On the subject of Keys prices in terms of Refined Metal

In Team Fortress 2, Mann Co. Supply Crate Key is an important currency in virtual item trading. We use Refined Metal to buy items of low value, and Keys to buy items of slightly higher value. It follows that an exchange market between two currencies exists.

This page examines the behavior of the market based on data collected from the past 6 months. The data consists of approximately 15,000 trades of various quantities. Theses data were collected through constant monitoring of a subset of 5000 users of TF2 Outpost, TF2 Trading Post, RAWR Trade Club and TF2-Trader. 

Fig. 1: Cached scatter plot of trades from 6 months’ timeframe (Click image to enlarge)

General observations

The following direct observations were found from the data:

1. The long-term trend is ever increasing.
2. Prices are "discrete", i.e. they usually sit exactly on 0.11 Refined (1 Scrap) lines, even in bulk deals.
3. There are 5 most traded prices throughout the summer of 2012, namely 2.33, 2.44, 2.50, 2.55, and 2.66.
4. Trades can happen at 5 different prices within 1 minute.
5. Standard Deviation of Keys in terms of Metal at any given time is 0.1666 Refined (1 Scrap + 1 Weapon)
6. Distribution is symmetric to the simple moving average.

Increasing trend. At which rate is it increasing?

The trend in summer appears to be linear. Using ordinary linear regression methods:

1. Price of Keys increases at 0.000667736751874752 Refined per day, or
2. Price of Keys increases at 0.004674157263123264 Refined per week, or
3. More neatly, Price of Keys increases at 0.02 Refined per month
4. Equivalently, Price of Keys increases at 0.12 Refined per 6 months

For instance, say Keys are at 2.66 Refined today, the annual rate of return of Keys in terms of Refined is 0.24/2.66 = 9% per year.

Do bulk trades have different average?

In general, no. By highlighting the trades with volume greater than 10 Keys in red, we can see that bulk trades conform to the simple moving average of all trades. This means that selling 20 Keys in a single trade will not earn them higher price than selling 20 Keys in 10 trades on average. The case is similar in buying 20 Keys in a single trade.

Data vs tf2spreadsheet.blogspot.com

We begin comparing the simple moving average and the mode (most traded price) from the data against the price range supplied by this price guide.

2012-09-16 Changed from "2.55" to "2.55-2.66" back and forth

Implications:

1. The pattern of trades on the day of 2012-09-16 as a result of expanded price range is no different than the patterns from 5 preceeding.
2. The market does not have dramatic reactions on expanded price range on price guide.

2012-09-30 Changed from "2.55-2.66" to "2.55" back and forth

Implications:

1. The pattern of trades on the two days as a result of contracted price range is no different than the patterns from 5 preceeding.
2. The market does not have dramatic reactions on contracted price range on price guide.

2012-10-02 Constant range of "2.55-2.66" for extended period

Implications:

1. Such range represents the two most traded price (2.55, 2.66) throughout October, thus the price check is accurate.
2. The third most traded price of 2.50 Refined (2 Keys for 5 Ref) still occurs.
3. At the end, the change from 2.55-2.66 to 2.66 is reasonable given that Simple Moving Average solidly crossed 2.60 Refined.

2012-11-02 Increasing of price from "2.66" to "2.88-3" during Halloween Event

Implications:

1. The first change on 2012-11-02, from 2.66 to 2.66-2.88 is an understandable preemptive move.
2. The second change on 2012-11-03, from 2.66-2.88 to 2.77-3, completely ignored the fact that vast amount of trades are still happening at 2.66. The moving average is at the lowest end of the range.
3. The last change on 2012-11-06, from 2.77-3 to 2.88-3, is a complete bias in favor of Keys owners. The moving average isn't even in the range.

Data vs backpack.tf

Backpack.tf only had 5 accepted suggestions of Keys price since September 2012.

2012-09-28 Changed from "2.55" to "2.50-2.66"

Implications:

1. This is basically adopting price ranges over average price.
2. The two prices are equivalent. There is no impact on the market.

2012-10-18 Changed from "2.50-2.66" to "2.55-2.66"

Implications:

1. The sudden exlusion of 2.50 Refined in the range is random. There is no explanation of why this needs to be done in October and not at other times before.

2012-10-28 Increasing of price from "2.55-2.66" to "2.66-2.88" during Halloween Event

Implications:

1. The first change on 2012-10-29, from 2.55-2.66 to 2.66, was based on the rising trend. The moving average was not 2.66 yet. This suggestion had extreme net downvotes.
2. The second change on 2012-10-30, from 2.66 to 2.66-2.77, reflects the two most traded price the next few days. A speculation that was accurate. This suggestion had minor net downvotes.
3. The last change on 2012-11-03, from 2.66-2.77 to 2.66-2.88, is a realization of the fact that the spread is larger than expected. This suggestion had minor net upvotes.

Tips for price checkers

1. Accept the fact that every item of value has price dispersion, i.e every item has price spread. For example, Earbuds will always have 1 Key spread because most people do not like to pay Key + Ref. Even the average is stable at 26 Keys, trades occur at 25 Keys and 26 Keys with equal probability between the two.
2. Accept the fact that price of each item has a central tendency, i.e. prices of different trades of the same item tends to the average. A more generalized statement is that if you know the "two most traded price" (2.66, 2.55) of a certain item, the "average of all trades" is always between 2.66 and 2.55, regardless of the third most traded price (2.50).
3. Actual trades show that TF2 Outpost is not overpriced. There are more trade advertisements of "extreme prices", but what they bought/sell for is not equal to their "advertised/claimed value".
4. Actual trades show that Unusual trade servers are not underpriced. The average recorded from servers are sometimes higher and sometimes lower than TF2 Outpost and Trading Post.
5. Power sellers and resellers form a good sample. They work hard to find the best deals. Thus you can find out the proper range just by looking at them.
6. Just because you never experienced something, does not mean that things don't happen. Tons of people sold keys at 2.66 Refined over 6 months ago and there were indeed such buyers who can afford.
7. The impact of price guide / price list making sudden changes to a certain item is overestimated.The general public does adjust their trading pattern according to such websites. Changing the value for 1-2 days has virtually no impact on the market.
8. Price (specficially equilibrium price) is determined by the market, not by opinions. The voting system for Keys did not work during the Halloween Event.

Interpretation of Volume

Volumes are supposed to be interpreted this way in elementary economics. However, given that volumes in the data can also be affected by the Steam API downtime, server load of trading sites and our own data-logging server, such implications from volume are not always reliable.

Conclusion

The Keys market is amazing with lots of trades and interactions. Long-term data gathering provens to be useful in many ways.

Transparency in prices is a key element for efficient markets.

Contact

base64 http://steamcommunity.com/id/base1024/

2012-11-09