Tour De France - 30000 points

ok chaps, here is another method.

the EUR/USD has more cycles in it than the tour de france. one of the main ones that jumps out at me is the 10-day cycle.


something clearly changed in the Eur/Usd market around the start of january 2003. which might be why the equity curve on this little beggar is so bizarre. it weaved around during 2002, then all of a sudden, it just took off, and hasnt really looked back over the past 2 years.


here are the stats of the Swiss Roll:-

Overall Stats
Points 29224
Trades 694
Wins 372
Win % 53.6%
Average Win 256.52
Average Loss -205.59
Profit Factor 1.44
Max Drawdown 8051
Profit per Trade 42.1

Since 1/1/03
Points 31575
Trades 459
Wins 260
Win % 56.6%
Average Win 292.53
Average Loss -223.53
Profit Factor 1.71
Max Drawdown 3517.00
Profit per Trade 68.79


this system is based on daily data, rather than weekly, hence the increased number of trades, and again, trades on "closing" prices at 9pm.


i havent coded any "silly" stop-loss rules, or buying on dips etc, after the balls-up last time around, but im sure stops, and stop-and-reverses could improve profitability further, and reduce the drawdown, which still looks a bit eyewatering.


Da Rulez:-

again, it is based on a day scoring method.

if today was an up day, score -1
if today was a down day, score +1

then, do a sum of the last 7 days. if this figure is positive, then go long, if negative go short.

simple eh?

so where do the cycles fit in?

well, since we seem to move in 10 day cycles, we hold each days trade for 10 trading days, and then on the tenth day, we flip the trade and reverse it for 10 trading days.

so for example if on the 1st november we got a signal to go long. we would hold this long until the 15th november (10 trading days - we exclude weekends) and then at 9pm on the 15ths, we would flip the trade (hopefully banking profits on the way) and fire in a short to hold until the 29th november.

and thats it. of course it results in multiple positions being held at once, and thus may require a rather substantial trading pot to operate fully, and take all signals (have been looking into filtering the entry. more on that later), and it may get a bit complicated to work out what trade needs opening and closing when, but it is nothing that a decent trading log shouldnt solve.


im still looking for a way to get 40,000 points from a day-day system over the last 3 years, and this is pretty much the closest i have come.


suggestions on a postcard to the usual address.

ps, other cycle period still show profitablity, but the 10 day one seems to offer the smoothest equity curve (yes i know it still is wonky) and the best returns vs drawdown.


toodle-oo,

FC

forgot to mention.

10 period cycle also exists on weekly charts, but needs slightly different entry criteria to catch the extremes...

FC

Surely this is biased to the date when you started to trade.

i.e. once you have the first 10 trades all you are doing is flipping every 10 days.

To me this is a fundamental flaw unless you can persuade me otherwise...

JonnyT

not really, as you only flip the trade once.

each entry is based on the 7 day rolling bias.


try if for yourself.


delete some of the rows from the start of the sheet and see the difference?


ill post an example in a few mins..


FC

ok, here is an example.

i picked 19th may 03 as a start date (i think)

still works nicely, IMHO.


FC