Guild Wars 2

First of all, great research Samu and well presented. +1 from me.

I was thinking of doing this myself but presentation wise, you did a way better job than i would have ever done.

One quick question to clarify: Which items from the bucket are categorized under crafting in the graphs? MIght be nice to color code the items from the bucket with the categories you chose for the graphs.
As you mentioned, the diversity of the bucket is quite debatable and i have some suggestions on how to extend your bucket, if you plan on continuing your research:

In my opinion its a little bit too focussed on end game crafting and i would have included more t1-4 common, fine and rare crafting mats as they have significant value for craft leveling alts in general and t2-4 common mats in ascended crafting.

It would also be nice to see some data on basic and high level cooking mats (potatoes, carrots, onions, omnoms, truffles, lemongrass) as well as some favourite food items.

As crafted rare lvl 80 weapons are the main supply for pre cursors through forging, it might be nice to see a graph for them to compare to pre cursor prices.

Vanity items like weapon and armor skins might also be interesting to have a look at but i guess its a bit harder as they have been introduced at different times.

Concerning pre cursors:

There are 3 significant price variations, the first was Lost Shores, where alot of pre cursors dropped from the event and the prices actually went down.
Then prices rose significantly until April. During that time, some people made alot of gold with Lodestones through exploitable means and cornered the pre cursor market, buying up a good chunk of the supply ( i know that discussing exploits is not allowed, if this is also true for exploits that have been resolved, I hope a mod will contact me and i will edit my post). The decline in Lodestone prices starting in January, even though the demand was higher as more people started crafting Legendaries from the Lost Shore drops, supports my theory.
However, during that time, Anet also banned alot of accounts that where involved in that and alot of pre cursors that we stored in those account inventories and banks simply were deleted from the economy, resulting in the price rise until April.

From April 2012 until now, pre cursor prices have actually been quite stable, i think, with the exeption of August/September, where prices rose again 15-20%.
This has to do with the “farming events” of the Jubilee and Scarlets Invasions of course. Those events were introduced to prepare the player base for ascended weaponcrafting, which was right around the corner but the majority of the player base actually used their new wealth to buy pre cursors and go for Legendaries instead.
They also raised the drop rate for pre coursors at some point in the first 6 months i think but i am not 100% sure, when that was.

As i already said, i appreciate your work on this, unfortunately its hard to determine the real rate of inflation, as we dont know the average income per player.
It would be nice to know the average income/hour of gameplay for each month but it will be hard to measure (even for John Smith) because the value of lootdrops makes up the mayority of wealth generated compared to coin drops.

This is exceptional work Samuirai. Thank you, +1 and bookmarked. Wanze, you always provide great feedback, thanks for your input as well.

An index showing which patch is which by title would eliminate flipping back to the LS page and checking dates.

I applaud the idea and publishing the results in a clear and intuitive fashion. You have clearly demonstrated that the GW2 trading market is incredibly efficient and “works” as well as anyone could hope. In particular, your charts make it easy to show some important ideas:

• Gold prices for certain popular items haven’t increased as much as people think they have.
• Prices follow clear patterns relative to changes in availability and interest (supply and demand).

I will be pointing to your site when I discuss the TP with friends or when discussing certain topics on various forums.

However, I would quibble with your use of the word “inflation,” with the choice of items for the market basket, and your weighting of those items.

• First, inflation has a technical meaning for economists and a vague meaning for the general public. I would recommend using some other term.
• If you are going to use that word, the charts should also include how easy it is to obtain gold, something that, as you note, is much more difficult to measure. For example, the cost of T3 cultural armor is fixed, but it’s far easier to obtain the gold required, so arguably, T3 is “cheaper” now than when the game launched.
• The specific items you have chosen don’t reflect the playing experience of huge elements of the player base: relatively few people craft much, even fewer people work on legendaries (they are luxury items rather than required), and the bundle doesn’t include the sort of stuff that people seem to talk about wanting/using frequently.

I would propose instead that you consider three different bundles:

• Essentials: this covers things that everyone seems to feel the need to do for each character, specifically gear up in full exotics (trinkets, one set of armor, three sets of weapons) and expand their personal inventory. For this, I’d look at:
  • Average cost of named exotics (they aren’t directly affected by material costs and are easy to track).
  • Some average for weapons (there are a variety of ways to do this).
  • Un-upgraded trinkets (they seem to be a lot cheaper than crafted ones).
  • Cost of four, 15-slot bags (whatever is cheapest, usually leather, but in the past it’s also included cloth).
• Basic Cosmetics: This covers dyes and skins. Nearly everyone tries to obtain a few specific colors and a few better-looking skins. A reasonable proxy for color is the price of unID dyes and a stand-in for skins could be the price of the mystic weapons (mystic spike, etc), since they are among the most plentiful.
• Luxuries: Things that no one ever needs, but plenty of people covet. I would consider looking at the five priciest non-exclusive dyes, the average price of legendaries (which also covers lodestones, precursors, and T6 materials), and the average price of permanent items that only drop from the BL Chest.
• Other: Your current system already covers a couple of things well and I’d like to see that continue, e.g. materials.

Again, I love what you have done and I look forward to seeing more.

Thank you all for the feedback.

@Illconceived Was Na.9781

I thought about adding some skins to the item list but the problem I had was that those were introduced later. And then it is very difficult to compare previous numbers. For example if I would include the Dreamhistle Weapons, I couldn’t include almost a year of gw2.

It is easy for me to generate those charts when I got a list of items. If somebody wants to write down a list with exactly the items and quantities, I would generate a chart and post it here:

item_id, item_name, quantity

That would be cool

The individual charts are awesome and really informative. It’s well presented. It’s a great resource.

From a technical standpoint I would caution against using the term inflation on your aggregated chart, as you are clearly only looking at a subset of goods and their weights are ad hoc; inflation means something pretty specific and this is only a very loose, biased measure of it.

Of course from a marketing perspective please by all means call it inflation; most people have no idea what the word means but they know it’s sensational and that they should care, and you might as well ride that train.

I would definitely recommend putting some vertical lines on the graphs for major patches to make it more clear exogenous shocks are responsible for large movements; you could basically run an event study on a patch and show its effects.

About the term inflation: I am not an economist :S I have studied computer science. So I am pretty sure I do stuff pretty wrong. As you guys have noted, apparently I have no idea what the word “inflation” means
I basically just try to be logical. But if you guys could correct me and tell me what exactly I have measured I would love to learn more.
Especially if this could maybe used as a reference for annoying economy discussions, so I would prefer if my source is as correct and as understandable as possible.

I would also have no problem to add another section or change my texts with credits if somebody else wants to describe it better.

Samuirai.4561:

About the term inflation: I am not an economist :S I have studied computer science. So I am pretty sure I do stuff pretty wrong. As you guys have noted, apparently I have no idea what the word “inflation” means
I basically just try to be logical. But if you guys could correct me and tell me what exactly I have measured I would love to learn more.
Especially if this could maybe used as a reference for annoying economy discussions, so I would prefer if my source is as correct and as understandable as possible.

I would also have no problem to add another section or change my texts with credits if somebody else wants to describe it better.

Bah, a couple things you learn on the internet is that everyone fancies themselves as an expert economist (ex. “your definition is wrong” but won’t give you the right one), computer programmer and statistician (ex. “your data sucks. everyone knows you must do a million trials to get data or else just RNG”) even if they are not. Cue the “but I have X degree posts”. I’m not. From a “consumer perspective” I’d argue that the basket of goods/price index method that you used provides the most useful measure of inflation experienced by players and is in accordance with government methods of reporting inflation
Even the wikipedia page suggests "However, most economists today use the term “inflation” to refer to a rise in the price level." Don’t get down on yourself or hung up on the pedantic posts.

Good work. Interesting. Useful.
Thanks.

thehipone.6812:

Don’t get down on yourself or hung up on the pedantic posts.

I dont see pedantic posts in this topic.

My comment about the word “inflation” is strictly related to the rhetoric, and not the definition: using the word generates a lot of passion and thus it tends to distract from anything else being discussed.

I think it’s perfectly fine for non-economists to use the word to mean “stuff costs more gold now,” even though professionals (and internet amateurs) would argue that’s technically incorrect. However, since no two people quite agree on the word’s meaning, I try to avoid using it when discussing things like GW2’s economy.

Samuirai.4561:

@Illconceived Was Na.9781

I thought about adding some skins to the item list but the problem I had was that those were introduced later. And then it is very difficult to compare previous numbers. For example if I would include the Dreamhistle Weapons, I couldn’t include almost a year of gw2.

Good point. What I was thinking (and didn’t clarify) is to use a type of average of the available skins:

• there will always be a “current” skin, costing 1 ticket
• there will always be a “most recent” skin, that just started costing multiple tickets.
• the rest will be available forever at a set cost (3, 5, or 7 tickets).

I imagine with some trial and error, one could weight these in relation to each other. It will be a lot more guesswork than the standard US bundle of goods, but even that involves some trial & error.

Samuirai.4561:

It is easy for me to generate those charts when I got a list of items. If somebody wants to write down a list with exactly the items and quantities, I would generate a chart and post it here:

item_id, item_name, quantity

That’s a very generous offer and a great idea. I’d love to see some more discussion around what types of bundles are worth looking at over time and what those bundles should contain. I’ll give it some more thought and put together a list based on the categories I outlined in my earlier post and we can see if that shows us anything interesting about the market’s evolution.

I doubt we can ever find a true market basket of goods that reflects everyone’s needs. Instead, what I’m hoping is that we can setup multiple bundles that reflect large subsets of the community. Those who play have played one main since launch have different spending habits than a new player with multiple alts or collectors (those who want every dye, every mini, every piece of cultural armor, …).

Samuirai.4561:

I basically just try to be logical. But if you guys could correct me and tell me what exactly I have measured I would love to learn more.

What you have is a measure of inflation, and mathematically you are on the right track. It is the first real attempt at measuring inflation that I’ve seen, and you’ve done a good job of it.

The problem is that your measure of inflation has a lot of statistical bias, stemming from two things – an incomplete (unrepresentative) sample of goods, and an ad hoc weighing of the volumes of those goods. In fairness, addressing these problems is hard; unfortunately, if you don’t address these problems, a measure of inflation might not be very useful and could imply erroneous conclusions.

To understand how it can go badly, look at the Queen’s Jubilee patch. In general it is understood that this was inflationary, but that did not mean that all prices went up. Vicious claws, for example, crashed hard during that patch, and on a more general level, cores and lodestones dropped steadily across that time frame. If you only looked at those goods, you wouldn’t conclude that there was inflation, but that there was deflation; similarly, if you looked only at goods that went up, you’d conclude there was a lot more inflation than there actually was. This is why you have to make sure what you are including is representative; if you leave things out, your sample is biased by whatever those things did.

The weights are also really important; even if you have everything included, your aggregate prices are going to be a linear combination of those individual prices, and as I’m sure you know if you take a linear combination of a bunch of different curves that are moving with some independence of each other you can get pretty much anything you want out of the aggregate. So you really need some principled way of weighing the different samples to approximate their real trade volumes.

On the bright side, you are on the right track, and both of these issues can be addressed. You do have a measure of inflation, and issues of its precision and biases are solvable. It’s a useful tool, I’d just be wary of taking the aggregated chart as a measure of inflation at face value.

As a first pass at approaching this in a principled way, I’d suggest gathering price time series data on as many items as possible and hammering at it with principle component analysis / eigenvalue decomposition to identify the relevant buckets of goods (and pick representative samples), and I’d then hit the volume time series data of those items with some sort of Poisson regression to get a measure of volume.

Fire me a PM if you’re interested in pursuing something rigorous like this.

Samuirai.4561:

Thank you all for the feedback.

@Illconceived Was Na.9781

I thought about adding some skins to the item list but the problem I had was that those were introduced later. And then it is very difficult to compare previous numbers. For example if I would include the Dreamhistle Weapons, I couldn’t include almost a year of gw2.

It is easy for me to generate those charts when I got a list of items. If somebody wants to write down a list with exactly the items and quantities, I would generate a chart and post it here:

item_id, item_name, quantity

That would be cool

I was just wondering, if you are working on on additional research buckets at the moment?

I think it would be interesting to see data for a bucket each for common, fine and rare mats, as well as cooking ingredients and gemstones (jeweling).
Or buckets related to profession crafting ingredients.