period could be either (1) fixed duration like 2 weeks or may be a month or (2) a function of the value it is traded for. For example, a riven traded for 20 plats would have cool-down period of 24 hours whereas a riven traded for 1000 plats would have a cool-down period of 2 weeks or more before it can be traded again. Of course, for rivens traded with other mods/items/rivens there should be a fixed cool-down period of, say, 2 weeks for all the rivens involved in that particular trade irrespective of their platinum value. Of course, the duration used here are not suggestive but only for illustration only.

[Note: The cool-down period may be flexed like the platinum cost of rushing a build in the foundry, but lets not get there, not for now at least.]

Trying to manipulate riven trades by frequently re-trading the same riven (before its cool-down expires) would then require the riven to be traded at lower values thereby negating the motivation itself.

It may not be the most assured method but I believe this method has (1) reasonable checks and balances to discourage the riven prices from being manipulated statistically (2) can be implemented within reasonable development effort and resources. (3) has no impact/inconvenience for average users who trade rivens for their own personal use or as gifts.

P.S. I am relatively new and not a regular on these forums so excuse me for being repetitive if such ideas have been discussed before.

P.P.S. I know this thread may not be the right place to discuss this but if there is enough public interest around this then mods please free to fork this out as a separate thread if I am not reachable.

  • Upvote 3

Share this post


Link to post
Share on other sites
On 2019-04-09 at 10:23 AM, Kontrollo said:

Here are some more thoughts on the matter. As previously shown, it's easy to infer some values that you devs probably don't want to expose. Because it's easier to show the following with lower numbers, I'm going to use the veiled Riven stats for Nintendo Switch again (w3), specifically Veiled Rifle Rivens.

Data (important values bold):

  Reveal hidden contents

    {
        "itemType" : "Rifle Riven Mod",
        "compatibility" : null,
        "rerolled" : false,
        "avg" : 52.38,
        "stddev" : 51.21,
        "min" : 10,
        "max" : 652,
        "pop" : 15.4639,
        "median" : 50
    },

Inferred: Number of trades total: 195

 

A clear indicator that something is wrong with this information is the relation of mean ("avg") and standard deviation ("stddev"). For this kind of data, even at a quick glance you'd expect the standard deviation to not be higher than about 1/3 to 1/2 of the mean, as explained previously (more precisely, it would be mean - min). But here we have stdev/mean = 0.98. Surprisingly however, the median = 50 is not much lower than the mean = 52.38. So I'm not entirely sure if the median is a good (i.e. non-biased) indicator here, either.

Just to get an idea, I've generated a random normally distributed sample of 195 values with the given mean and standard deviation, and from that I created a histogram with 20 bins. This is meant to show what actual data could look like with these indicators (mean = 52.38 / stdev = 51.21):

  Reveal hidden contents

A0AMFQ3.png

A histogram with 20 bins of randomly generated numbers that are normally distributed. The key point here is that these random numbers have the same mean and standard deviation as the provided Riven data. But if we pretend that these are Riven trade values, it's clear that this does not look realistic at all.

We know that the actual data has a min = 10 and a max = 652. Having a look at the graph it's clear that this definitely isn't a good way to summarise the trade data -- everything below 10 is essentially garbage, and there's a lot of that in this example.

Pistol Rivens have an even bigger problem with mean = 60.92, stdev = 314.95 and max = 2500. But in this case, the median is clearly a much more robust indicator with a value of median = 20.

The other categories have more sane values:

  • Shotgun:
    • min: 20
    • mean: 53.71
    • stdev: 17.29
    • median: 58
  • Kitgun:
    • min: 5
    • mean: 25.85
    • stdev: 8.72
    • median: 25
  • Zaw:
    • min: 12
    • mean: 30.5
    • stdev: 8.64
    • median: 30
  • Melee:
    • min: 5
    • mean: 32.55
    • stdev: 8.14
    • median: 30

 

Conclusion:

Aggregating and presenting the data through standard deviation and mean -- implying a normal distribution -- is usually a good idea because a lot of things can be modelled that way.

I've had a look at veiled mods where you'd think the prices would be distributed roughly that way because there aren't any variable stats. However, as shown above that's not necessarily the case and conclusions drawn from that might be misleading.

 

So here are some things to consider:

  • Process the data in a way that gets rid of these outliers. They're either clear attempts at manipulation or other "dubious" transactions (high end: plat transfer? / low end: special rebates between friends?). They're definitely not representative of the overall market.
     
  • Present the aggregated data in another way. If we had the information to create Box Plots instead of trying to shoehorn this data into normal distributions, that'd probably be a better way overall.

Thank you so much for putting this information forward. A guy from the riven mafia has been posting a cleaner version of the data that DE is posting here on a weekly basis to make the data more visible.

I have accused him of using that to support his practice of manipulating the market by simply including the max riven prices. It's frustrating that they get away with this and ruin an aspect of the game in the process.

He just called me a liar and speculative, but your post backs up my suspicions clearly. I'm not good at statistics past realizing that something is wrong with the data to begin with.

 

DE posting this data as-is is only helping the people trying to manipulate the prices. Riven trading in general needs to be more accessible, transparent, and visible in order to resolve the manipulative practices that are hurting the game overall.

  • Upvote 2

Share this post


Link to post
Share on other sites
1 hour ago, Lobosandia said:

Thank you so much for putting this information forward. A guy from the riven mafia has been posting a cleaner version of the data that DE is posting here on a weekly basis to make the data more visible.

I have accused him of using that to support his practice of manipulating the market by simply including the max riven prices. It's frustrating that they get away with this and ruin an aspect of the game in the process.

He just called me a liar and speculative, but your post backs up my suspicions clearly. I'm not good at statistics past realizing that something is wrong with the data to begin with.

 

DE posting this data as-is is only helping the people trying to manipulate the prices. Riven trading in general needs to be more accessible, transparent, and visible in order to resolve the manipulative practices that are hurting the game overall.

You're welcome.

Note that for veiled Riven prices you should just use the Median. When the trade numbers are high enough (they are on PC), it is not easily manipulated. Here are the veiled stats of week 4. First on the Switch for comparison, because it has sane numbers for mean and standard deviation:

Switch Week 4

itemType compatibility rerolled avg stddev min max pop median
Kitgun Riven Mod   False 27.16 9.48 10 60 3.6213 27
Melee Riven Mod   False 32.34 9.37 4 60 11.3723 30
Pistol Riven Mod   False 21.25 8 7 50 6.6074 20
Rifle Riven Mod   False 49.09 15.38 14 100 15.629 50
Shotgun Riven Mod   False 51.82 16.75 15 100 2.1601 50
Zaw Riven Mod   False 32.32 16.12 3 120 5.2097 30

 

On PC these are clearly not useful, the standard deviation and max numbers are too high. But the Median remains stable:

PC Week 4

itemType compatibility rerolled avg stddev min max pop median
Kitgun Riven Mod   False 31.24 109.75 2 4000 2.7861 27
Melee Riven Mod   False 33.1 108.06 2 4200 9.2007 30
Pistol Riven Mod   False 27.74 160.97 2 4300 4.4012 15
Rifle Riven Mod   False 61.01 68.43 2 4000 10.5553 61
Shotgun Riven Mod   False 51.02 37.03 2 1051 2.228 50
Zaw Riven Mod   False 33.32 91.24 2 2500 3.6374 30

 

(Also note: For unveiled mods it's entirely possible that the max prices are that high because perfect or near perfect rolls can be incredibly rare.)

 

 

Inferred total number of trades week 4:

  • PC: 67549
  • PS4: 14875
  • XB1: 7599
  • Switch: 1574
  • Upvote 2

Share this post


Link to post
Share on other sites
On 2019-03-29 at 6:43 PM, [DE]Rebecca said:

All of this data is made up of Riven + Platinum trades. We do not include '1 Platinum' trades in our averages, etc, because we consider these to be between friends and not reflective of Market reality

What about multiple rivens in a single trade?  We've been seeing some weird numbers, specially veiled rivens being sold for thousands of plat. Are these rivens being traded alongside other valuable rivens?

Share this post


Link to post
Share on other sites
Posted (edited)

Week 5: At least on PC & PS4, the veiled Rivens look manipulated again, Shotgun Riven mods being the only exception. So this is a friendly reminder to use the median and not the mean (avg in the stats) for prices.

 

Week 5 median values of veiled Rivens

  Kitgun Melee Pistol Rifle Shotgun Zaw
PC 29 30 15 60 50 30
PS4 17 20 13 20 30 15
XB1 20 25 20 30 40 20
(Switch) 27 30 20 50 50 30

 

Also @devs: I'd like to reiterate that it might be better to have the necessary information to create box plots. And maybe don't produce any data for Rivens that don't have less than a certain number of trades per week, plus add a date ranges in that case (i.e. publish only every X weeks, but mention the interval).

Mean and standard deviation aren't good indicators in every case, and the mean is more easily manipulated/influenced by outliers than the median.

 

Edit: Week 5 totals. After deriving these, it looks like all of the Switch data is identical to last week's:

  • PC: 61921
  • PS4: 14941
  • XB1: 6951
  • (Switch: 1574)
Edited by Kontrollo
  • Like 3

Share this post


Link to post
Share on other sites
Posted (edited)

I'm a super programmer 😀

166.jpg

Windows cmd + wget + jq json parser

Spoiler

@echo off
SetLocal EnableExtensions EnableDelayedExpansion
title RIVEN PRICE

:start
    set platform=KUVA_FARM
    set tmp=
    echo.
    echo   1 - PC    2 - PS4        3 - XboX    4 - SWICH
    echo.
    choice /c 1234 /m " "
    if Errorlevel 4 (
        set platform=SWI
        goto :run
    )
    if Errorlevel 3 (
        set platform=XB1
        goto :run
    )
    if Errorlevel 2 (
        set platform=PS4
        goto :run
    )
    if Errorlevel 1 (
        set platform=PC
        goto :run
    )

:run
    echo.
    set /p WEAPON="RIVEN MOD WEAPON: "
    for /f "usebackq delims=" %%a in (`echo "!WEAPON!" ^| jq -r ". | ascii_upcase"`) do set WEAPON=%%a
    echo.
    wget -nc -nd -q -P %temp% http://n9e5v4d8.ssl.hwcdn.net/repos/weeklyRivens%platform%.json
    set file_DE=%temp%\weeklyRivens%platform%.json
    for /f "usebackq delims=" %%b in (`type %file_DE% ^| jq -r ".[] | select(.compatibility==\"%WEAPON%\") | .compatibility"`) do set tmp=%%b
    if not defined tmp (
        echo Weapon not found
        goto :choose
    )
    echo ----------------------------------------
    type %file_DE% | jq -r ".[] | select(.compatibility==\"%WEAPON%\") | select(.rerolled==true) | \"    \" + .compatibility + \" RIVEN MOD\", (\"MEDIAN OF PRICE:    \" + (.median | tostring) + \" plat\"), (\"MAXIMUM PRICE:        \" + (.max | tostring) + \" plat\"), \"Rerolled?        \" + (.rerolled | tostring)"
    echo ----------------------------------------
    type %file_DE% | jq -r ".[] | select(.compatibility==\"%WEAPON%\") | select(.rerolled==false) | \"    \" + .compatibility + \" RIVEN MOD\", (\"MEDIAN OF PRICE:    \" + (.median | tostring) + \" plat\"), (\"MAXIMUM PRICE:        \" + (.max | tostring) + \" plat\"), \"Rerolled?        \" + (.rerolled | tostring)"
    echo ----------------------------------------
:choose
    echo.
    echo    1 - Again    2 - Exit
    echo.
    choice /c 12 /m " "
    if Errorlevel 2 goto :end
    if Errorlevel 1 goto :start

:end

 

maybe someone will come in handy😀

Edited by (PS4)highway_street
  • Like 1
  • Haha 1

Share this post


Link to post
Share on other sites
Le 14/04/2019 à 22:07, Deflektor a dit :

Suggestion:Stats combination rating like +CC +MS +DMG -Harmless 10 stars or tiers  , +Status duration +damage to infested +impact -CC 1 star or tier or anything that will simply identify the quality of the roll.Most new players dont roll rivens cause they cant value properly the rolls and many scammers exploit this ignorance.Though the ideal option would be an auction house even with trade slots restrictions and some basic data like average price just to get an idea like rubico unrolled 1.5k tatsu 1k.As it is now there always be a window for riven mafia or likeminded people to abuse and scam new players and its really sad considering that Warframe has the reputation of a great and supportive community.This small toxic part of the playbase must be disarmed atleast in my humble opinion.

PS.Sorry for the syntax english is not my main language.

Irrelevant since all weapons don't necessarily need the same value to really shine. For instance an atomos or an Amprex would benefit a lot from punchthrough as opposed to many other weapons, an Ocucor would benefit a lot from more ammo clip etc.

All depends upon a lot of factors and the rivens you're talking about are so rare to really care about and so expensive that new players won't be able to buy some anyway. Auction house wouldn't help since it prevents absolutely nothing at all, people would still be abused and we'd got bots everywhere to snipe what the riven mafia would need to acquire.

Truth is that auction house also sucks, i'm still wondering why people keep talking about that since it never worked that well in any game before. It's also all but casual friendly - no one wants to spend his life on this kind of things, but many greedy people would.

Share this post


Link to post
Share on other sites

Week 6 has some funny numbers again, so I'm not even going to go into details. The only doubtlessly useful indicators are veiled Riven medians as before.

 

Week 6 median values of veiled Rivens

  Archgun Kitgun Melee Pistol Rifle Shotgun Zaw
PC 80 30 30 15 62 50 34
PS4   15 20 13 20 30 15
XB1   20 25 20 30 50 20
Switch   25 35 20 50 55 30

 

Inferred total number of trades week 6:

  • PC: 57816
  • PS4: 14363
  • XB1: 7381 
  • Switch: 1294
  • Upvote 1

Share this post


Link to post
Share on other sites
Posted (edited)

It seems that statistics from multiple rivens in a single trade are being included, which might be skewing the data of all these weeks.

Here is an example from week 5:

itemType

compatibility

rerolled

avg

stddev

min

max

pop

median

Pistol Riven Mod

SEER

FALSE

231,91

464,78

2

1888

0,0371

20

Melee Riven Mod

DARK DAGGER

FALSE

168,14

454,87

5

1888

0,0452

25

Melee Riven Mod

SILVA & AEGIS

FALSE

133,05

340,86

7

1888

0,0921

50

1888 is an odd number to have as a max value for 3 different weapons. Was that a coincidence? Or this value represents a single trade involving 3 rivens?

Edited by Agropeu
  • Like 2

Share this post


Link to post
Share on other sites
Posted (edited)
il y a une heure, (PS4)highway_street a dit :

Thank you very much!
Can you make the on/off option for column "rerolled"
For example: display only rerolled=true rivens and sort by other columns.

 

If you type True or False in the search bar, you can show only rerolled or not rerolled riven and when you double-click on the title of the column, you can sort by price, name or the others columns (if i correctly understand).

Edited by NeoGuardian

Share this post


Link to post
Share on other sites
Posted (edited)
On 2019-03-29 at 2:53 PM, NightmareT12 said:

I wonder what actual impact this will have in the coming few weeks. Nonetheless thanks for setting this up guys!

I assume this would give us a range of "reasonable" prices for rivens of any particular weapon (depending on stats), and players can price accordingly.

For example, a good Furax Riven is NOT worth the same as a good Amprex Riven or whatever.

However, I don't know if players will account for popularity when pricing, although they should. In my experience, sellers never acknowlege how unpopular a weapon is. Sellers will still want 700p for a Cestra or Anku Riven, and consider anything less as "low-balling". Then one Tenno gets so desperate they pay that price, and now the average price for all those Anku Rivens skyrockets.

EDIT
Just saw that Tetra rivens have min-max sales of 10P - 4500P. I don't know what to do with that information haha

Edited by MechaTails
  • Upvote 2

Share this post


Link to post
Share on other sites

Actually semlar did a great job, why not allow him to carry on??

so far I think DE spent a lot of time to come up with something which is less useful compared to what semlar did

i don’t think semlar spent such a long time to come up with what he had

 

 

 

Share this post


Link to post
Share on other sites
On 2019-03-30 at 6:08 AM, korndolorous said:

Can we have Median, Standard Error of the Mean in the next phase? since it would be better for riven pricing

I think price of riven should base on riven score, because roll or unrolled doesn't matter to me, for example 

riven has [ + dmg to grineer , + damage to corpus , + damage to infested ] should have 0 or 1 score. (junk tier) and [ + crit chance , + crit damage , + multishot, - damage to infested ] should have 10/10 or 9.5/10 score (god tier), while both of them are unrolled the first one will give you 0 - 100 (or more depend on popularity of weapon), the second will give you 200 - 5000 or more base on popularity and weapon stats

so the price is base on stats tier and weapon popularity

  • price of junk / common tier usually base on number of rolled 0 / 1-4 / 4-8 / 8+
  • price of  decent / good / god tier base on stats of riven  

** there are 3 tier of stats, and 5 tiers of riven

Stats Tier 

  • High tier / will give weapon more damage like(damage, multishot, critical chance, critical damage, slide crit , - zoom, - dmg to infested, etc. )
  • Medium tier / will give weapon utility like ( element damage, physic damage, - recoil, etc.)
  • Low tier / will make weapon go bad ( damage to faction, - damage to faction, magazine size, - damage, - multishot, etc.)

some time element / physic damage and utility stats are high tier like - recoil on grakata

Weapon Tier / ( + positive stats , - negative stats)

  • S or god tier      / +3 high -1 high  ,  +2 high -1 high for some weapon ( 2000+ ) 
  • A or good tier    / +2 high -1 high  ,  +3 high ( 1000+ )
  • B or quite good / +2 high + 1 low or medium  ( mostly 400+ )
  • C or common   /  +1 high 
  • F or junk tier     /  +0 high 

Conclusion so the data that we collect should be based on score on each riven

Note :   price are base on stats on each weapon for example + critical chance are good on crit weapon, but useless on status weapon

 

 


 

i doubt that a bane of everything riven should be considered junk, faction damage is actually quite strong, just not used much due to having to swap and the investment costs however on a riven with all 3 its alot less costly and it has the benefit of never needing to be swapped removing all bad aspects of it. 

you were right with utility stats being hightier in some cases like exergis is best with mag cap (50%) damage or multi and either reload or status chance with a neg.

price based on stats is not bad but what happens to off meta rivens that not everyone realises is good eg at one point long ago when i started playing it was almost impossible to sell a -impact riven and now -ips is everywhere cuz a certain youtuber showed how good it was, look last night i had a clan mate say strun wraith isnt worth a crit hybrid build, till they saw me 2 shotting 145s with mine, heck at one point cuz i saw everyone selling crit astilla rivens i thought crit was the way to go, ppff i got a full status one and never looked back, it was by far stronger than my dual crit riven.

about the prices, i feel like people should be able to put whatever prices they like but just have a max cap of like 4k i guess. its probs obvious given i started a certain thread about super inflated riven prices but i honestly believe no riven is worth 10k hell even 4k is a stretch.

Share this post


Link to post
Share on other sites
On 2019-04-27 at 4:00 AM, 000l000 said:

Irrelevant since all weapons don't necessarily need the same value to really shine. For instance an atomos or an Amprex would benefit a lot from punchthrough as opposed to many other weapons, an Ocucor would benefit a lot from more ammo clip etc.

All depends upon a lot of factors and the rivens you're talking about are so rare to really care about and so expensive that new players won't be able to buy some anyway. Auction house wouldn't help since it prevents absolutely nothing at all, people would still be abused and we'd got bots everywhere to snipe what the riven mafia would need to acquire.

Truth is that auction house also sucks, i'm still wondering why people keep talking about that since it never worked that well in any game before. It's also all but casual friendly - no one wants to spend his life on this kind of things, but many greedy people would.

tbh based on how well riven.market and warframe market works and auction house kinda thing might not be that bad but, hopefully its alot better than those sites if its ever done.

Share this post


Link to post
Share on other sites
Posted (edited)

Is the value for "pop" broken this week? It seems to be wildly incorrect on all platform, e.g. from PC:

 

Quote


	{
		"itemType" : "Rifle Riven Mod",
		"compatibility" : null,
		"rerolled" : false,
		"avg" : 54.92,
		"stddev" : 17.84,
		"min" : 2,
		"max" : 600,
		"pop" : 100,
		"median" : 55
	},

 

EDIT:

Nevermind, I see the initial thread post has been updated:

Quote

pop: Popularity of this Riven Type being traded within all traded Rivens. Popularity of this Riven Type being traded in relation to the most popular Riven Type in the list. Effective May 12 Data.

In general, I'm not sure what the benefit of changing this even is. It's a ratio based on one type of item sold as opposed to the entire population of items sold. We would have calculated this ourselves if we wanted with the old system, and with this new system in place it feels more like we're losing information about what's actually going on in ways of Riven trading, which (from what I understood) this whole project was supposed to help alleviate.

If this is the route it's going to take, could the ratios at least be decimals? Having them as whole numbers isn't very informative since well over a majority of the items (483 out of 576) are listed with a pop of 1, which doesn't really tell you much beyond "these traded less frequently" yet with what actual scale is unknown. For instance, a non-rolled Fluctus has a pop of 1 this week. Last week, with the old system, a non-rolled Fluctus had a pop of 0.2914%, which lets you estimate an amount of 128 total traded. With this new system, if we assume a similar amount was traded (let's say 90 for this sake), a pop of 1 with 90 actually traded would be the same pop as some other item with 2 actually traded.

Edited by MatNova
Removed extra samples from JSON data

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...