MNB 10.4 Streamlined by Snailbeard

[snailbeard]

Progress so far

The results are overall good but not exciting (on one currency)
Good points: average profit to loss > 2

Bars in test 5666
Ticks modelled 11218
Initial deposit 10000.00
Total net profit 1069.17
Gross profit 1898.55
Gross loss -829.37
Profit factor 2.29
Expected payoff 62.89
Absolute drawdown 57.90
Maximal drawdown 584.03 (5.30%)
Relative drawdown 5.30% (584.03)
Total trades 17
Short positions (won %) 12 (83.33%)
Long positions (won %) 5 (60.00%)
Profit trades (% of total) 13 (76.47%)
Loss trades (% of total) 4 (23.53%)
Largest
profit trade 457.68
loss trade -213.88
Average
profit trade 146.04
loss trade -207.34
Maximum
consecutive wins (profit in money) 6 (1001.05)
consecutive losses (loss in money) 2 (-425.27)
Maximal
consecutive profit (count of wins) 1001.05 (6)
consecutive loss (count of losses) -425.27 (2)
Average
consecutive wins 3
consecutive losses 1

It would be interesting to know what would happen if we could do basket backtesting!
If it placed x10 trades with the same win/loss ratio then it would be very exiting

ToDo
Re-enter deleted PO at better price
Experiment with partial close at pivots
Experiment with zig-zag based range predictions

[snailbeard]

Epiphany

If you have been following my earlier posts
then you won't suprised that this thread
has come to a conclusion or rather a realization:

We need to throw away almost everything:

The other indicators will keep you out of the
market when you can be in, they'll take you in
the wrong direction when it is time to reverse.

I can image CJ nodding sagely.

Here is what currently works for me ( see attached )

And for the future:
If we can reliably find the Fractal-Fibonacci levels and channels
then we can also throw away the pivots, the Bollinger and the ZigZag.

So it is time to move to another thread where it would be more appropriate to explore such ideas,
and possibly to a modern software development environment as well.

[snailbeard]

Despite my interest in a more frequent trader, I still bang away at this one now and again.
It could be a nice little earner if we can overcome its main weakness!

I spent several days developing some code which can close trades in parts.
Although the code now works and removes a fraction say every 20 pips.
The overall result is disappointing with this strategy.
The number of partial exits and the gaps between exits are set at run time.
It could be that these parameters need to be optimised.
The main contribution to the disappointment is caused by the same issue that plagues this EA:
not knowing that the trend is reversing.
The idea of the partial close was to win some pips then breakeven before a reversal all the way back to the stop-loss. However, there are too many reversals which never get to the first partial close so the losses are bigger in proportion to the wins.
It seems more effort is needed in understanding reversals near the end of a swing.

In a bad week the drawdown for 10 pairs could be 50%!

After struggling with Empty4 and large quantities of M1 data, I can now back test from 2007 but only on AUDUSD, because GBPUSD M1 seems to overwhelm CrappyT4.

Summary of back test
M1 mid-2012 to mid 2013

Bars in test 412058
Ticks modelled 820853
Modelling quality n/a
Mismatched charts errors 0
Initial deposit 10000.00
Total net profit 858.62
Gross profit 2484.62
Gross loss -1626.00
Profit factor 1.53
Expected payoff 31.80
Absolute drawdown 25.00
Maximal drawdown 551.92 (4.98%)
Relative drawdown 4.98% (551.92)
Total trades 27
Short positions (won %) 19 (73.68%)
Long positions (won %) 8 (62.50%)
Profit trades (% of total) 19 (70.37%)
Loss trades (% of total) 8 (29.63%)
Largest
profit trade 213.24
loss trade -229.20
Average
profit trade 130.77
loss trade -203.25
Maximum
consecutive wins (profit in money) 5 (560.10)
consecutive losses (loss in money) 2 (-395.52)
Maximal
consecutive profit (count of wins) 560.10 (5)
consecutive loss (count of losses) -395.52 (2)
Average
consecutive wins 3
consecutive losses 1

[nonlinear]

Snail, applaude your efforts -- so difficult to find a consistently winning system