userque & NUGT/DUST

Sort by newest Sort by date
Troubleshooting update.

I keep copies of the last 26 versions of anything of importance I'm working on. Upon inspection, I saw where I broke stuff. (There was more involved than my previous post indicated. I realized that after conducting longer backtests, which showed Houston that there was a problem.)

Fixed stuff and added new features. Currently hill climbing (optimizing). Optimizing on a meticulous scale. Not even one quarter of the way finished with what I believe will be the likely complete search field.

Optimizing is simply plugging in different parameters in some orderly fashion and noting which performs best. It is used with complex systems such that a determination can't simply be made with maths etc as to what should be the best configuration.

I start optimizing with 'easier'/'quicker' parameters first, then move to longer ones. When not being meticulous, I start with the likely 'best' parameters first.

I'm testing essentially all configurations, to ensure the best possible result...even if I won't finish until after the opening bell...it's currently good enough to trade with right now.

To break any ties between two or more configuration settings, I simply expand the time frame of the testing until only one 'best' is left. Currently:

The time frame is from 4/6 through 3/16
The best possible result (trading perfectly) would turn $1000 into about $1934.40
The best settings so far yield $1311.46 from $1000...again, not 1/4th done yet with the settings search. (The worst settings so far were in the $800's, btw)

Anyway, the best-thus-far likes being in NUGT right now (what I'm in). Instead up posting an update every time the system finds a better configuration, I'll post the best-so-far, and the position I'll be taking or not (circa the open), before the opening bell.
 
Seeing that my last question in the racists' thread is still unanswered, and;
I haven't been crucified (Jesus) or assassinated (King) yet, I'll catch you up.

Almost finished generating the backtest data going back through Feb. I've been very meticulous in the optimization, so it was slow-going this weekend. I believe I've got the best parameters and have put the backtesting balls to the wall.

Should certainly have backtest data through Jan. available later today.

EDIT:
Next, the TSP model...that'll take a while.
 
Oh, and I'll do some a solid today...in case my existence is permanently cut short on this website today. :)

I normally only give advance notice to a loyal few, but, the system issued a NUGT signal for today's close.

I'll be holding NUGT overnight.


__________________
"Like anybody, I would like to live a long life. Longevity has its place. But I'm not concerned about that now. I just want to do God's will. And He's allowed me to go up to the mountain. And I've looked over. And I've seen the Promised Land."



__________________________________________________ __________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
 
Backtest data for this year:

1=tomorrow will close up, so go long, (1)=opposite, 0=hold previous position

EDIT: Running another backtest with a small parameter change to see if I can improve on this.

Code: 
Date         Open    High     Low   Close         Volume  Actual Forecast Balance
  4/7/2016  20.59   20.86   20.52   20.78    79301771.00    1.00    1.00  1841.26
  4/6/2016  19.88   20.23   19.80   20.15    35891037.00    1.00    1.00  1785.44
  4/5/2016  20.06   20.19   19.56   20.13    74601422.00    1.00   (1.00) 1787.21
  4/4/2016  20.05   20.05   19.51   19.53    65764779.00    1.00    1.00  1733.94
  4/1/2016  19.46   20.12   19.32   20.12    58369174.00   (1.00)  (1.00) 1684.54
 3/31/2016  20.48   20.56   19.96   19.97    46237376.00    1.00    0.00  1697.29
 3/30/2016  20.23   20.50   19.85   20.23    62118354.00   (1.00)   0.00  1675.75
 3/29/2016  19.63   20.60   19.40   20.54   113458281.00   (1.00)  (1.00) 1650.84
 3/28/2016  19.62   19.66   19.09   19.42    35286724.00    1.00    1.00  1560.82
 3/24/2016  19.25   19.60   19.07   19.46    64669516.00   (1.00)   0.00  1564.04
 3/23/2016  19.78   19.88   19.02   19.03   137219672.00    1.00    1.00  1529.48
 3/22/2016  20.90   21.00   20.35   20.58    41735369.00   (1.00)  (1.00) 1422.35
 3/21/2016  20.50   20.78   20.31   20.57    39053220.00    1.00   (1.00) 1423.04
 3/18/2016  20.36   20.93   20.35   20.61    66406771.00   (1.00)   0.00  1420.29
 3/17/2016  21.05   21.42   20.37   20.43   110884035.00    1.00    0.00  1432.91
 3/16/2016  19.22   20.96   18.98   20.86   156219544.00   (1.00)  (1.00) 1403.97
 3/15/2016  19.11   19.55   18.86   19.53    73297418.00    1.00   (1.00) 1506.57
 3/14/2016  20.15   20.23   19.11   19.12   122947826.00    1.00    0.00  1539.58
 3/11/2016  20.24   20.58   19.84   19.98    85600278.00   (1.00)   0.00  1476.05
 3/10/2016  19.84   20.44   19.75   20.38    90204671.00   (1.00)  (1.00) 1447.64
  3/9/2016  19.01   19.72   18.81   19.51    79357761.00    1.00    1.00  1385.84
  3/8/2016  20.62   20.73   19.39   19.42   120144618.00    1.00    1.00  1379.45
  3/7/2016  20.00   20.73   19.92   20.40   109608538.00   (1.00)  (1.00) 1316.22
  3/4/2016  19.84   20.88   19.56   19.71   138290747.00    1.00    1.00  1271.70
  3/3/2016  19.16   20.02   19.10   19.82    97936124.00   (1.00)   0.00  1264.68
  3/2/2016  18.72   19.18   18.59   19.05    77623685.00    1.00   (1.00) 1317.95
  3/1/2016  19.44   19.50   18.52   18.57   119848346.00    1.00    1.00  1284.74
 2/29/2016  18.96   19.40   18.92   19.38    72095907.00   (1.00)   1.00  1340.78
 2/26/2016  18.95   19.34   18.64   18.69   109818142.00    1.00    1.00  1293.04
 2/25/2016  19.01   19.46   18.94   19.39    55242660.00   (1.00)   1.00  1341.47
 2/24/2016  19.44   19.85   18.80   19.11    90886017.00    1.00    1.00  1322.10
 2/23/2016  18.89   19.06   18.65   18.92    67141313.00    1.00   (1.00) 1335.51
 2/22/2016  17.82   18.69   17.75   18.53    49907659.00    1.00   (1.00) 1364.22
 2/19/2016  18.63   18.89   18.29   18.38    79024434.00    1.00    0.00  1353.18
 2/18/2016  17.52   18.91   17.44   18.90   106526884.00   (1.00)   0.00  1391.47
 2/17/2016  17.54   17.91   17.34   17.82    77754592.00    1.00    0.00  1311.95
 2/16/2016  17.94   18.31   17.20   17.21   142438923.00    1.00    1.00  1267.04
 2/12/2016  17.85   18.85   17.78   18.84    80692582.00   (1.00)  (1.00) 1166.15
 2/11/2016  18.38   18.65   17.78   18.36   168723889.00    1.00    1.00  1136.44
 2/10/2016  16.72   17.16   16.15   17.14    68760182.00    1.00   (1.00) 1223.53
  2/9/2016  17.72   17.82   16.75   16.75   100734573.00    1.00   (1.00) 1252.70
  2/8/2016  17.50   17.90   17.44   17.47   114317717.00   (1.00)   1.00  1306.54
  2/5/2016  15.75   17.12   15.57   17.05   113856701.00    1.00    0.00  1275.13
  2/4/2016  15.80   16.36   15.77   16.15   122759253.00    1.00    1.00  1207.82
  2/3/2016  14.44   15.35   14.42   15.35   117143236.00    1.00   (1.00) 1274.23
  2/2/2016  14.49   14.56   14.12   14.29    59458751.00    1.00    1.00  1186.24
  2/1/2016  14.33   14.65   14.30   14.65    48295585.00   (1.00)  (1.00) 1157.79
 1/29/2016  13.83   14.24   13.74   14.21    46570702.00    1.00   (1.00) 1194.78
 1/28/2016  13.87   14.18   13.73   13.86    44926613.00    1.00    0.00  1225.74
 1/27/2016  13.87   14.21   13.71   14.20    57258462.00   (1.00)  (1.00) 1197.07
 1/26/2016  13.53   14.00   13.52   13.97    69292956.00    1.00    1.00  1177.69
 1/25/2016  13.26   13.40   13.16   13.38    47896051.00    1.00    1.00  1127.95
 1/22/2016  12.73   13.17   12.69   13.03    30033286.00    1.00    1.00  1098.44
 1/21/2016  12.70   12.98   12.45   12.91    30793576.00    1.00   (1.00) 1108.75
 1/20/2016  12.78   12.92   12.52   12.85    57296457.00    1.00   (1.00) 1113.95
 1/19/2016  13.21   13.24   12.40   12.47    76419061.00    1.00   (1.00) 1148.96
 1/15/2016  13.60   13.61   13.06   13.08    35420238.00   (1.00)  (1.00) 1097.77
 1/14/2016  13.46   13.59   13.00   13.13    56266599.00   (1.00)   1.00  1101.96
 1/13/2016  13.54   13.80   13.36   13.61    38364928.00   (1.00)  (1.00) 1064.42
 1/12/2016  13.80   13.80   13.37   13.61    66902549.00    0.00   (1.00) 1064.42
 1/11/2016  14.57   14.61   13.82   13.92    69796877.00   (1.00)   0.00  1041.23
  1/8/2016  14.50   14.64   14.26   14.52    66017860.00   (1.00)   0.00   999.92
  1/7/2016  14.50   15.01   14.38   14.88    78644599.00   (1.00)   0.00   976.30
  1/6/2016  14.22   14.35   14.15   14.25    39864724.00    1.00   (1.00) 1021.45
  1/5/2016  14.13   14.15   13.91   14.02    29913361.00    1.00    1.00  1004.97
  1/4/2016  14.12   14.24   13.86   14.09    45197133.00   (1.00)  (1.00) 1000.00
 
Last edited:
Didn't have time to explain earlier.

The system performed much better in March than it did prior. This is because I did all of the optimizing using March data. Once the parameters were optimized, I used them to backtest through January.

It still made a profit every month, I believe.

The system dynamically adjusts to price action with internal optimizations; but some things can only be tested via trial and error. I don't want to code this part into the system yet because it would be a lot of work and only useful for backtesting.

I can manually do the optimizing going forward. And I'll next code that functionality into the system.

IOW, I believe that in reality, it'll perform more like it did in March during the test, going forward. :)
 
IMAGE: Old School (GDX):
iHrdML.png
 
SIGNAL: HOLD

I now have a corps of 20 variations of the algo tracking GDX. The one that performs the best gets to generate the official SIGNAL. However, I do see the signals of the others. They are mixed right now. About a third wanted to go to dust. I am still holding NUGT.

Code: 
04-06-16 ENTERED NUGT AT 59.15
03-30-16 ENTERED DUST AT 03.14---04-06-16 EXITED DUST AT 03.16 MAJOR upgrade
03-28-16 ENTERED NUGT AT 54.25---03-30-16 EXITED NUGT AT 60.38
03-21-16 ENTERED DUST AT 03.19---03-28-16 EXITED DUST AT 03.60
03-18-16 ENTERED NUGT AT 67.65---03-21-16 EXITED NUGT AT 65.20 optimized, fixed 1 bug
03-15-16 ENTERED DUST AT 03.83---03-18-16 EXITED DUST AT 03.08 minor system update, fixed 2 bugs
03-14-16 ENTERED NUGT AT 54.37---03-15-16 EXITED NUGT AT 56.80
03-11-16 ENTERED DUST AT 03.61---03-14-16 EXITED DUST AT 04.04 major system update
02-29-16 ENTERED NUGT AT 58.25---03-11-16 EXITED NUGT AT 61.59
02-24-16 ENTERED DUST AT 04.51---02-29-16 EXITED DUST AT 04.22
02-04-16 ENTERED NUGT AT 36.78---02-24-16 EXITED NUGT AT 55.95 major system update
02-02-16 ENTERED DUST AT 13.05---02-04-16 EXITED DUST AT 08.77
01-26-16 ENTERED TNA  AT 44.93---01-27-16 EXITED TNA  AT 44.85
01-25-16 ENTERED TZA  AT 61.15---01-26-16 EXITED TZA  AT 60.51
01-20-16 ENTERED TNA  AT 42.98---01-25-16 EXITED TNA  AT 44.79










_______________________________________________________________________________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
 
I should note this for the record:

I've analyzed historical signals from the algos and have seen where all 20 algos agreed on a direction, yet, they were all wrong. In these cases, the market was wrong. :)

I mean, something (news?) unpredictable (at least for my current algos) probably happened, imo.
 
SIGNAL: NUGT

Still holding nugt.

Code: 
04-06-16 ENTERED NUGT AT 59.15
03-30-16 ENTERED DUST AT 03.14---04-06-16 EXITED DUST AT 03.16 MAJOR upgrade
03-28-16 ENTERED NUGT AT 54.25---03-30-16 EXITED NUGT AT 60.38
03-21-16 ENTERED DUST AT 03.19---03-28-16 EXITED DUST AT 03.60
03-18-16 ENTERED NUGT AT 67.65---03-21-16 EXITED NUGT AT 65.20 optimized, fixed 1 bug
03-15-16 ENTERED DUST AT 03.83---03-18-16 EXITED DUST AT 03.08 minor system update, fixed 2 bugs
03-14-16 ENTERED NUGT AT 54.37---03-15-16 EXITED NUGT AT 56.80
03-11-16 ENTERED DUST AT 03.61---03-14-16 EXITED DUST AT 04.04 major system update
02-29-16 ENTERED NUGT AT 58.25---03-11-16 EXITED NUGT AT 61.59
02-24-16 ENTERED DUST AT 04.51---02-29-16 EXITED DUST AT 04.22
02-04-16 ENTERED NUGT AT 36.78---02-24-16 EXITED NUGT AT 55.95 major system update
02-02-16 ENTERED DUST AT 13.05---02-04-16 EXITED DUST AT 08.77
01-26-16 ENTERED TNA  AT 44.93---01-27-16 EXITED TNA  AT 44.85
01-25-16 ENTERED TZA  AT 61.15---01-26-16 EXITED TZA  AT 60.51
01-20-16 ENTERED TNA  AT 42.98---01-25-16 EXITED TNA  AT 44.79










_______________________________________________________________________________________________________________________________
DISCLAIMER: Userque's trading signals are not affiliated with TSP Talk or Buy Low Sell High, Inc. The information is for educational purposes only! The information is not advice or recommendations. The information may be revised at anytime. Userque does not give investment advice. Do not act on this data. Do not buy, sell or trade any stocks, ETF's, indices, funds, or any other instruments/vehicles based on this information. Userque may trade differently than discussed or posted in this forum.
 
Small Status Update:

After I finish optimizing the tsp system and adding code to automate running multiple models:

I want to test the algo with the volatility ETF's (TVIX, XIV, etc.). They are extremely unique instruments and I wonder how well the algo can predict them.
 
userque said:
Small Status Update:

After I finish optimizing the tsp system and adding code to automate running multiple models:

I want to test the algo with the volatility ETF's (TVIX, XIV, etc.). They are extremely unique instruments and I wonder how well the algo can predict them.
Click to expand...

Will you be running it based on those ETN/ETF charts or what they track (like GDX[GDM equiv] for NUGT/DUST)? I believe it makes a difference and each of them do volatility very differently.
 
RealMoneyIssues said:
Will you be running it based on those ETN/ETF charts or what they track (like GDX[GDM equiv] for NUGT/DUST)? I believe it makes a difference and each of them do volatility very differently.
Click to expand...

I'll use just one for both long and short.

For the system, it doesn't matter how the volatility is calculated. What matters is whether they are correlated (or inversely correlated).

If two instruments are 100% correlated, their peaks and valleys match. A signal for one is the same signal for the other. Just as GDX is correlated to NUGT; and inversely correlated to DUST.
 
userque said:
I'll use just one for both long and short.

For the system, it doesn't matter how the volatility is calculated. What matters is whether they are correlated (or inversely correlated).

If two instruments are 100% correlated, their peaks and valleys match. A signal for one is the same signal for the other. Just as GDX is correlated to NUGT; and inversely correlated to DUST.
Click to expand...

Will be interesting to see what ETF/ETN's you find for volatility similar to the gold miner pairing; I don't remember seeing any that correlated that nicely.
 
RealMoneyIssues said:
Will be interesting to see what ETF/ETN's you find for volatility similar to the gold miner pairing; I don't remember seeing any that correlated that nicely.
Click to expand...

No. :)

I load GDX into the system to trade NUGT/DUST.

I could load MSFT to trade MSFT.

I could load TVIX to trade TVIX/XIV
 
GDX is correlated to NUGT/DUST, so I can use one GDX model instead of two NUGT and DUST models.

So on and so forth.
 
You must log in or register to reply here.

Similar threads

userque
  • Article Article
Introduction
Replies
0
Views
2K
userque
userque
userque
  • Article Article
userque & NUGT/DUST
Replies
0
Views
6K
userque
userque
E
Neural Network Artificial Intelligence System (No Chat)
4 5 6
Replies
104
Views
14K
Birchtree
Birchtree
350zCommTech
Citizens For Financial Responsibility Petition
Replies
9
Views
2K
350zCommTech
350zCommTech
mlk_man
Investment Trading System Scams Exposed
Replies
4
Views
11K
remmy
R
Back
Top