There is a 5 file limit to attach in this foum?
Anyway here is the atr_results.zip file...
I'll drop V3 on my Neurex feed and see what I get from that. I'll post results at the end of the month.
Andy
Statement update... TSD_TR_0[1].2_midweekswitch.mq4 EA
Testing on 2 different brokers: Alpari and MIG
Sada
I will test out the system, so that we don't duplicate settings, what is everyone else using as settings. I can dedicate a computer to demo trade it for a while, what currency pairs does it work best on?
Any info on the EA would be appreciated.
By the way jdun, which EA version are you talking about?
Sada
Still testing midweek switch ?
I'm testing default settings on (6) pairs and EA appears to be working properly. Only problem right now seems to be the losses incurred during a trend retracement. Perhaps obtaining weekly trend direction on current and previous bars instead of previously closed bars during retracement ?
Ronald
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
This is an EA I found on the Yahoo Metatrader Group.
The EA is a modified version created by "TR" from the orginal TSD EA (also posted in this forum)
Settings: default all, D1 charts
There are 3 versions floating around:
Version 0[1].2 might be the same as version 0.2 but I didn't check if that was true. I got 0[1].2 from one of the tester of the EA and it's the same one he and I are using. (attachement statement_the other_tester and statement_me_myself_and_I)
There is also Version 3...here is what the author posted when he attached this version:
"Well folks, after running TSD-TR_0.2-midweekswitch.mq4 on 8 pairs for a
week, I compared the demo and backtest results and can report that
everything matches within 1-2 pips. I had to use the logs to verify the
pending trades in backtest that never were hit, as I had wrongly assumed
that the account history report would include those--maybe it only does in
MT3. This means that there may have been additional pendings set in demo but
not backtest that my comparison would not have found, but that seems
unlikely. So it still looks to me like the backtests I've been doing are
accurate enough to rely on.
However just before and at the open of this week I again saw that trade
operations were returning a success code even as the journal was showing
"trade disabled" before market open, and no trade but instead a relogin
after market open. This was after I added print statements before and after
trade ops so I could be sure the code was actually running. So I feel
confident to say that the code returned by trade operations is at this point
totally unreliable and we can't even be sure it only happens in these two
instances.
I've seen other methods such as the new "lockSemaphore" method (which is
just using a global regarding which I've previously expressed my
reservations), and the IsTradeAllowed method. But even assuming those
methods work, we still need a reliable return code to handle problems other
than trade collisions!
So for now it seems to me the only complete solution for TSD is to run the
weekly code on each tick, checking that the trades are where they should be
based on the current signal. This brings a few new problems and would
require tweaks to the system, including a reliable way to ensure a trade
isn't reopened after a stopout. So for now I've decided to try and get MQ to
fix this return code problem, and in the meantime simply reattach the
experts to the charts at the open of a new daily bar, and check the signals
manually. It's looking at this point like I'll likely be live with max 2
pairs anyway. Once I am live, then I'll consider making my versions reliable
for unattended running.
So to be clear: the experts I've posted so far are not reliable for
unattended live or demo use, you must reattach the experts to the charts at
the open of a new daily bar, and verify the signals manually. Hopefully I'll
remember to mention this whenever I release new versions.
As a preview of what I'm working on, here are results after adding a
ATR-based trailing stop. It works on closed bars only (when run on 15M--the
trailing code needs to be modified to work on any timeframe) so backtesting
should now be fully accurate aside from news and weekend related slippage. I
haven't yet verified that it works at open prices only though, so I'm not
yet 100% sure. But here you go anyway. (Files attached.)
GBPUSD as good as ever, and EURUSD now much improved! I might just go live
with this version after analyzing the results a bit more.
If I can ever get to it, I'll add quite a few inputs that I can using in a
massive optimization run on say, 8 months of data for each pair, and thereby
finally be sure which pairs are worth continuing with. The pairs that stay
in will likely have the optimized inputs tested on Alpari MT3 data from June
2004 after conversion to 15M and bid-base, to be sure it's not just a
curve-fit. Full report will be forthcoming...if I can get to it...
- TR"
This EA has been tested by the other tester since the end of october and I have been testing it since the beginning of last month. As you can see we are using different brokers and our trades are almost the same on the same pairs. What is the big deal if they are? well a lot of EAs have different results depending on what brokers platform you use so to have them almost identical pips wise is re-assuring if you want to go live with it.
I am going to start forward testing version 3 by monday and see if there is any impact on the overall result and reliability.
It would be great if some other forum members here decides to join the ride in demo-ing this so we could compare our results.
Sada
PS: the atr_result.zip is the backtest results from TR for the version 3.
PSS: Again, I would like to stress that it would be nice if people decide to test this EAs out to POST AND COMPARE their statements...come on, make the effort and stop being lazy & selfish!