CL151

Results for the week ending November 7, 2014:
- BasicTM: 2 wins ; 1 BE ; 1 (max) loss ; net -320
- AlternateTM : 1 wins ; 1 BE ; 2 losses (incl. 1 max loss) ; net -680

Max-loss trade on Wednesday was pretty upsetting, as price moved 2-ticks beyond our stop, then turned around and made a trend-change 1pt higher (which would have been the trade exit, if it had not been stopped).
 
Update re. the CL151 Market-Model statistics.

The first table is the trend-count on a monthly basis – I have highlighted in yellow the months below 100, as you can see there was a severe decline in trend-count starting in November 2013, with 4 months in 2014 down to the 50-60 range (May to July). The good news here, is that September, October and so far November are back above the 100 mark (October significantly above it), which translates directly into higher trade counts.

market-model stats.1.png


The second table is the % of CT-trends (trends which need to be faded for a win, as patterns 151x, 302 & 153 do). The long-term average for CT-trend is 64.9%. Except for January & October, all other months in 2014 have been under the long-term average. There is no doubt that all CT patterns have had a rough time from February to August, in particular in February, June & July were all under 60%, with July making an 8-years low at 46.3%, almost 9% under the prior low (54.8% in December 2012). With 66.3%, October has popped above the long-term average, which I take as a very positive sign, as I am expecting the pendulum to swing to the other side at some point.

market-model stats.2.png


The last table is the % of TR-trends (trends which need to be traded in the direction of the trend for a win, as patterns 152, 154 & 155 do). As expected, this is the mirror image of the prior table: the long-term average for TR-trend is 35.1%. Except for January & October, all other months in 2014 have been over the long-term average. There is no doubt that all TR patterns have had a good time from February to August, in particular in February, June & July were all above 40%, with July making an 8-years high at 53.7%, almost 9% above the prior high (45.2% in December 2012). With 33.7%, October has come down under the long-term average.

market-model stats.3.png



What does it all mean, and how can we make the most of it?


If the current (October) market behavior stays for a while, we can expect a trade-count more in line with prior years, as well as improved win% for the CT patterns. It seems to me this is the right time to activate the 2 patterns that I did carve out of the 151 lower-half (patterns 1511 & 1512) late August, but did not include in the default configuration then (they have been in the software, unchanged, since v06a was released).

As a result, I have decided to activate those patterns 1511 & 1512 for my own trading.
 
Results for the week ending November 14, 2014:
- BasicTM: 5 wins ; 1 BE ; 1 loss ; net +1175
- AlternateTM : 3 wins ; 1 BE ; 2 losses ; net +1330

Way more trades than any week I can remember in the last few months ... of the 7 trades for BasicTM, 3 were from the patterns just activated (2 wins + 1 BE).

AlternateTM got a sour exit, for 1-tick and/or less than 100-contracts traded in excess of a system threshold in a pullback, triggering an early exit when that trade would have otherwise made 66 more ticks. Of course, this will always happen once in a while (sometimes for the better, sometimes like today for a not so good outcome).
 
Results for the week ending November 21, 2014:
- BasicTM: 1 win ; 1 BE ; 2 loss ; net -1420
- AlternateTM : 2 BEs ; 2 losses ; net -1680

Double frustration last week - 1st I had a "phantom" trade overnight, after a corrupt reload of historical data - didn't see it until that trade was already down ~-500 (my own account VPS was the only one to have this phantom trade, backup PC was flat, as was the other VPS I operate - I reloaded historical data manually on backup PC first, still flat, then I reloaded on my VPS, and after restart the system got flat). 2nd frustration we had a 2nd max loss in less than a month, stopped 3-ticks before the extreme of the move - the exit at next trend-change would have been 70-ticks better.

I released a maintenance version of all dom993trading systems, introducing "Client Alerts" - these
are emails sent to a configurable ClientMonitoringEmail upon the following operations events:
  • Strategy Start : alert sent once the strategy is alive, ie. has seen its 1st live bar
  • Strategy Stop : alert sent when the strategy is terminated
  • Connection(s) Lost : alert sent when 1 or more broker/datafeed connections are lost
  • Connections OK : alert sent when all broker/datafeed connections are back to connected
  • DataFeed Inactive : alert sent when no transaction has been reported for the instrument for a period of X seconds – the purpose of this Alert is to detect potential silent failures of the broker/datafeed connection (ie., in the Connected state, but actually not working)
  • DataFeed Alive : alert sent when a transaction is reported for the instrument, after sending the prior alert
Working on another DataFeed Alert: "IntradayGap", which will alert on suspicious intraday gaps (potentially erroneous historical or live data).
 
Results for the week ending November 28, 2014:
- BasicTM: 3 wins ; 3 losses ; net -900
- AlternateTM : 2 wins ; 1 BE ; 3 losses ; net -770

Finished testing the "IntradayGap" Alert(s) - there's gonna be 2 of them:
  • one to flag real-time Intraday gaps, with 2 settings (1 for RTH, 1 for outside RTH).
  • one to flag Intraday gaps in the Historical data, only for the current week AND only if that Intraday gap wasn't seen as a real-time gap - the idea here is to detect corrupted reloads of historical data
I'll be beta-testing those new Alerts this week on my backup machine (with tighter settings) as well as on my VPS (with normal settings). Planning to release this update at the end of the week, if all goes ok.
 
You must log in or register to reply here.
Back
Top