TSCI.pngTradescapes Standard Edition


The Tradescapes Standard Edition offers core analysis options for traders in financial markets who employ indicators, strategies, and algorithmic signaling for entering and exiting positions. Each of the options within the standard version are designed to solve specific problems in entity selection and signal design that are intrinsic to any computerized system using financial modeling or prediction, signaling algorithms, or multicomponent indicators or strategies. The intention is to answer the key analysis and design issues effectively and with confidence.

The Standard Edition procedures are those which involve directly signaling on the entity to be traded using a single time horizon target and where binary (simple) entries and exits are generated for trades.

Tradescapes

This is the main functionality of the product, and the starting point for any analysis or design.

Determining Which Entities to Algorithmically Signal

A binary signaling algorithm will generate a specific count of entries and exits across a given span of bars. Depending on the bar density, these will represent a period of time. The count of transitions determine the number of trades within that period, and the average trade length is a general estimate of the time horizon of the signaler. To be effective in trading the ordered movements within the time series, most algorithms will employ indicators with a specific lookback period and various forms of moving average or fitting in order to manage the 'noise' in the price movements, the chaotic changes in price that result in the fast cycle trades known as whipsaws in the trade.

In order for a signaling algorithm to be effective, it must be able to map most of the ordered movements in prices in some fashion. This can be directly done with various smoothing and modeling approaches, or indirectly using a sympathetic signaler, an indicator or model that employs multiple indicators whose turns seek to strongly correlate with those of the target entity to be traded. In either case, the algorithm itself or its components will involve a lookback period, and a lag or latency in the information will result. One of the most important steps in developing a computerized signaling system is to determine how lag tolerant a given entity is, since the signaling will invariably have some mix of lag and inaccuracy.

Using tradescape signal analysis, you should quickly get a sense for what you can expect from your favored signaling algorithms in terms of lag. With this grasp of your resources in hand, you can view the tradescapes of a collection of entities to determine which of them have a potential to be effectively traded with your algorithm. You set a desired reward-pain ratio (or if desired a simple return) and the tradescapes will show you the historical behavior of the individual financial instruments. The TradeStation interface allows up to 50 entities to be simultaneously analyzed. The standalone version allows up to 100. In an astonishingly brief period of time, you can know which entities would have had success, by your own yardstick, across some period of past performance.

Determining the Time Horizon For Algorithmically Signaling a Given Entity

An individual tradescape is a 3D surface or contour plot of the outcome of 700 different backtests carried out at different time horizons and different measures of lag. The entire trading response surface is mapped. An ideal signaling algorithm, drawn from the EM (expectation modeling) science, is used to generate as close to a fully accurate signal as possible at each measure of lookback or information content, and those signals are then artificially lagged with no scatter. A tradescape is thus the ideal trading landscape for the ordered component of price movements.

When the history used is sufficient, a clear picture will emerge as to the measure of information that is optimal for signaling that entity. Some entities must trade very quickly, with a fast time horizon, so as to effectively capture the ordered movements in prices. Others must trade more slowly. Many will have multiple "sweet spots" and "dead zones" where trading will be especially effective or particularly poor.

By visual inspection of a tradescape, one can often know in seconds the time horizon most wisely used for signaling. Once that determination is made, the lag fraction the signaler will need to achieve at full accuracy is also immediately evident. A real-world signal algorithm will suffer inaccuracy from many sources, the whipsaws being perhaps the most prominent evidence a signaling failure. Unless you have an exceptional algorithm that offers both low latency and high accuracy, or one that can effectively trade both chaos and order, it will generally be understood that you will need to realize a lower median lag at the turns of the signaler than the tradescape surface reflects in order to realize the same reward-to-pain. In effect, you know how good you are going to have to be in your signaling before you even begin a signal analysis or design.

Standalone

  TradeStation

Primary Signal

Secondary Signal

Multiple Panels

Purpose

7901.png Tradescape

  TSTradescape

Traded Entity

None

Different Entities

Study symmetric signals of one or more entities

The Tradescape procedure is the primary starting point for all direct signaling applications that use EOD data.

Asymmetric Tradescapes

In a very simplistic sense, entry and exit points will tend to have a different mix of the human dynamics governing price movements. The nature of the price turns in one direction may be very different than those in the other direction. One of the first checks the trader should make before analyzing a real world signal against a tradescape is whether or not asymmetric signaling is needed.

Asymmetric Signaling

As with any optimization where there is a different parameter for the entry as opposed to the exit, the asymmetry of a signaler adds a good measure of complexity. A great deal of time can be spent in such multi-parameter optimizations, and they may be fraught with issues of over-optimization or over-fitting. A few fat-tail events that are perhaps unlikely to repeat in the same manner may unduly factor into such optimizations.

You can think of asymmetry as varying from the fast to enter-slow to exit strategy to that of the slow to enter-fast to exit approach. For example, the turtle breakout system with a 55-bar entry and a 20-bar exit would have an asymmetry greater than one since it is slower to enter and faster to exit.

 In a symmetric tradescape, the same time horizon is used to generate both the upside and downside transitions. In an asymmetric tradescape, separate time horizon signals are used for generating the upside and downside transitions.

A Meaningful Answer with Respect to Asymmetry

An asymmetric tradescape analysis is usually a panel of tradescapes where the information content used for entries and that used for exits is varied in different asymmetry ratios. With an asymmetric tradescape panel, you look at the entire trading landscape at each different asymmetry and you look for a decided advantage over the symmetric case. With a little experience, you will know whether or not an asymmetric signaler is going to be of value and you will have a good idea as to the asymmetry needed in your signaler.

Standalone

  TradeStation

Primary Signal

Secondary Signal

Multiple Panels

Purpose

7902.png Asymmetric Tradescape

  TSAsymTradescape

Traded Entity

None

Different Asymmetries or Entities

Study asymmetric signals of one or more entities

The Asymmetric Tradescape procedure is a set of tradescapes whose sole purpose is to answer if there is a robust and real advantage to be had from using an asymmetric signaler.

Progressive Tradescapes

One of the most critical questions that arise in any signal design is the robustness across time. Before any signal design is even begun, one can answer this crucial question using a time-sequence of tradescapes.

Robustness in Performance Across Time

Experienced traders may be able to look at the equity curves and have a good sense for robustness across time, and much can be inferred from the drawdowns in the mark to market equity, but a more crucial question rests with the entity to be traded. How well behaved is it across the different market states? If one were to select the target time horizon from an aggregate ten year EOD tradescape, for example, what would five different two-year periods look like in their respective tradescapes?

With respect to performance, one would expect a significant variation across the different time segments as the overall markets move through their different cycles. However so, one wants to see robustness in that time horizon across all of these states. That means you would like to see a higher reward than pain, or at least a positive return, at this single selected time horizon in each of the time panels in a progressive tradescape. If there were multiple time horizons that could have been viably traded, a progressive tradescape sequence may reveal one of those sweet spots to be considerably more robust across time.

Addressing Asymmetry Across Time

A progressive tradescape specifies the asymmetry of the signaling. If the asymmetry selected from the aggregate analysis period does not result in the desired robustness across the time segments, one can readily revisit the symmetric case or a shift upward or downward in asymmetry to see if the desired robustness across time can be found.

Overlapped Time Segments

Whenever time bands are separately analyzed, the reduction in information available to each can be an issue, especially when there is only a modest count of trades in the various time segments. When this is a factor, progressive tradescapes can optionally use overlapping time segments to map the progression through time more accurately. This will double the trade information in each individual tradescape at the cost of 50% of that information being carried forward into the next time segment.

Standalone

  TradeStation

Primary Signal

Secondary Signal

Multiple Panels

Purpose

7903.png Progressive Tradescape

  TSProgTradescape

Traded Entity

None

Different Time Segments

Study variations in time of the signaling on one entity

Intraday Tradescapes

Each of the tradescape procedures process data of any bar density. You can process 1 minute bars as easily as EOD data. A tradescape is entity specific, and it will also be data sampling specific.

Intraday Tradescapes

Intraday tradescapes are a convenience that allows a panel of different bar densities to be automatically constructed and analyzed using just one intraday input data stream. By using a reduction algorithm with a single input data stream, a fixed period is represented for each of the bar densities generated. An asymmetry may optionally be specified.

The purpose of an intraday tradescape is to identify the data density and time horizon for best trading the measure of order that can be realized from intraday price movements. We have seen vast differences between entities with respect to optimal data densities and time horizons.

While the built-in reductions are designed for 1min, 5min, or 10min bar intraday input data, the reduction series are generic and will be applied to any density of bars in the input data series.

Standalone

  TradeStation

Primary Signal

Secondary Signal

Multiple Panels

Purpose

7911.png Intraday Tradescape

  TSIntraTradescape

Traded Entity

None

Different Bar Reductions

Study different bar rates for signaling of one entity

For intraday trading, this is the typical entry point tradescape procedure. Once the bar density and time horizon have been selected, the other tradescape routines are used.

Trading Signal Analysis

This procedure evaluates one’s own real-world trading signals against the backdrop of an entity’s tradescape or asymmetric tradescape. This is the signal design procedure for the standard edition.

Performance and Signal Analysis

This signal analysis procedure is used exclusively to analyze single stage direct signaling. The tradescape is formed by directly signaling the traded entity.

Your real-world trading signals will be plotted atop the tradescape at their respective estimated time horizon and lag fractions. A brief summary of the signal analysis for each trading system is shown when moving the mouse over the special point representing that particular trading signal. The full signal analysis, including the equity curve, is shown when clicking on the point representing that particular trading signal.

In terms of the performance and signal evaluations, you are presented state-of-the-art leading edge analyses that form the core of the tradescapes technology.

TradeStation

If you use TradeStation, the process of generating and analyzing a trading signal is about as simple as it can be, although some experience with the platform's scripting language will be needed. The TradeStation tutorials cover this process in extensive detail.

ASCII Files with Signals

If your are generating signals from another platform or using your own code, the signal will need to be written into the columns of the ASCII file that also contains the price data. This can readily be done within Excel. The Standalone tutorials cover this approach.

Standalone

  TradeStation

Primary Signal

Secondary Signal

Multiple Panels

Purpose

7913.png Trading Signal Analysis

  TSSigEval

Traded Entity

None

Not Available

Assess a trading signal against the backdrop of a tradescape

The Trading Signal Analysis procedure is used for all signal analysis and design in the standard edition product.

Overfitting and 'Touching the Data'

Experienced traders understand the danger of overfitting and overoptimization. The features we have included in our standard edition of Tradescapes acknowledge the importance of 'touching' the data to be traded as minimally as possible in the process of signal design. We consider it essential. In our tradescapes paradigm, we 'touch' the data 'one' time whenever we perform a full-accuracy trading map or a panel of such maps. Each tradescape analysis is designed to represent what we regard as a single 'touch' of the data.

You touch the data once to determine the time horizon to trade and to see how good you must be with your signaler in terms of the lag tolerance you are granted for that entity. You must still find or build the real-world signal that delivers that time horizon and lag fraction, but the tradescape design can essentially be finished with this single 'touch' of the data. The direct-signaled options in the standard edition manage exactly this, helping you to avoid the pitfalls of overfitting and overoptimization. We 'tune' a signaler for the robust time horizon we target. We may further 'tune' it to decrease the lag. By designing for time horizon and lag you avoid the overfitting of parameter optimizations that look for the best backtest performance. In fact, once you know what you can expect in time horizon and lag fraction from a given signaler, it is even possible the first equity curve you see may be the one that summarizes the final design.