A mere 28.8% in a year, that’s a pittance. Based on the same setup but testing more combinations I came upon this. Big surprise is the SMA 70 / 90 being the most profitable, and the 480 might be a surprise if i was more familiar with the scale of numbers that characterise the DAX30. But the huge surprise was the profit of USD 5329 off a base of $3000 in 1 year. I ran the visual results version of the top line and have posted it below. It accurately represents the first line of the numbers produced here.
|5329.65|TakeProfit=480 |MA_Period=90|
|5329.65|TakeProfit=480|MA_Period=90|
|5106.08|TakeProfit=480 |MA_Period=70|
|4965.93|TakeProfit=470 |MA_Period=70|
|4909.25|TakeProfit=450 |MA_Period=90|
|4825.78|TakeProfit=460 |MA_Period=70|
|4685.68|TakeProfit=450 |MA_Period=70|
|4545.58|TakeProfit=440 |MA_Period=70|
|4488.85|TakeProfit=420 |MA_Period=90|
|4405.43|TakeProfit=430 |MA_Period=70|
|4265.28|TakeProfit=420 |MA_Period=70|
|4208.65|TakeProfit=500|MA_Period=90|
|4125.23|TakeProfit=410 |MA_Period=70|
|4068.5|TakeProfit=390 |MA_Period=90|
From here the results tail off steadily. The top 32 results from 144 tests all use 70 or 90 SMA and all use fairly large TPs (all 300 or above with the better ones above 400 - the pattern in these top 32 is very clear.
So am I curve fitting (that seems to be a dirty word around here)? Not at all. I am simply saying that from this date to that date based on data freshly downloaded from the server and seemingly accurately modelled, a TP of 480 and an SMA of 90, with an emergency stop based on a candle closing on the wrong side of that SMA produced xx result in USD.
That’s all I’m saying. Is it a useful predictor of future performance? Possibly. But until I have done the same test over a similar period further back in time eg the 12 months of 2009, or 2013 or whenever and found that the numbers are similar or totally contradictory, these numbers stand alone and say nothing about the future.