Monthly Article
Topics for November
PAGE 1

This Issue
November 2000

Various Topics
Page 1

Tech Talk
Page 2

Market Statistics
Update &  IPO's
Page 3

Notice:
The views and information expressed in this document reflect the opinions and experience of the author Robert C. Pelletier.  Neither CSI nor the author undertake or intend to provide tax advice or trading advice in any market or endorse any outside individual or firm.  All recommendations are provided for their informational value only.  Readers should consult competent financial advisors or outside counsel before making any software purchase or investment decision.  CSI does not stand behind or endorse the products of any outside firms.

Copyright (c) 2000 Commodity Systems Inc. (CSI).  All rights are reserved.

Topics discussed in this month's journal.

 Holiday Schedule TSPE Tutorial

Holiday Schedule

CSI will be closed for voice communication on Thursday, November 23rd for the Thanksgiving holiday. US exchanges will be closed, but data from other exchanges will be available as usual. The CSI staff wishes you a Happy Thanksgiving!

TSPE Tutorial

TSPE differs from most analysis programs you may have encountered in two important ways: 1) it evaluates your trading system, not the market you wish to trade and; 2) its input data is your trading record, not market data. Therefore, in order to use TSPE, you must have a specific, logical trading approach to evaluate and you must have a track record for trading that approach, which can be either actual trading results or derived from a simulation.

TSPE makes a couple of important assumptions about your input data (profits and losses) which must be accommodated to produce valid results. Looking at the first assumption, the trading results you want to evaluate must not have been derived through what some technical analysts call "curve fitting." The term "curve fiting" comes from the mathematical procedure whereby one can fit an Nth degree polynomial through N+1 points such that the resulting equation will define a path that exactly visits all N points in a sample. Consider the straight line, which is a first degree (one parameter) equation. Two factors can define a straight line that will always pass through two given points. The straight line includes an origin (A), and a slope (B). An equation using the known factors A and B will always force a straight line through those two given points.

To satisfy the first assumption (requirement), the user should try to avoid any attempt to certify technical trading systems that use extensive repetitive searches of all possibilities to uncover the one best procedure that exactly matches the data. To a great extent, this restriction might remove many alternative schemes. Consequently, TSPE is designed to reverse some of the potentially exaggerated performance through the parameter-count adjustments and small-sample penalties discussed below.

The second assumption is that sample results derived from the trading system are independent. This means the results of each trade have absolutely nothing to do with the results of previous trades. For example, a system that doubles-up your position when certain profits or losses are attained would not be suitable for evaluation with TSPE. This is because the distribution of profits and losses is assumed to be random and inherently defined through the sample.

If your trading record complies with the above assumptions, you have what you need to input your trade-by-trade results. This tutorial example will guide you. If you don't have a suitable trading record, we suggest you develop one or ask your system developers for their experience so you can evaluate it with TSPE.

Launching TSPE

From your computer's START menu, select Programs, then Unfair Advantage, and finally click "Trading System Performance Evaluator." The TSPE main screen offers the version number, a parachutist achieving a safe landing and a policy statement from CSI encouraging use of TSPE to certify your trading system before venturing into circumstances where you may incur substantial risk.

Click the [Next] button to proceed to TSPE's "Import Trades" screen. This is where TSPE accepts a list of the profits and losses (P&L, with or without commission and slippage) derived from the simulated or actual results of your trading system. The trading results can be entered 1) manually, 2) through an ASCII file with the trade results shown on separate lines, or 3) as a TradeStation® System Report. In the example at hand, the manual method was used, so I clicked the [Manual] button. The other two options are covered in the on-line manual.

This takes you to the "Enter Data" screen that includes a "Profit/Loss" entry box at the right. This is where you will show the individual trade outcomes, with or without commission and slippage. Click in the first empty entry box below the "Profit/Loss" title and enter the dollar amount of the first profit or loss from your trading record. To include commission and slippage here, the typical ask-minus-bid spread in dollars and the commission should be subtracted from the trade-by-trade results entered. If commission and slippage are excluded, you can enter guidelines for these calculations later.

 Sample Track Record 300 5000 500 -900 2100 700 1800 1500 -2000 900 1100 -1300 1300

As you are entering data, you may review the summary statistics shown on the left side of the screen. These include total profit, average gain and loss, maximum drawdown and other details derived from your P&L record. Although this table tells a story about your input data series, it is not a projection of future performance. The information displayed here is dependent upon the sequence of trade results and is, therefore, not appropriate for evaluating profit potential. The differences between the summary statistics and TSPE's ultimate capital analysis results reveal through simulation the importance of TSPE's requirement for trade independence. The independence assumption is critical to the forthcoming evaluation, which will investigate exhaustive random orderings of your trade results forcing sobering reality into your input.

Contract Size Multiplier

The next screen has to do with the Contract Size Multiplier. It offers you a chance to use a TradeStation input file, a manual input file, or an ASCII input file that is not presented in dollar P&L terms. Here you should enter the value TSPE would need to multiply the P&L stock or contract's tradeable units to convert those P&L units into dollar terms. For example, if your trading record represents profits and losses in cents per bushel, as in a 5000 bushel corn contract, you would show a multiplier of 5000. This would make a 1 cent profit or loss into a 5000 cent profit or loss or \$50. Similarly, a 10 cent loss would represent a 50,000 cent (\$500) loss. The multiplier in this example would be 5000. If profits and losses are expressed in dollars, the default multiplier would remain at 1. Click [Next] to proceed.

Choose Commission

If you did not adjust your trade records to reflect commission and slippage costs, you may do so on the Choose Commission screen which appears next. Enter the average per-trade dollar commission expense (by modifying the default) and the nominal dollar slippage that you would expect to pay to overcome the ask-bid spread. Change the default entries to zero (0) if these were factored into the profit and loss statistics you entered. Click [Next] to continue.

Editing Sample Size and Parameter Count

The next screen reports upon the trade sample size and invites you to enter the parameter count from the trading system used to generate the trading record. The sample size affects the degradation imposed on the P&L statistics. Large samples tend to bolster the credibility of the profit statistics and small sample sizes tend to support degradation in performance. The sample size shown is the number of trades you reported in your profit-and-loss record. The default value is an accurate representation of the input data. Although you can edit the sample size, we recommend against doing so.

The parameter count is important because it describes the level of curve fitting that may have been adopted to achieve the trade-by-trade results. The more parameters or constraints employed by a trading system, the less likely the proposed system will perform in actual trading. Your TSPE evaluation will demonstrate that you should choose a system that consumes minimal parameters. A system with just one (1) parameter would be far more likely to rate well with TSPE than a system with several more. Enter a parameter count that describes your analysis approach.

Correcting for Parameter Count and Sample Size

TSPE accounts for hindsight bias by replacing every profit shown in the trading record with the profit value divided by the calculation shown for the selected adjustment category. In the event a loss is examined, each loss shown in the trade record is replaced with the product of the loss times the correction category selected. Losses are inflated and profits are reduced according to the correction chosen.

You may select from six different approaches in trade-by-trade degradation by simply clicking on the down arrow opposite the "Total Correction" box and making your choice:

 Total Correction Conservative Normal Moderate Generous Aggressive None Calculation N**[(P+1)/(N-1)] N**[0.62*(P+1)/(N-1)] EXP**[(2P+1)/(N-1)] EXP**[(P+1)/(N-1)] EXP**[1/(N-1)] 1.0

[The EXP term above is the Napierian natural logarithmic base of 2.71828, and ** is the symbol for exponentiation. The symbol P is the integer parameter count and N is the integer sample size.]

Corrections are most severe for the "Conservative" correction and are incrementally less severe for "Normal," "Moderate," "Generous" and "Aggressive," respectively. If no correction is desired, as might be the case for actual market performance, "None" would be chosen. Click [Next] to proceed.

Frequency of Profits and Losses

TSPE's first graphic output is a display of the Frequency of Profits and Losses with and without degradation. Three color-coded graphs are shown. If slippage is part of the trade statistics, only the original raw values chart and the "corrected" chart are shown. This display may help to verify the input statistics and offer some appreciation for the level of degradation provided. Click [Next] to supply the input for further evaluations.

Choosing a Reasonable Goal

The "Choose Goal" screen displays, showing TSPE's suggestion for what might be a reasonable profit expectation based on your profit and loss entries. You can accept this value or not, and may wish to enter any value that reflects a reasonable reward that you are totally committed to achieve by risking your personal capital.

The suggested Profit Target is designed to, hopefully, obtain some separation of the various curves displayed on the forthcoming Capital Analysis display. Feel free to return to this screen by clicking the [Back] key to experiment with different goals. It is common for the user to move between the "Choose Goal" display and the "Show Capital Analysis" display (next) in search of a feasible goal and time target. Multiple iterations may be necessary.

A nominal time target of ten (10) trades is suggested as a reasonable number of trades to consider before abandoning a pursuit of expected profit. The value entered will depend upon your patience in achieving success and the time it takes to consummate a trading round trip. Ten trades could take many months if average trade duration is lengthy. Please modify this input as you may see fit, but do not increase it beyond a reasonable threshold.

Understanding Probabilities

TSPE's results are all presented in probabilities, so it is important to understand this term before attempting to interpret the chart. For some readers the term "probability" may be sufficient for an understanding of this output. For others the term may be intimidating. Probability, in a statistical setting, is simply a number between zero and one. A probability of one (1) reflects certainty. For example, there is a probability of 1.00 that at dawn the sun will rise. A probability of zero means that the event will never occur. In between 0 and 1, the chance that an event will occur is some measurable amount. A probability of 0.10 would translate into a situation where the proposed event will happen 10% of the time, etc. A probability of 0.50 tells us that the event will happen half of the time.

Probability theorems tell us that, from the relative frequency of all possible events, a particular outcome will occur some computed percentage of the time. If TSPE tells you that the probability of avoiding bankruptcy with a given capital stake is 94%, then there will be a 6% (1.0-0.94) chance that your account will be forced into bankruptcy with that capital stake. Such a large (6%) chance of loss would not generally be acceptable to most traders.

Show Capital Analysis

Click [Next] from the "Choose Goal" screen to reach the final, and most important, TSPE screen representing Capital Analysis. TSPE continues to simulate trading over and over again, even after displaying the chart. You can see this by the slight movement and flicker on your screen. After allowing simulated trading for a few seconds, click [Stop] to end the sampling. Once stopped, you can view a stable chart or go back to earlier screens where you can change your input to see how a different mix of profits and losses or a different goal, etc. will change your result. See the caption under the chart for a full description.

In viewing the Capital Analysis Chart, the bottom or green colored curve is an important analysis result. This reached-goal curve should level off near 90% or more for a marginally acceptable result. Reducing the goal should increase your chances of achievement. Measure your chances of success (reaching your goal) versus the level of capital required and compute your percentage rate of return. There is a strong chance that the bottom curves may overlap if the likelihood of preserving your capital is minimal and your goal is small. If this occurs, click [Back] and  experiment with the goal and trade duration values.

 Chart caption:     The graphic display shows the probability of achieving your stated goal on the left scale, with varying levels of capital stake shown on the horizontal scale. The three curves plotted on the display (from top to bottom) nominally define the probability of avoiding bankruptcy (red), the probability of preserving trading capital (purple), and the probability of reaching the stated goal (green).     The right-hand scale shows the probability that none of the objectives are reached for the level of capital placed at risk. Notice that the probability of avoiding bankruptcy will increase to some asymptotic level as more capital is supplied. It should be reasonably easy to compute the maximum level of capital necessary. Please note that the probability of a non-event is (1.0-p) where p is the probability the given event will occur.

TSPE offers help in interpreting the Capital Analysis chart through the use of your mouse. Simply hold your mouse stationary on any point of the graphic image for an in-depth probability statement. For example, in the displayed chart, positioning your mouse on the vertical line 10^3 would produce the comment, "With capital of \$1000 and a goal of \$3,000 within 10 trades, a win probability of 43.9%, a breakeven probability of 46.3% and some capital left probability of 46.6%." A reduction in desired goal, an increase in capital stake, or an increase in the time target will generally improve results. This display should help traders decide what is best for their own situations.

Choosing Capital Required

Every trader should address the idea of avoiding bankruptcy; therefore the top red line should be every analyst's primary focus. Find the point on the top curve where you have a 99% or greater chance of avoiding bankruptcy before commencing with your trading. Conclusively identifying the required capital stake is paramount to success. A lesser capital stake may result in too small a chance of achieving your goal (left scale) and a high chance of not preserving your capital (right scale).

Your trading system should pay for itself with positive results. If it doesn't regularly provide an achieved goal that exceeds the capital cost of financing the trading experience, then quickly consider another system or approach. Fortunately, all of this can be done on "paper" in just a few minutes with TSPE, before committing precious capital. We wish you a successful season of trading and are confident that TSPE can help you in your efforts.

PAGE 1