In last week's issue, we pointed out Mike's achievements with Dakota R/T in our Gallery of Customer Successes.  This time we'll switch back to Dakota End-of-Day and take a look at what Bill has accomplished.  Using history since January, 2002, he hypothetically profited $5,400 on a $3,000 investment on the Rydex Velocity 100 fund, which, if I'm pushing the right buttons on my calculator, is about 38%/year simple return. His equity curve is about 93% of a straight line.  The top plot in the chart is price, the middle is his signal and the bottom is his equity curve.  Click the image to go to the Gallery of Customer Successes to see bigger images.

Summary:
Security: Rydex Velocity 100 (RYVYX)
Bot: User Modified Standard Bot
Gains: $5,400 in 4.75 years on $3,000 invested
Consistency: 93% straight equity curve
User: Bill V., Texas, using Dakota EOD Standard

What can YOU do with Dakota?
 

Tech Talk

BioComp Profit: Cook's Method of Indicator Optimization
In BioComp Profit, technical indicators are used as inputs to predictive models that profitably forecast a security's price movement in the future.  The models are essentially converting the indicators' movements into profitable signal movements.  There is a clear advantage if the indicators themselves are created to be profitable, that they "speak" in equity terms, and if they were used as signals, they would make money.  Now, if they did, then you would be done.  You would just use the indicator as your trading signal.  But finding an indicator that is tradable straight away is a real challenge and a rare occurrence.  However, if you have a handful of somewhat profitable indicators, then a model can merge them into a profitable signal.  Finding profitable indicators is where "Cook's Method" comes in.

Cook's Method searches through combinations of different indicator functions and parameter values (e.g., periods of an RSI) looking for the right permutations to maximize the equity performance of the result. These combinations of functions include such things as Change, Momentums, RSIs, ZScores, Normals, Slopes and various smoothing functions like SMA, XMA and AMA.  It is important to point out that this optimization must be done "in-sample" and validated "out-of-sample" to see if the equity behavior of the resulting indicator continues reasonably well. Also, optimizing input indicators on all data gives an unfair advantage to the model's signals and contaminates the model's out-of-sample testing as well.

Once you have found reasonably profitable input indicators, the result is clearly to your advantage.  The modeling process needs to do less "work" and your signals are typically well improved.  Many Profit users will swear by the results of this method.  It is so good, we considered making a massively parallel "Seti" style project to search for optimal indicators and some users have created their own "Cook's Method" search engines, including the "Profit Data Miner" and the "Klondyke Package".  We also took some Cook's Method results from Profit to create trading bots for Dakota...  Because it works.