[DCX] The Digital Currency Index Project

[Smidge]

I hereby want to make an announcement for a Digital Currency Index (WORK IN PROGRESS) that can be used to indicate the current state of the market, as benchmark etc.

Feel free to shoot holes in it as you desire by just replying, but please make sure you take a look at the discussion that has been going on in this thread. I am excited about any feedback and looking forward to developing this project beyond it's Alpha stages.

If anyone with more data, capabilities etc. wants to pick this up and take it to new heights, I would be happy to cooperate and have lots of ideas (sub-indices, certificates, options...). Please contact me.

Cheers,
Smidge

-------------------------

[DCX] The Digital Currency Index Project

http://dcx.smidge.com

What is it?

DCX is the attempt to create an index like the DAX, FTSE etc. that you know from real world markets for the digital currency community.

Why do we need this?

Primarily as a benchmark. You can compare the performance of a stock, fund or any other security to this index, which should be a broad indication of securities that make up most of the digital currency market. In it's final form, you could regard it as "The Market" and check how a security (or your own portfolio) has performed against it, just like in the real world markets.

A fund manager for example can announce their fund's NAV being up 10% in just a week, which sounds pretty good. If "The Market", meaning a roundup of all securities, is up 25%, then the fund's performance relative to the market is pretty bad.

What is it's value?

It started on Friday, August 2nd, 22:00 CET with a value of 100.00. You can check it's current value at http://dcx.smidge.com or via the API. It's already up a few points at the time of this post

Why 8 decimal places?

Because we're used to it
This might be too much and could change pretty soon.

How often does this update?

Currently "only" once per minute. It cannot be used for (professional) daytrading yet, which is also not its current purpose. There will be improvements on this soon though, so that you can speculate on short term movements.

There is not enough data in there for me to plan my next move!

It's been running a few hours! Just give it a few days or weeks to accumulate more data.

The current chart looks like a bunch of crap

I know. There will be improvements on this soon, please note that the project is in Alpha stage. The current focus is accumulating data rather than displaying it in fancy ways.

It is also displayed as a Candlestick Chart, which describes the high/low and the opening/close values of the current data period (1 hour) at the same time. There is a lot of documentation out there on how to read it in case it confuses you.

Plus, you can always create your own charts by using the API!

Why are there no funds in it?

A fund traditionally invests in other securities, which means a fund itself is a compilation of standalone securities. Including it would lead to dilution and/or overweight of some standalone securities.

Some securities are dominating the index

That is correct. This lies in the nature of the current market and is not a problem of the index itself, I would describe it as "If ASICMINER coughs, the market gets a cold". As the index is made for the very long run, we will hopefully see worthy competition arise that evens out this imbalance.

How does it work?

The index is calculated using the Laspeyres method. It is also used to calculate the traditional indices like DAX, FTSE etc. The formula is described in detail below, plus here is a simplified example (S1-S3 being the securities included, price0 and price1 are their previous and current values, mkt cap is number of shares times the price):

Name	price0	price1	mkt cap
S1	40	40	200
S2	50	55	300
S3	60	60	400

The value of 103.19 means that if the securities would perform in the way described in the table, the index would rise by 3.19% in the given timeframe. In detail, security S2 has a market cap of 300 (1/3 of the total 900 in the index) and gains 10% in value. This pushes the index up by 3.19% when using the Laspeyres method.

In which intervals are price0 and price1 from the example above calculated?

The interval is set to daily, which means the deviation (not the index value!) is "reset" at 0:00 CET and it begins using the new reference point from then on (for the next day).

Is there an API?

A very basic one, yes:

Current value: http://dcx.smidge.com/ticker
History: http://dcx.smidge.com/ticker/history

How do the next steps & planning look like?

That depends very much on community input. On my todo list is currently:

• Get as much community input as possible
• Add more securities to get a broader, more realistic depiction
• Update more often to make daytrading possible/more effective (this is also a problem of the securities not being traded frequently enough)
• Update to better charting software
• Future music: Create sub incides based only on mining, funds etc.; Issue put/call options, create certificates etc.

-------------------------

Thanks 4 reading & please also take a look at my main project Smidge.Com - A virtual, actively managed, multi-asset digital currency fund: https://bitcointalk.org/index.php?topic=259772

[Smidge]

reserved

[Smidge]

reserved

[ThickAsThieves]

Good work!

[somestranger]

Very cool. I can't wait to see it when it gets a few weeks of data.

[Smidge]

Anyone have a realistic description of all relevant securities out there with their number of shares and market cap? I am currently only accumulating data from btct.co. The goal would be to create a waterproof composition of the index to really have it reflect and be regarded as "The Market".

[Smidge]

102.7 seems to be a hard resistance today! Let's make a snapshot of it for the archives...

[DrGregMulhauser]

The index is calculated using the Laspayres method. It is also used to calculate the traditional indices like DOW, DAX etc.

(Side note: Laspeyres -- not Laspayres -- method. Also, most major indices are market cap weighted, but the Dow is price weighted.)

Since the disadvantages of market cap weighted indexes are well known, have you considered an equal weighted index? Just a-wundrin'...

[Smidge]

Hi DrGregMulhauser,

The index is calculated using the Laspayres method. It is also used to calculate the traditional indices like DOW, DAX etc.

(Side note: Laspeyres -- not Laspayres -- method. Also, most major indices are market cap weighted, but the Dow is price weighted.)

Both fixed! Thank you.

Since the disadvantages of market cap weighted indexes are well known, have you considered an equal weighted index? Just a-wundrin'...

The basic question is, should the index describe the market 1:1 (market cap weighted) or should it regard all included stocks as equal, ignoring their market cap? I chose the previous for DCX, because I see the index like a portfolio of securities and that is how it should behave. As stated above, I think it is the best setup for the long run, where we ideally want to have several more or less equal sized stocks competing along lots of small ones.

Edit: One solution could be to set a maximum weight that an individual security can have in the index (in the DAX, that is 10%, but there are 30 securities in there altogether). I think 25% would be enough. Thoughts?

[DrGregMulhauser]

The basic question is, should the index describe the market 1:1 (market cap weighted) or should it regard all included stocks as equal...

There could be more than one basic question; one might be what sort of description would be most useful? For some people, equal weighted would probably be more useful, while for others some type of fundamental weighting might also be useful. For some ETFs, for example, it's earnings weightings that count.

...I see the index like a portfolio of securities and that is how it should behave...

I'm not sure I follow. Any index can behave like a portfolio of securities; there's more than one way to build a portfolio, and more than one way to have an index behave like a portfolio.

...long run, where we ideally want to have several more or less equal sized stocks competing along lots of small ones...

Why would that be ideal?

Edit: One solution could be to set a maximum weight that an individual security can have in the index (in the DAX, that is 10%, but there are 30 securities in there altogether). I think 25% would be enough. Thoughts?

Caps like that, although arbitrary, seem like at least one good way of helping an index provide more of a glimpse into something other than just its largest one or two constituents. If you're coming at it from the standpoint of how an index should behave, and you believe it should be market weighted, then a cap like that would be all wrong, but on the other hand it could make it more useful.

[chairforce1]

this seems pretty cool nice work

[Smidge]

(I've put this in the FAQ above)

Some input about variance and reference points:

<dexX7> smidge: in which intervals are p1 and p0 calculated? this seems to be critical, because if you do the calc.. well, let's say every new trade, the difference is very small (as only one number changes) and the result oscillates very close to 100
<smidge> dexX7 very good question, i currently plan on setting the interval to daily
<smidge> as you described, the variance would be too low
<dexX7> smidge: would you then take the end of day (or whatever interval) price of each asset or some averaged value?
<smidge> dex the last price of the day would be used as reference for the next

[Smidge]

The basic question is, should the index describe the market 1:1 (market cap weighted) or should it regard all included stocks as equal...

There could be more than one basic question; one might be what sort of description would be most useful? For some people, equal weighted would probably be more useful, while for others some type of fundamental weighting might also be useful. For some ETFs, for example, it's earnings weightings that count.

I agree that it should focus on usefulness for the community.

...I see the index like a portfolio of securities and that is how it should behave...

I'm not sure I follow. Any index can behave like a portfolio of securities; there's more than one way to build a portfolio, and more than one way to have an index behave like a portfolio.

What I mean by this is that individual securities should not behave like they have a "distorted" value because we equal weight them. A compromise could be capping the security's weight at a certain percentage.

...long run, where we ideally want to have several more or less equal sized stocks competing along lots of small ones...

Why would that be ideal?

Because the index would be more diversified and not tied to one or a few security's performance, creating less overall volatility.

Edit: One solution could be to set a maximum weight that an individual security can have in the index (in the DAX, that is 10%, but there are 30 securities in there altogether). I think 25% would be enough. Thoughts?

Caps like that, although arbitrary, seem like at least one good way of helping an index provide more of a glimpse into something other than just its largest one or two constituents. If you're coming at it from the standpoint of how an index should behave, and you believe it should be market weighted, then a cap like that would be all wrong, but on the other hand it could make it more useful.

Thanks a lot for your input. I think the ideal setup for the BTC world lies somewhere in between. I am looking forward to discussing this further down the road and making the necessary adjustments.

[Smidge]

Added CIPHERMINE to the index with a weight of currently 6.86% (about 542k LTC/15k BTC mkt cap)

https://www.litecoinglobal.com/security/CIPHERMINE

[dexX7]

Yay, I really support this.

I played a bit around with the Laspeyres formula and I'm not sure, if this is correct: I took BTCGARDEN, LABCOIN, ACTIVEMINING, COGNITIVE, ASICMINER-PT (no BASIC because split and simplicity...) and used the current number of issued shares as basis. This is not ideal, but I don't have data on historical numbers of issued shares and this is experimental anyway.

The share prices were grouped into 1 hour groups for each asset and I calculated the index output like this:

p₀ = BTCGARDEN price_n * 2000000 + ... + ASICMINER-PT 1 hour price_n * 26006
p₁ = BTCGARDEN price_{n +1} * 2000000 + ... + ASICMINER-PT 1 hour price_{n +1} * 26006

index value = p₁ / p₀ * 100.


Here are the results:

Daily, 1H groups with volume weighted average price


Daily, 1H groups with 1H closing price


Daily, 1H groups with line on close


So basically the Laspeyres formula can be used to track down changes in market cap and a steady value around 100 means that there is not much price fluctuation.



I also created market capitalization charts. Those represent simply the sum of the market capitalization of the above mentioned assets:

Daily combined market cap


Daily combined market cap with line on close



Brainstorming:

• Exclude company held shares
   
• Get historical outstanding share data
   
• Which assets should be included?
   
• What is a more realistic model for a representation of the market?
   
• Fetch data from other exchanges
   

I don't have a very financial background and to be honest, I'm not even sure, where this leads, but one thing is for sure: it's time for the next level! A general overview of the Bitcoin markets is only the beginning and this thread by Smidged is the initial experiment to accomplish this.

Any input and discussion is very welcomed.

[Smidge]

Hi dexX7,

Yay, I really support this.

I played a bit around with the Laspeyres formula and I'm not sure, if this is correct: I took BTCGARDEN, LABCOIN, ACTIVEMINING, COGNITIVE, ASICMINER-PT (no BASIC because split and simplicity...) and used the current number of issued shares as basis. This is not ideal, but I don't have data on historical numbers of issued shares and this is experimental anyway.

The share prices were grouped into 1 hour groups for each asset and I calculated the index output like this:

p₀ = BTCGARDEN price_n * 2000000 + ... + ASICMINER-PT 1 hour price_n * 26006
p₁ = BTCGARDEN price_{n +1} * 2000000 + ... + ASICMINER-PT 1 hour price_{n +1} * 26006

index value = p₁ / p₀ * 100.

Highly interesting. I need a more flexible charting engine (ClarkMoody style would be awesome), plus focus on getting historical data in there as a priority.

Brainstorming:

• Exclude company held shares
   
• Get historical outstanding share data
   
• Which assets should be included?
   
• What is a more realistic model for a representation of the market?
   
• Fetch data from other exchanges
   

I don't have a very financial background and to be honest, I'm not even sure, where this leads, but one thing is for sure: it's time for the next level! A general overview of the Bitcoin markets is only the beginning and this thread by Smidged is the initial experiment to accomplish this.

Any input and discussion is very welcomed.

I think historical data is a quick win, plus I received the suggestion to add Moving Averages.
Get data from other exchanges would increase the number of price fixings (=liquidity).

[Smidge]

Added change in percent to previous day's close (currently +1.01% to yesterday).

[EskimoBob]

Cool idea but not sure why are you calling it currency index?
It looks more like "The BTCT Mining Index"

[Smidge]

Cool idea but not sure why are you calling it currency index?
It looks more like "The BTCT Mining Index"

Hi EskimoBob,

Currently, it is more or less a Mining Index. But it is made up of the securities that reflect our market. This may change over time as new business models get implemented, for it is a very long term project.

[Smidge]

Up for almost 48 hours, nice recovery today with +2.86%.

The forces of the market at work... beautiful