Several financial data sources I mentioned in my previous post offer yield data series. But yield data series are often unpractical as we want to know how much money we would make if we invest in some bond fund or index. There exists one simple way how to change yield data series into price data series.
Imagine yield data set with monthly granularity and following first few entries:
31. jan. 1926 2,5 %
28. feb. 1926 3,3 %
31. mar. 1926 3,7 %
We would now like to change this yield data set to price data set with desired duration. We can start with price data set entry 1 on 31.1.1926 and then calculate entry for price data set for february 1926 as:
(price entry on january 1926) * (1 + (yield entry on january 1926) / 12 + [ (yield entry on january 1926)-(yield entry on february 1926) ] * (index duration) )
So we could calculate price index for february 1926 as interest gain for one month and gain or loss caused by change in interest rates multipled by our index duration. We could imagine that yields from example are for 10 year interest rates (bond index will then have duration around 7). Then we would calculate our price data set entry as:
1 * (1 + 2,5%/12 + (2.5% - 3,3%)*7) = 0.946
It is very handy to have price data series as we could then use them as a proxy for bond funds in our money management and portfolio management calculations.
22 September 2010
15 September 2010
Financial world roulette
I explained, that it is not worth playing negative expectation game on roulette example. But we have also seen there is high probability to win small amount of money even in that negative expectation game.
Risk isn't worth it. But there exist money managers which do so on regular basis.
Imagine you are money manager for small money market mutual fund. You are fighting with your peers for customers in fearful competition. You have 1 000 000 $ in your fund and you earn 3% income on your investments after 364 days. At 31st of December you see that you are in the middle of the crowd.
So you will take 100 000$ from you fund and will go to some kind of financial casino. You will do one simple bet which have 90% chance for success and will pay you 10 000$ and 10% chance for failure - in which case you will lose 100 000$.
You will win with big probability and finish year with 4% performance which place you among top money managers for that year. People are happy, more of them will come to your fund and you see your assets under management increase to 10 000 000$. So you will do same trick next year and you are again star. Press calls you next Soros. Money are flowing and you are getting rich from management fees.
Do you see how investment stars are born? And how they often end? Do you see similarity to business world and company management?
As Warren Buffet once said - "You never know who is swimming naked until the tide goes out."
So we want to invest in mutual fund. Or hedge fund. Or some company via its stocks. What could we do if we do not want to invest with such money managers?
Know them! Know their characters, know their strategies, know how they invest. Look on their fingers and ask questions. Learn stuff they do and do our own risk due diligence.
Trust is something they must deserve.
Risk isn't worth it. But there exist money managers which do so on regular basis.
Imagine you are money manager for small money market mutual fund. You are fighting with your peers for customers in fearful competition. You have 1 000 000 $ in your fund and you earn 3% income on your investments after 364 days. At 31st of December you see that you are in the middle of the crowd.
So you will take 100 000$ from you fund and will go to some kind of financial casino. You will do one simple bet which have 90% chance for success and will pay you 10 000$ and 10% chance for failure - in which case you will lose 100 000$.
You will win with big probability and finish year with 4% performance which place you among top money managers for that year. People are happy, more of them will come to your fund and you see your assets under management increase to 10 000 000$. So you will do same trick next year and you are again star. Press calls you next Soros. Money are flowing and you are getting rich from management fees.
Do you see how investment stars are born? And how they often end? Do you see similarity to business world and company management?
As Warren Buffet once said - "You never know who is swimming naked until the tide goes out."
So we want to invest in mutual fund. Or hedge fund. Or some company via its stocks. What could we do if we do not want to invest with such money managers?
Know them! Know their characters, know their strategies, know how they invest. Look on their fingers and ask questions. Learn stuff they do and do our own risk due diligence.
Trust is something they must deserve.
10 September 2010
Roulette investment
Roulette is an perfect example of negative expactation game. House has an advantage in each game you play (2.7% in case of european roulette and 5,3% in case of american roulette).No matter which betting system or money management you want to use you will broke at the end with 100% certainity in case you will not stop playing at some moment. There is great article on wikipedia regarding popular martingale system which shows why it is imposible to win in the long run even with this system.
But there is one way how it is posible to earn moeny even if it doesn't make sense to play roulette. The thery is simple - you need to have absorbing barier. It is posible to come to casino with 100$ and realistically expect to win for example 20$ - in case you will put al of your 100$ (or main part of it) in risk for that 20$ (divide 100$ to 5 stakes and put each on one Six Line). It is stil negative expectation game - if 1000 players come to casino then casino will get their part of the profit as around 18,9% of players will lose their 100$ and 81,1% of players will leave with 120$ (which yield profit 18.9% * 100 * 1000 - 81.1% * 20 * 1000 = 2 680$ for casino). But nearly 81% of players finished with profit.
Does it make sense to engage is such activity. Not at all. You risked 100% of you money to get 20%. Risk for doing so isnt worth the outcome. Your probability for winning 20$ is 81,1% (30/37 in european roulette with one 0) and your probability for losing 100$ is 18,9%. What is expected outcome?
81.1% * 20 - 18,9%*100 ~ -2,68$
So it is negative expectation game and longer you will play it more probable it is that you will broke. But it is possible to gain positive revard on short time frames.
So there is a posibility to gain some money on negative expectation game in short periods of time simply due to a luck. Do you see here some paralel between financial world and gambling world?
But there is one way how it is posible to earn moeny even if it doesn't make sense to play roulette. The thery is simple - you need to have absorbing barier. It is posible to come to casino with 100$ and realistically expect to win for example 20$ - in case you will put al of your 100$ (or main part of it) in risk for that 20$ (divide 100$ to 5 stakes and put each on one Six Line). It is stil negative expectation game - if 1000 players come to casino then casino will get their part of the profit as around 18,9% of players will lose their 100$ and 81,1% of players will leave with 120$ (which yield profit 18.9% * 100 * 1000 - 81.1% * 20 * 1000 = 2 680$ for casino). But nearly 81% of players finished with profit.
Does it make sense to engage is such activity. Not at all. You risked 100% of you money to get 20%. Risk for doing so isnt worth the outcome. Your probability for winning 20$ is 81,1% (30/37 in european roulette with one 0) and your probability for losing 100$ is 18,9%. What is expected outcome?
81.1% * 20 - 18,9%*100 ~ -2,68$
So it is negative expectation game and longer you will play it more probable it is that you will broke. But it is possible to gain positive revard on short time frames.
So there is a posibility to gain some money on negative expectation game in short periods of time simply due to a luck. Do you see here some paralel between financial world and gambling world?
09 September 2010
Risk premium or asset class?
Lets say I have a company which buys and sells bread (wholesale). This company is traded on stock exchange and we consider this company as a part of equity asset class. What does it exactly mean - equity asset class? It means, that I am owner of the future gains of company.
So lets take a better look on what is that company in reality doing. That company is in reality doing arbitrage strategy (or spread strategy) based on price difference between two parts of country (buys bread in one part of the contry in big quantity and sells it to the public in other part of the country in smaller quantities). What this company does is that it trades some underlaying comodity (corn in this example) - as an allegory. And this company is mostly sensible to price of corn.
Now lets say we have trading algorithm which trades gold. Algorithm is trying to exploit some market ineffeciency on gold market and lets call it "ABX Alpha Gold" as an example. Company which is mining gold would be sensitive mostly to same market risks as our "ABX Alpha Gold".
Algorithm and company are often interchangable components of same machine - our portfolio. What is important is how are companies/algorithms sensitive to change of some input parameter (price of some raw material, change in interest rates, S&P index drawdown etc. etc.). Input parameters are also sensitive to other input parameters (price of gold could be sensitive to change in CPI as an example).
For us - portfolio managers who manage our own portfolios - is important to understand all of the risks in those companies/algorithms and understand what is source of their income. We could often pack them together in one group. And those groups don't have to be called asset classes. Beter name for those groups would be risk premiums then asset classes.
Algorithm/company could be part of more risk premium groups. But we would try to find most characteristic springs of several important risk premiums and work only with them. Then we would apply money management and pack those risk premiums together to one optimalized portfolio.
So lets take a better look on what is that company in reality doing. That company is in reality doing arbitrage strategy (or spread strategy) based on price difference between two parts of country (buys bread in one part of the contry in big quantity and sells it to the public in other part of the country in smaller quantities). What this company does is that it trades some underlaying comodity (corn in this example) - as an allegory. And this company is mostly sensible to price of corn.
Now lets say we have trading algorithm which trades gold. Algorithm is trying to exploit some market ineffeciency on gold market and lets call it "ABX Alpha Gold" as an example. Company which is mining gold would be sensitive mostly to same market risks as our "ABX Alpha Gold".
Algorithm and company are often interchangable components of same machine - our portfolio. What is important is how are companies/algorithms sensitive to change of some input parameter (price of some raw material, change in interest rates, S&P index drawdown etc. etc.). Input parameters are also sensitive to other input parameters (price of gold could be sensitive to change in CPI as an example).
For us - portfolio managers who manage our own portfolios - is important to understand all of the risks in those companies/algorithms and understand what is source of their income. We could often pack them together in one group. And those groups don't have to be called asset classes. Beter name for those groups would be risk premiums then asset classes.
Algorithm/company could be part of more risk premium groups. But we would try to find most characteristic springs of several important risk premiums and work only with them. Then we would apply money management and pack those risk premiums together to one optimalized portfolio.