chalbersma (OP)
|
|
August 18, 2013, 10:44:59 PM Last edit: August 28, 2013, 02:59:09 AM by chalbersma |
|
Latest News: Website is Currently Not as it was. I'm Migrating it to Wordpress (already looks much better). Current index can be grabbed from http://index.mwguy.us/data/mwgi_latest.json------------------- Announcing the MWGI or Mid West Guy Index. This is a Dow like index. It's still in it's alpha stages and the index itself is subject to change. Looking for feedback on the index (especially it's contents) and feedback on the concept in general. I've got a long todo list still but feel free to check it out. The Index: MWGICurrent Index Makeup: AM1 (HAVK) LTC-GLOBAL (LTCG) BASIC-MINING (BTCT) S.MPOE-PT (BTCT) S.BBET (BTCT) CRYPTO-TRADE (BTCT) BITVPS (BTCT) ESECURITYSABTC (BTCT) COGNITIVE (BTCT) CIPHERMINE (LTCG) OPCU (LTCG) TABITA (LTCG) ART (LTCG) S.WIHEE (LTCG) HIM (HAVK) KCIM (HAVK) S.MG (HAVK) VTX (HAVK) NASTY-PT (BTCT) <OPEN> - <OPEN> - <OPEN> - <OPEN> - <OPEN>
|
|
|
|
Deprived
|
|
August 18, 2013, 10:52:02 PM |
|
Would help if you told us what the concept is that you want feedback on.
I'm struggling to find any purpose behind an index of various securities - some of which are denominated in BTC, others in LTC, others in USD and one in EUR. Unless you take steps to remove the impact of exchange-rate movements such an average would in no way reflect anything to do with the behaviour of securities.
And why include pass-throughs to securities listed on MPEX? If you include those then include the real securities themselves (i.e. MPEx prices).
You need to explain what the index is intended to mean/represent before anyone can usefully comment how it does/doesn't achieve that.
|
|
|
|
Rannasha
|
|
August 18, 2013, 10:59:12 PM |
|
All securities have weight 1? According to the website, the index is calculated by simply adding shareprices. That seems weird.
|
|
|
|
chalbersma (OP)
|
|
August 18, 2013, 11:11:18 PM Last edit: August 18, 2013, 11:25:00 PM by chalbersma |
|
Would help if you told us what the concept is that you want feedback on.
I'm struggling to find any purpose behind an index of various securities - some of which are denominated in BTC, others in LTC, others in USD and one in EUR. Unless you take steps to remove the impact of exchange-rate movements such an average would in no way reflect anything to do with the behaviour of securities.
And why include pass-throughs to securities listed on MPEX? If you include those then include the real securities themselves (i.e. MPEx prices).
You need to explain what the index is intended to mean/represent before anyone can usefully comment how it does/doesn't achieve that.
In terms of concept looking for feedback on the included stocks including ideas of new indices that people want to see and of course the calculations themselves. Right now I simply add up all the stock prices, should it be an average of them like the DOW or take into account market share like some other indices? I like my method as an alpha but come beta I want to have the method that will provide people the best representation of the market. There may not be a good purpose to stock indexes. But things like the DOW or the S&P can give one a feel for how the market as a whole is doing. I'm not to familiar with MPEX. I was looking for an api to pull data from it but I couldn't seem to find one. I'd rather not scrape the website to grab the stock price. That doesn't seem like a best practice. If they have an api and it's just not documented please tell me . That's kinda the makeup of all the passthroughs on my list. Bitfunder hasn't launched their API yet (according to their website). I'll in all likelihood be adding bitfunder assets when they launch it. I didn't see anything about a planned MPEX api though so I went with the passthroughs there. Index ideally will provide a general state on where the bitcoin market is at any given time.
|
|
|
|
chalbersma (OP)
|
|
August 18, 2013, 11:13:09 PM |
|
All securities have weight 1? According to the website, the index is calculated by simply adding shareprices. That seems weird.
That's how I've put it together to start off with. There's definitely other methods and if you think one is best please please tell me. The one that hands the most utility to people is the ideal one. The only way to figure out which method is best is to ask people what's best.
|
|
|
|
Addz
Newbie
Offline
Activity: 34
Merit: 0
|
|
August 20, 2013, 02:43:10 AM |
|
Id love it if you added some of the mining bonds from LTC global such as buy a hash or LTC miner
|
|
|
|
chalbersma (OP)
|
|
August 20, 2013, 03:37:36 AM |
|
Id love it if you added some of the mining bonds from LTC global such as buy a hash or LTC miner
A Bond Index is definitely one of the considerations for indices (along with a fund index, and a miner index).
|
|
|
|
Rannasha
|
|
August 20, 2013, 05:26:11 AM |
|
All securities have weight 1? According to the website, the index is calculated by simply adding shareprices. That seems weird.
That's how I've put it together to start off with. There's definitely other methods and if you think one is best please please tell me. The one that hands the most utility to people is the ideal one. The only way to figure out which method is best is to ask people what's best. I would definitely opt for a market-capitalization-weighted index. Right now, prices on BTCT range from well below a bit-cent (Labcoin, ActM) to several coins (AM). With price-weighting, even a doubling of, for example, the Labcoin price will move the index considerably less than 1%, while a 2% change in AM-PT price probably would achieve this.
|
|
|
|
chalbersma (OP)
|
|
August 20, 2013, 01:00:26 PM |
|
All securities have weight 1? According to the website, the index is calculated by simply adding shareprices. That seems weird.
That's how I've put it together to start off with. There's definitely other methods and if you think one is best please please tell me. The one that hands the most utility to people is the ideal one. The only way to figure out which method is best is to ask people what's best. I would definitely opt for a market-capitalization-weighted index. Right now, prices on BTCT range from well below a bit-cent (Labcoin, ActM) to several coins (AM). With price-weighting, even a doubling of, for example, the Labcoin price will move the index considerably less than 1%, while a 2% change in AM-PT price probably would achieve this. X Post from Reddit- Ya right now I do. For when I move to BETA I'm considering four methods: 1) Avg Top 25 Stock Price 2) 1% Model, Add together 1% market cap for each of the 25. 3) I like the S&P 500 Method ( http://www.investopedia.com/ask/answers/05/sp500calculation.asp). 4) Add them up. It's easy on the server! Of course I'd like some feed-back on this. I'm keeping historical data atm. I'm looking to finalize my data structures to support multiple indices more efficiently and then I'll start collecting and maintaining historical data.
|
|
|
|
Rannasha
|
|
August 20, 2013, 01:44:36 PM |
|
All securities have weight 1? According to the website, the index is calculated by simply adding shareprices. That seems weird.
That's how I've put it together to start off with. There's definitely other methods and if you think one is best please please tell me. The one that hands the most utility to people is the ideal one. The only way to figure out which method is best is to ask people what's best. I would definitely opt for a market-capitalization-weighted index. Right now, prices on BTCT range from well below a bit-cent (Labcoin, ActM) to several coins (AM). With price-weighting, even a doubling of, for example, the Labcoin price will move the index considerably less than 1%, while a 2% change in AM-PT price probably would achieve this. 1) Avg Top 25 Stock Price 2) 1% Model, Add together 1% market cap for each of the 25. 3) I like the S&P 500 Method ( http://www.investopedia.com/ask/answers/05/sp500calculation.asp). 4) Add them up. It's easy on the server! 2 and 3 are the same (other than that 2 limits to the top 25). They may give another value, but they'll always be proportional to eachother. 1 and 4 are the same as well, with the exception that you limit it to the top 25. Essentially there are 3 easy ways to compute an index (and infinitely many more complex variations): 1. Add prices together. Super-simple, but a very poor representation of what is going on in the market, since stocks with a high base price completely outweigh stocks with a low base price, independent of the size of the company behind the stocks. 2. Add prices together, with prices of each stock normalized to their starting price. Basically set a starting time for the index and evaluate the prices relative to the price at the starting time. Advantage of this method over the previous one is that the base price of the stock doesn't matter, only its change since the starting time of the index. 3. Add market caps together (the S&P 500 method). This I consider the most informative index-method, as it weighs large companies more than small companies. With each method, you can divide or multiply the final index by some constant to obtain a nicer number. This doesn't change the functionality of the method.
|
|
|
|
chalbersma (OP)
|
|
August 20, 2013, 06:56:24 PM |
|
All securities have weight 1? According to the website, the index is calculated by simply adding shareprices. That seems weird.
That's how I've put it together to start off with. There's definitely other methods and if you think one is best please please tell me. The one that hands the most utility to people is the ideal one. The only way to figure out which method is best is to ask people what's best. I would definitely opt for a market-capitalization-weighted index. Right now, prices on BTCT range from well below a bit-cent (Labcoin, ActM) to several coins (AM). With price-weighting, even a doubling of, for example, the Labcoin price will move the index considerably less than 1%, while a 2% change in AM-PT price probably would achieve this. 1) Avg Top 25 Stock Price 2) 1% Model, Add together 1% market cap for each of the 25. 3) I like the S&P 500 Method ( http://www.investopedia.com/ask/answers/05/sp500calculation.asp). 4) Add them up. It's easy on the server! 2 and 3 are the same (other than that 2 limits to the top 25). They may give another value, but they'll always be proportional to eachother. 1 and 4 are the same as well, with the exception that you limit it to the top 25. Essentially there are 3 easy ways to compute an index (and infinitely many more complex variations): 1. Add prices together. Super-simple, but a very poor representation of what is going on in the market, since stocks with a high base price completely outweigh stocks with a low base price, independent of the size of the company behind the stocks. 2. Add prices together, with prices of each stock normalized to their starting price. Basically set a starting time for the index and evaluate the prices relative to the price at the starting time. Advantage of this method over the previous one is that the base price of the stock doesn't matter, only its change since the starting time of the index. 3. Add market caps together (the S&P 500 method). This I consider the most informative index-method, as it weighs large companies more than small companies. With each method, you can divide or multiply the final index by some constant to obtain a nicer number. This doesn't change the functionality of the method. Thanks for the advice.
|
|
|
|
dexX7
Legendary
Offline
Activity: 1106
Merit: 1026
|
|
August 20, 2013, 07:31:38 PM |
|
Hi chalbersma, I'd favor a capitalization-weighted index, but instead of using last ticks, it might be a good idea to use volume-weighted average prices to buffer low volume spikes. What do you think about.. - One index for each exchange or one which combines them all? - Exclude company held shares? What could be used to determine the freefloat? For some stocks it's public knowledge, for others not.. - Any idea how to recover data about outstanding shares in the past? Bitfunder maintains shareholder lists and the number of outstanding shares on BTC-TC is included in the dividends, but all in all it's still incomplete.
|
|
|
|
chalbersma (OP)
|
|
August 20, 2013, 08:54:28 PM |
|
Hi chalbersma, I'd favor a capitalization-weighted index, but instead of using last ticks, it might be a good idea to use volume-weighted average prices to buffer low volume spikes. What do you think about.. - One index for each exchange or one which combines them all? - Exclude company held shares? What could be used to determine the freefloat? For some stocks it's public knowledge, for others not.. - Any idea how to recover data about outstanding shares in the past? Bitfunder maintains shareholder lists and the number of outstanding shares on BTC-TC is included in the dividends, but all in all it's still incomplete. I think a capitalization-weighted index is probably a good idea based on feedback at this point. Volume-weighted averages depress price spikes and while to a point that's good. It may not be reflective of the market. A second index that takes into account volume could definitely be considered. Initially I think I'll do a combine index. I've been designing the infrastructure with multiple indices in mind so a HAVK index or a LTCG or a "Burnside" index is definitely something I'd like to place there. Excluding company held share's may not be the best idea since most don't seem to disclose exactly how much is company held and it sort of hard to define. I could ask stocks to report it but that seems like a version 2 idea. Also I'm not sure how to handle Nasty mining and Asic Miner. Their share's aren't on the markets for the most part they run a custom platform. Should I count their non-market shares as part of the market cap or should I stick to just the share's that have been placed on the markets. In that case also do I count just one market or do I treat them all equally? Questions I'm still wrestling with.
|
|
|
|
|
|