In this issue:
I. Forecasting sales of high volume consumer goods
II. Commentary by Steve Ward
III. Sum Since (date) Indicator
IV. “Cocoon” trading signals top 80% accuracy
V. Golf handicapping with GeneHunter
VI. One way to stop this newsletter
*******************************************************
I. Forecasting sales of high volume consumer goods
As a consultant to Nestlé Italy, Filippo Vasta from Mbym s.r.l. developed a GeneHunter application that is an adaptive exponential smoothing algorithm for forecasting sales of high volume consumer goods such as chocolate bars and cookies. His model exceeded 96 percent accuracy for the year, with average monthly accuracy for the twelve coming months ranging between 80 and 95 percent.
Vasta used GeneHunter in an Excel spreadsheet to develop the application. According to Vasta, “the innovation comes from searching for seasonal adjustments, the initial value of the time series, and the promotional plan.” His model separates the time series into a baseline component derived from trends and seasonality and an incremental sales component.
Vasta says that the user does not need to make any assumptions about promotional efficiency as this is calculated by the model as a result of and in the context of the calculation of all other parameters.
“Once all the data is calculated, the forecast is made by projecting the baseline and calculating the future incremental sales based on assumptions for the promoplan for the coming periods, i.e., the proportion of customers interested in the promo each month for the coming year,” according to Vasta. He added that the forecast is clearly affected by the accuracy of the data.
Vasta said the same algorithm can be used for optimizing promo spending plans over a range of SKU’s (Stock Keeping Units) within a product family.
For more information about this forecasting model, please contact Filippo Vasta, MbyM s.r.l., at in**@mb**.it.
*******************************************************
II. Commentary by Steve Ward
Ever wished NeuroShell Trader Pro had an indicator you really want? A while back I told you about a neat little programming language called IBASIC PRO (www.pyxia.com), and how easily it could be used to program indicators for NeuroShell Trader. At that time it was only about $70, but the price has now come down to $49.99! You can’t afford NOT to buy it at that price. It is very easy to program, and we already have an example on www.ward.net.
Even you C and C++ programmers might like IBASIC because you can use C operations like ++ and +=
In addition, we have recently made a useful indicator with it (see article III in this issue), and the source code is being provided.
If you ever programmed in BASIC or ANY language in college or after, you can easily program indicators with IBASIC PRO. Get the compiler and get rolling – we’ll help you.
*******************************************************
III. Sum Since (date) Indicator for NeuroShell Trader
We recently had occasion to want to sum up a time series, not in a window or from the beginning of the chart (we already have indicators for those), but since a certain date. Our need was to sum up the daily returns from a trading strategy we wrote as a series of indicators instead of using the Trading Strategy Wizard. We wanted the total only for an out-of-sample period starting after we had built the indicator. No doubt you will think of other uses of this indicator, so we are making it available on www.ward.net in the Tips and Techniques section. The IBASIC source code can be downloaded too for those who want to modify or learn from it. The logic itself wasn’t hard to write, but figuring out how to convert dates to internal computer Julian dates took a little time.
*******************************************************
IV. “Cocoon” trading signals top 80% accuracy
Kenneth Branch has taken a different approach to forecasting stocks and indexes by looking for trades in indexes and stocks while they are in the “cocoon” stage. Branch says the term cocoon represents the underlying forces that develop within the markets. “It is not apparent to most of these participants, because they are focused on the known forces, which are moving the market at that time. In this cocoon stage, there is a large supply of any given security that may be easily purchased at low prices before they become known to most of the market.” His forecasts are more than 80% accurate in predicting movements in the Dow, S&P 500, NASDAQ, etc.
Branch has also taken a different approach to becoming a trader. He began studying the markets while in federal prison. He entered prison at age 18 on drug related charges. The fact that he was in a federal prison was his first lucky break. Unlike many state prisons, federal Inmates can receive newspapers and magazines. Branch noticed several inmates had subscriptions to the Wall Street Journal and were immersed in studying the markets. As Branch became more involved in the markets himself, he studied the book “Stan Weinstein’s Secrets for Profiting in Bull and Bear Markets.” He manually charted stocks and indexes for six years (he still has the charts). This led to his ability to recognize patterns in different indexes and stocks and he subsequently developed daily trading models based on indicators.
In 2000 he set up a trading account on the outside through a friend. Branch continued to develop his models until his release from prison at age 30. He then started his company called Cocoon Trade, Inc. based in Lanham, Maryland.
Branch applauded the NeuroShell Trader’s ability to build and backtest trading systems against past market data, in any time interval chosen. He quickly expanded the daily models developed on his manual charts to intraday and weekly forecasts. “The primary focus is on the major market indices, with an understanding that, statistically, any given stock moves along with the broader market most of the time. This approach simplifies our analysis a great deal,” said Branch.
His system called Buttafli9 is the result of over eight years of in-depth analysis of the correlations that exist between the different markets. “These correlations remain intact until the underlying forces (cocoon) move to bring the relationships out of whack, thus signaling a possible reversible move,” according to Branch.
Branch said his company offers clients forecasts of how the markets will move in different time intervals. He has developed forecasts for 14 indexes and more than 40 stocks, along with E-Minis, the large S&P contract, and ETF’s such as SPY and QQQQ. He has two basic models that he applies. The first model is a ratio analysis that measures the relationships between different indexes and stocks. The second model looks at momentum in an attempt to measure the strength or weakness of a move.
His models generate a combined 34 – 45 signals a month from his intraday, daily, and weekly charts. For his intraday charts he uses hourly bars because he believes smaller time increments have too much noise. These trades usually last about two hours from entry to exit. His daily trades can take up to three days but most exit the next day. His weekly charts are used for forecasting longer term trends.
Branch spends approximately 12 hours each day on his models. He gets up at 5 a.m. to check the overnight futures trading and analyze how the market will open. He continues his analysis throughout the day and after the markets close. His web site www.CocoonTrade.com offers signals for indexes and some stocks. If a potential client would like signals for a particular stock or index, he/she can send Branch the name of the stock and Branch will respond if his formulas will work for that particular stock.
Branch offers potential clients a free two week trial to view the signals posted on his web site. For more information call him at 301-484-1228 or send email to in**@co*********.com
*******************************************************
V. Golf handicapping with GeneHunter
One of our users, Brad Stern of AIS Futures Management of Wilton, Connecticut, has come up with a clever use of GeneHunter, one we wish we had thought of when we were building GeneHunter examples. He built a system to create foursomes from groups of golfers such that all foursomes would have about the same average handicap.
Brad is a trader who utilizes NeuroShell Trader Professional and GeneHunter, but recently he found a way to use his artificial intelligence skills to help his golf club. Brad explains, “At every event at my club and most other clubs, they make groups of four. They want them to be even so they divide the players into 4 quartiles by handicap, then randomly pick one player from each category for each team. Obviously, one group could get the best player from each category and one group the worst four. As there are prizes and betting at events, it would be much fairer to make groups with exactly equal averages of handicaps but also keeping a player from each quartile so that one group doesn’t have 2 great players and 2 poor players while another group has 4 mediocre players.”
So Brad built a GeneHunter application to solve the problem. He gave each golfer a number and used an enumerated chromosome to represent them. GeneHunter rearranges the golfers into foursomes. The fitness function, the standard deviation of the averages of foursome’s handicaps, is minimized. There are two soft constraints which work together to assure that each foursome contains one member from each of the four groups.
Our thanks to Brad for making his spreadsheet available. GeneHunter users will find it on www.ward.net under Tips from GeneHunter Users.
*******************************************************
VI. One way to stop this newsletter from coming
It is really easy. Just change your email address and don’t tell us.
*******************************************************