Forex Automaton

During February 2011, the second month of the second year of live performance, the system continued on auto-pilot without parameter changes. This document consists of a summary section reporting the figures of merit for the forecasting quality, followed by 14 subsections, dedicated to the individual exchange rates tracked by the system. Those contain our usual green-yellow-blue-red color-coded charts of the performance (2 for the two hypothetic strategy triggers for each currency pair, 28 in total) and details pertinent to the specific currency pairs.

Looking at the most important figure of merit, namely the correlation coefficient between predicted and actual returns for close, the forecasting system eked out a marginally positive result this month. The main contributors to the positive performance were USD/JPY where the system was able to quickly readjust to the yen rally in the second half of the month, and AUD/JPY.

For comparison with the previous month, you may want to take a look at the January 2011 performance review.

Figures of merit

The figure of merit used, Pearson correlation coefficient between the forecast and real logarithmic returns on a day scale, is a measure of forecasting quality.

By construction, the Pearson correlation coefficient is a quantity bounded between -1 (the forecast move and reality are total opposites) and 1 (the forecast is perfect). A success or lack thereof on every trading day makes a contribution to this quantity.

In order to make a large positive contribution, one needs a coincidence of a large move in a currency pair with a large forecast move in the same direction. Since a hypothetic rational operator of the system will not pursue small forecast moves, understanding this to be a noisy system, where a forecast with large magnitude is more likely to result in a successful trade.

Why use Pearson correlation coefficient instead of running a model portfolio? First, we do run model portfolios -- see Demi -- but that's a separate product involving additional optimization choices. Second, this independent figure of merit, characterizing the forecasting component regardless of the money management strategy used, allows one to split the complex problem of trading system development into independent tractable pieces, and solve each pieces of the problem separately, eventually combining a winning forecasting system with a winning money management one, having an independent quality assurance process for each.

For those interested in the relative frequency of being "right" vs being "wrong", these outcomes, judged by the prediction success for close (as opposed to high and low which are a lot easier to predict), are counted.

Performance tables

In the tables, data for the "month" column are taken from the "month" column of the last day's forecast, which reports 24-trading-day running average of the Pearson correlation figure of merit. It must be noted that the "month" column is for rough orientation only as it gives the average over 24 most recent trading days which may not necessarily be the month. In contrast, the conclusions on the basis of the "since inception" columns are strict. The mean over markets for the Pearson correlation coefficient for the month was negative. End of February 2011 data are taken from the February 28 update. End of January 2011 data are taken from the January 31 update.

Discussion of performance

As Table 0.1 indicates, this month's effect on the performance figures for daily close since inception was net negative.

The mean value of correlation coefficients for high and low went up as they did every month since January 2010. The improvement is due to the v1.0 upgrade performed in June 2010 and the v0.5 upgrade performed in January 2010.

The system performed very well for USD/JPY and AUD/JPY, with EUR/AUD, GBP/JPY, CHF/JPY and GBP/CHF being other net positive performers (in descending order of performance). EUR/AUD maintains its status as the most predictable currency pair.

The following 14 subsections are dedicated to the specific currency pairs, using a particular method of charting to illustrate performance. Two charts will be presented for each. The first chart will show all trade ideas. This corresponds to trading in every pair every day.

The second chart will highlight the days when forecasts for all three components of a day's candlestick were pointing in the same direction. I call this level zero (L0) requirement or L0 trigger. The rest of the days will be colored in black.

Surprisingly, unlike other months with net positive correlation for close, th L0 trigger did not enrich the stream of outcoms with wins, as Table 0.4 indicates.