Mathematical and Statistical Methods for Actuarial Sciences and Finance

258 D. Pelusi

01−Jul−20031.05 01−Oct−2003 01−Jan−2004

1.1

1.15

1.2

1.25

1.3

Time

Euro−Dollar exchange rate

01−Jul−20071.3 01−Oct−2007 01−Jan−2008

1.35

1.4

1.45

1.5

Euro−Dollar exchange rate

01−Jul−2007 01−Oct−2007 01−Jan−2008

−0.2

−0.1

0

Time

Profit

Fig. 1.Trading phrase using the second semester of 2003 as the training period

note that there are considerable losses at the beginning and at the end of the trading
semester.
Table 2 summarises the algorithm application results to the semester pairs chosen.
From the observation of these results, we infer that there is a global similarity threshold
which lies between 0.61844 and 0.66299. For global similarity values greater than
this threshold we should obtain profits.

4 Conclusions and future work

In the technical analysis literature, some authors attribute to chart patterns the property
of assessing market conditions and anticipating turning points. Some works develop
and analyse the information content of chart patterns. Other papers have shown the
importance of choosing the best trading rules for maximum and stable profits. There-

Ta b le 2 .Profitability and similarity results of the semester pairs

Training semester Trading semester Profit GS

2 nd 2003 2 nd 2007 − 0. 1301 0.61844

2 nd 2004 2 nd 2007 0.2592 0.66299

1 st 2006 2 nd 2007 − 0. 1367 0.61634