In this issue:
I. Commentary by Marge Sherald, CEO
II. Commentary by Denham Ward, Lead Developer of NeuroShell Trader
III. Save 20% on purchase of NeuroShell Trader Upgrades, Updates, and add-ons
IV. A Trader tip from Dennis Voigt, developer of BMJ Fractal Analysis Indicators
V. One way to stop this newsletter
***********************************************************
I. Commentary by Marge Sherald, CEO
A. Interview with Marc Chaikin
In the October 2012 issue of Technical Analysis of STOCKS & COMMODITIES there’s an interesting interview with Marc Chaikin about his 40-year Wall Street career. Near the beginning of his brokerage career, he talks about becoming interested in technical analysis in 1969 after a two year bull market was followed by a bear market that made it harder for him to make money. The article is an interesting perspective on how computers have changed trading, evolving from hand drawn charts to the current use of smart phones and tablets. Chaikin credited the hand drawn charts with giving him a deep understanding of the meaning behind the indicators. Chaikin is well known for the technical indicators he developed: Chaikin Accumulation/Distribution Oscillator (part of Advanced Indicator Set 1), Chaikin Volume Accumulation Percent and Chaikin Money Flow Persistency (part of Advanced Indicator Set 2) or the Money Flow Index from the Volume category in NeuroShell Trader’s indicators.
B. MESA91 Released
John Ehlers has completely revised every indicator that was included in the MESA9 add-on. The new indicators in MESA91 have a modification to the cycle length measurement that accounts for the fractal nature of the cycle amplitude (longer cycles have a greater swing). All of the indicators have a new algorithm that extracts the Dominant Cycle from the spectrum. Basically this is a Center-of-Gravity averaging rather than merely selecting the period with the highest amplitude. This approach gives a smoother Dominant Cycle and therefore better looking adaptive indicators that use it.
In addition, the MESA91 indicators allow you to fine tune the dominant cycle in each instrument. The indicators include parameters for Lowest Period (LP) and Highest Period (HP). The particular values of LP and HP might change from symbol to symbol and may be optimized.
A new indicator called MESA91_EvenBetterSine is included in the set and it’s designed to give you trading signals in only a half period of the dominant cycle, compared to the Sine indicator, which requires the full dominant cycle before it can determine market trend.
When I was revising the Trader example charts for these new indicators using updated data, I found it easier to build models that made a profit on out-of-sample data compared to the previous version. Like all model building, it took some tweaking and sometimes I changed instruments, but overall I found these new indicators a valuable addition to my toolbox.
MESA91 requires NeuroShell Trader 6.3 RC3, available from www.ward.net. If you purchased the MESA9 indicators in the past 4 months, the update is free. Otherwise, the cost is $250 for the update. If you are buying the MESA91 add-on for the first time, the cost is $399.20 on sale until Wednesday, October 10 (list price is $499). Use the following order link to receive the discount: www.ward.net/60.asp. The prices on the order form will reflect the discount.
If you feel you are entitled to a free update to MESA91, send an email with your request for a free update to sales@wardsystems.com
***********************************************************
II. Commentary by Denham Ward, Lead Developer of NeuroShell Trader
Position Sizing and Pyramiding
NeuroShell Trader Power User versions allow much greater control over entering and exiting positions due to the inclusion of various Position Sizing methods and Pyramiding in the trading strategy wizard. However, this also creates some confusion with the inevitable question of “What is the difference between position sizing and pyramiding?” The short answer is that Position Sizing determines the total number of dollars or shares to be bought or sold whenever a long or short position is taken and Pyramiding breaks that number up into multiple entries and exits. The long answer requires a little more explanation.
The simplest form of position sizing is saying “I want to trade 100 shares every time I go long or short” and simply placing a buy order for 100 shares when you enter and a sell order for 100 shares when you exit. As the underlying stock price goes up, you pay more dollars to buy 100 shares and as the underlying price goes down you pay less. While Fixed Size is included in all versions of NeuroShell Trader, the Power User version introduces position sizing methods which dynamically change the amount bought or sold based upon current account equity, trade risk, history of winning/losing trades, current volatility and historical drawdowns. Simpler techniques like Percent of Account, trades a fixed fraction of your brokerage account balance and allows a winning system to grow exponentially by trading more as your account balance climbs and trading less if your account balance drops. Then there are even more complicated techniques like Volatility Risk which buys the number of shares that limits current Average True Range volatility to a specified fraction of account equity per trade. A full list and brief explanation of the Advanced Position Sizing techniques can be found in the NeuroShell Trader help file. For an in depth discussion of the techniques, we suggested you consult the books referenced in the help file, some of which devote entire chapters to explaining some of the more complex position sizing methods.
Once position sizing determines the total number of shares to be bought or sold, the next question is should we get in and out all at once or let Pyramiding further control market exposure by scaling in and out of positions with multiple buys and sells. For instance in a moving average crossover system, we could enter fully as soon as a crossover above occurs or we could reduce our exposure to whipsaws by using pyramiding to slowly build up the position as we grow more confident a trend is forming with each subsequent bar that the fast moving average is above the slow moving average. Likewise we could choose to exit all at once on a cross below or slowly scale out of a position as we get more confirmation that the trend is reversing. The NeuroShell Trader Power User version allows control over the pyramiding entry and exits separately, the max number of entries and exits to use for each position, whether subsequent entries and exits grows larger or smaller, and if the same condition causes a new entry/exit (for instance each MovAvg1 > MovAvg2) or if new conditions are required for each new entry/exit (for instance MovAvgCrossover, then separate conditions for price climbing 5%, 10% and 15% above the crossover point).
Position Sizing and Pyramiding can be used separately or combined together to create systems. Add in the NeuroShell Trader Power User’s ability to optimize the Position Sizing and Pyramiding parameters and you end up with an extremely powerful tool for creating trading systems with the optimal balance between growing profits and limiting market risk.
***********************************************************
III. Save 20% on purchase of NeuroShell Trader Upgrades and Updates, and add-ons
We’re offering a 20% discount on upgrades to the Trader Power User versions and updates to Release 6 of the Trader. We’re also offering a 20% discount on the WSG add-ons and MESA91.
These offers expire on Wednesday, October 10, 2012. Use the following order link to receive the discount: www.ward.net/60.asp. The prices on the order form will reflect the discount.
***********************************************************
IV. A Trader tip from Dennis Voigt, developer of BMJ Fractal Analysis Indicators
I have had mixed results using the position sizing optimization feature. Combining the position sizing optimization with normal trade optimization seems to result in overfitting in the optimization period with less than satisfactory results in the out-of-sample period.
So this is what I tried:
I optimized the E-Mini futures with a fixed size of 1 contract each. Once that optimization completed, I checked the detailed analysis and trading rules to see what parameters were found by the optimizer. I copied the trading strategy and pasted it back so I now have two identical trading strategies on the screen. I went back into the second (copied) trading strategy, navigated to the trading rules and fixed the parameters by inputting the values from the trading rules screen. I recalculated the strategy to make sure that the results are the same as the first optimized trading strategy. Next I clicked on modify trading strategy parameters and selected the sizing tab. I clicked off the fixed size and selected the fixed fractional with a parameter range of .01 to .05 and re-optimize. The result is a much better model with much higher returns in both the optimization period and the out-of-sample period. Interestingly, with this approach I am getting better results in the out-of-sample period than in the optimization period.
This approach is a two stage optimization. First we build a model that works with one contract. Next we expand the model to take advantage of position sizing. But the models are built sequentially rather than one model all at the same time.
Dennis Voigt
Bascomb Madison & James Software, LLC
15 Brickyard Road
P.O. Box 104
Essex Junction, VT 05452
Voice: (802) 879-3766
Fax: (802) 878-1709
info@bmjsoftware.com
http://www.bmjsoftware.com/
***********************************************************
V. One way to stop this newsletter
It is really easy. Just change your email address and don’t tell us.
***********************************************************