Forex Correlation. Trading System R&D.

Our primary goal is to create a public information service providing financial markets forecasts, based on our proprietary forecasting tools: an automated trading system -- a Forex Automaton™. A live forecasting system operating on daily time scale, Danica, is online since late December 2009. An hour-scale forecasting system, Heidi, and a daily signal EUR/USD system running a model portfolio, Demi, were launched in August 2010.

Our secondary goal is to discover, quantify and monitor the very existence of sustainable opportunities for profit-making via systematic trading. Or simply put, to monitor the degree to which these markets are more predictable than a "fair game" -- to a trader without access to insider information.

In April, Danica underwent a major upgrade. The nature of the change is described in the Annual Research Report. The objective of the change is extend the adaptivity of the algorithm to embrace both main types of second order correlation patterns: trending and reversion to the mean without addition of extra parameters.

The number of currency pairs we track has been reduced from 14 to 6. The format of these monthly performance reports has been simplified and made more visual: tables are replaced by figures.

Here is the April 2013 update of the ForexAutomaton correlation analysis report, with its distinctive emphasis on time-lag dependence of auto-correlations and intermarket correlations.The report follows structure and definitions explained in a separate document, which can be used as a user manual.The report deals with the following exchange rate time series: AUDJPY, AUDUSD, CHFJPY, EURAUD, EURCHF, EURGBP, EURJPY, EURUSD, GBPCHF, GBPJPY, GBPUSD, USDCAD, USDCHF, USDJPY.

Success of algorithmic trading has as much to do with the market as with the nature of the algorithm. While the algorithms may vary and may have terms such as artificial intelligence, adaptivity, machine learning, neural networks and the like associated with them, the phenomenology of what it is that the machines are "learning" all comes down to correlation measurements of varying order -- pairs, triplets, quadruplets... -- over a variety of time lags and time scales, performed on the market time series.

ForexAutomaton began its life on April 02, 2008, five years ago. Year V was exceptionally productive for the project. Here are the main highlights of this annual review:

• An Index of Correlation Strength (CERCSI) and a Predictability Index have been formulated and the history of these measures has been established.
• Index of Correlation Strength reveals that a regime switch in the market dynamics took place in Summer 2007.
• Danica, the daily forecast system, has confirmed its status of an "Anti-System", delivering its forecasts in the non-random and consistent opposition to the market reality.
• The challenge of regime switch has been addressed by implementing an algorithm, nicknamed Twisted Pair, to incorporate sensitivity to two basic types of memory effects: trending and reversion to the mean.
• Danica, the daily forecast system, and Heidi, the hourly system, have been upgraded with the Twisted Pair algorithm.

In April 2013, Danica was upgraded to version 2. This version is believed to greatly increase the number of pattern types the system is sensitive to, by providing the ability to adapt to the so-called reversion to the mean. The structure of the output has been changed, new types of quantities have been added. This document is intended to serve as an addendum to the original user manual, explaining the new output.

Internally, the system classifies each forecast and keeps track of historical profitability of each class. Strategy is based on these historical profitabilities, and consists in taking a position in accordance with the expected profitability, with the position size proportional to the ration of the expected return to the statistical precision with which such expected return is estimated. Return expectations in the class have to exceed one standard deviation for the forecats to result in a position. The quantity listed as position is the ratio of return expectation to its precision (standard deviation). A "-" symbol instead of a quantity in the position field indicates that the forecast did not qualify as a valid one to open a position.

P&L numbers for the strategy are listed in units, relative to the quote, for all time and the last 50 updates. The P&L numbers have been reset before launching the system and therefore they do not include back-testing performance. The P&L numbers are per-trade averages and are accompanied by statistical precision estimates.

A time-integrated study of hourly time-scale lagged correlation between USD/TRY and EUR/JPY reveals hints of a correlation whereby EUR/JPY leads and TRY/USD follows with a lag on one hour.