Profit mismatch between Stategy Optimiser and Fast Test

[philBell]

I'm using version 2.1.5 (01.06.2009)

Using the example 2SMA strategy, setting it for 20 and 40 MAs on EURUSD 15 minutes from 1 Aug 2008 to 1 July 2009, with LotSize 0.1, Stratefy Optimiser for a single set of parameters shows a profit of $29.3 in its statistics panel (and doesn't execute Print statements in the strategy source code, so I can't get more information out of it). With the same strategy settings, FastTest reports a profit of $1505.80. I would have thought these should be the same? Cancel this for now; I just tried it again and got $1585 for both - I giess I had either a date range or a parameter difference between the two cases; sorry.

Minor points : using the 2 SMA strategy code, setting overlapping ranges for the search of the two MA periods, the profit for period1 = 7 and period2 = 8 should be the same as the profit for period1 = 8 and period2 = 7 I would have throught, but they are different ($-4477 v $-4511 over all EURUSD data).

Setting the optimiser to vary period1 from 19 to 23 in steps of 2, and period2 also from 19 to 23 in steps of 2, gives an extra output record in the file after the ones you would expect (the extra record is for SMA1 period=19, SMA2 period=23). Would this be the values it decided where optimum?

It would be more convenient for importing the results from the optimiser into Excel if the currency, timeframe, lot size, and two periods on the end of each record where appended in the same format as the rest of the data, ie as values below a column heading, rather than in the current repeated "heading, value" pairs Currency,"EURUSD",Timeframe,"15",LotSize,"0.1",SMA1 period,"19",SMA2 period,"23" format.

It would be nice if running the optimiser didn't erase all the indicators from the price chart before it returns to Testing Mode (eg 2 MA indicators and a stochastic, which I put on for manual testing)

Phil Bell

[philBell]

Thinking about it (which, I agree, would have been better before making the post, but I thought I'd better add the retraction rather than just editing it out of the original, so you know it was my mistake) the point about different results when interchanging the two MAs (first "minor point") is not I bug, I think.

Trying Optimiser over all EURUSD data, which is where I mistakenly thought I'd seen the "problem", every combination I'd tried had returned a loss of around $-4500. I now see that this was may be because they'd all hit a margin call (at 90% of an initial capital of $5000) and stopped making further trades. The slight differences could be explained by the last trade in each run overshooting the margin call threshold by different amounts? Presumably, if a run has period1 > period2, when the strategy is designed on the assumption that period1 < period2, it starts making Sells when it should be making Buys, and vice versa?

Sorry.

Phil Bell

[philBell]

Sorry, I forgot. In the Optimiser's parameters panel at the foot of the left-hand column, it appears that the value of the last parameter registered by RegOption() (SMA2 for the example 2 Simple MAs ) isn't displayed.

[philBell]

I think I may have been mistaken about the difference in profits between the two results; I tried the test again and got the same results from both runs. Maybe I didn't have equal date ranges or equal parameters for the two cases. Sorry.

[Tantalus]

It seems to me I had a similar experience when I was new to FT... it takes a little time to learn the ropes!

B-

[philBell]

Thanks; yes it does.

Phil Bell