-------------
See this DanielsAg has last 7 years actual trade records for Ag ( corn, wheat ) , examining these may give some clues
http://www.danielsag.com/performance-history/
commoditity news letters: MRCI is here , so other letters are also seems credible. if we combine 'Guru bargain' concept ( proven historic trades offering at lower prices ) with above Above news letters we can get good returns.
example: 1/27/2010 crude OIL is down from 82 to 73 for last 10 days. MRCI sesional said trade for 1/12 to 3/15 for CRUDE OIL should give good return, since for year 2010 crude oil dropped to -50% of avg. profit of 'the Sesanoal' , I think it became 'Sesonal bargain trade'
( have Amibroker exploration based on seasonal Trade data:
- criteria: commodities that dropped to 30-50% of the the avarage profit for the seasonal based on historic seasonal return.
-------------
By Kurt Sakaeda
From Secrets of The World Cup Advisors, Edited by Chuck Frank & Patricia Crisafulli, Published by Traders' Library, www.traderslibrary.com
asr: this guy is Data analyst as mentioned so IT guy, so used my SQL , he may make 'Blue Book' available as service find out ...
---------
see other STOCK seasonal posted as TS code from this link
http://eminiforecaster.com/blog/672.php
asr: may be this eminiforecaster guy may be using ( developing as he said ) the method of combination that is a) his TS method b) Kurt Sakaeda method
----------
Although we can never get to an absolute assurance of our potential extremes in trading, we can accurately define the probabilities. The following is a rough outline of how to read them:
When the elements of a data set are pretty tightly grouped together forming a steep bell-shaped curve, the standard deviation is small.
When the elements are widely distributed forming a flatter bell curve, the standard deviation is larger.
To entice us to enter into a position, the standard deviation of a potential trade must be in proportion to the amount we expect to gain; if we are looking at a modest average profit of $500 and a large standard deviation of $10,000, the expected profit does not justify to exposure to extreme loss.
Ideally, we are looking for a high average price, a positive median profit, and a relatively low standard deviation. Typically, I find one standard deviation of up to five times the average profit to be in an acceptable range. I suggest bypassing trades carrying a standard deviation of five or more times the size of the average profit.
Average profit = $1,000
Standard Deviation = $2,500
Outcomes within one standard deviation:
Profit up to $3,500
Loss up to $1,500
A trade with a $1,000 average profit may typically carry a standard deviation — the potential profit or loss beyond the average — of $2,500. This means that in slightly more than two-thirds of all cases, it can be expected that the result of entering that trade will fall between a profit of $3,500 and a loss of $1,500.
Average profit = $10,000
Standard Deviation = $20,000
Outcomes within one standard deviation:
Profit up to $30,000
Loss up to $10,00
If another trade carries an average gain of $10,000 but a standard deviation of $20,000, the potential exists for a gain of $30,000 or a loss of $10,000 in approximately 68% of all outcomes. This means that in approximately 32% of all outcomes, the gain or loss could exceed our projections.
My Seasonal Method
My model calculates an average price for every day of the year in a given commodity contract (i.e. March corn or May silver), minus the days immediately following contract expiration. This average price is determined using daily settlement prices going back as many years as available from the data vendor I use. (I purchase Bridge CRB data, but there are several reliable vendors to choose from.) Because any annual date (say May 28) will fall on a weekend and possibly a holiday as we move across many years, I assign weekend and holiday prices based on the interpolation (estimated value between two known values) of prices on the trading days surrounding the weekend or holiday. It is simply our best guess at what would have transpired on those days based on surrounding values. This smoothes out the rough edges in my calculations.
I also have to make allowance for the fact that some years do not produce a full 365 days of data. In its first year of trading, a contract may be launched on, say, May 15. In other instances, trading may have been suspended for prolonged periods of time due to exceptional circumstances; the 1996 copper trading scandal due to rouge trading of Sumitomo Corporation’s Yasuo Hamanaka comes to mind.
Furthermore, I do not analyze data for the period of days extending from expiration day to the end of that month. This data is not relevant to my method, as my studies suggest a maximum “hold time” of 11 months.
Why use the settlement price as an indicator? Actually, tracking the average price in the opening range would work just as well. I use settlement prices because they are represented by a single price and are readily available.
Once I have the average settlement price for each day of the year, I calculate the price difference between every possible pair of days. This comprehensive analysis of 132,860 combinations (365 x 364) takes only about two minutes on my computer for each commodity contract. Then I calculate average profit, median profit, and standard deviation for the trade represented by the absolute high and low. As I currently track approximately 280 contracts, the run time for this analysis in MySQL using AWK language is approximately nine hours. There are several commercially available programs that can be used to generate these studies; I use MySQL because it is convenient and free. The download is free, but you may have to purchase support and training to learn how to run these studies.
Alternatively, you could simply scan for the highest and the lowest average settlement price to determine the single trade with the greatest potential for each particular contract. I prefer the prior method, as this provides secondary and tertiary trade opportunities within each contract, but for the most part I also rely heavily on identifying the highest high and the lowest low.
There are, however, occasional instances in which the high and low do not represent the best trading opportunity. This occurs when a comparable high or low can be identified closer to the anticipated exit date. A good example of this occurs in my data for January natural gas:
Chart 5.1 - January Natural Gas
As you see on this chart, the vertical column represents price (in this case $2.55 to $2.95 per MM BTU) and the horizontal line represents the calendar days of the year. The chart tracks the average settlement price on each calendar day included for all months of available data. In this case, we are examining 12 months of data through 2002.
(As I do not enter into positions on the days immediately following contract expiration, my charts cover approximately 350 days of the year. For example, soybean meal contracts expire on the business day prior to the 15th of the month. Consequently, my July soybean meal chart will not include data for July 15–31.)
In January natural gas, the annual composite low occurs on February 3, and the annual composite high occurs on December 16. However, we can make a faster and safer trade by looking to enter long at a nearly comparable low price on July 22. We may not achieve quite as low of an entry price, but we will theoretically need to hold this trade less than half as long. That frees up margin money for other trades in the interim, and reduces the risk of outside events adversely affecting the normal seasonal trend. In this scenario, I’d rather face the prospect of making a projected profit of $3,498 over five months than $3,714 over 11 months.
No comments:
Post a Comment