qldfrog weekly Skate inspired system

qldfrog · 23 April 2020

Skate said:
You are correct
@qldfrog as you are using Premium Data, "Norgate's old legacy format" (MetaStock format) you are using the correct format for the All Ordinaries "XAO".

Correction
I was referring to "Norgate Data Users" not "Premium Data Users"

The problem with a bit of history
If you subscribed to (NDU) both Australian + US stocks originally you would have had a .AU suffix. If you were an Australian-only subscriber then there would be no .AU suffix. This optional suffix arrangement resulted in causing an issue, especially when a AU+US subscriber dropped a US subscription then all of their Australian stocks lost their suffix. It broke various watchlists that had been created, it also broke the formulas that were relying on a suffix being present. So to overcome this problem about 18 months ago Norgate made the .AU suffix mandatory for all new subscribers of Australian data to avoid the confusion.

Using Norgate Data (NDU) instead of Premium Data
With AmiBroker significantly-enhanced capabilities Norgate Data (NDU) capitalises on this by adding additional features well beyond those in Premium Data, so it pays to migrate over.

There is a catch (with regards to historical data)
There is one very important difference, with my original Premium Data I purchased historical data separately. NDU works as a subscription-only model and historical data cannot be purchased separately. The historical data is in the old legacy format (MetaStock format) meaning, they are not compatible.

Skate.

Thanks for the precision as my subscription renewal time approaches and I am using the data much more, I might bite the bullet and upgrade? to NDU

jjbinks · 23 April 2020

qldfrog said:
Thanks for the precision as my subscription renewal time approaches and I am using the data much more, I might bite the bullet and upgrade? to NDU

You will still have access to your old premium data.
Maybe back up the most recent premium data updater before u switch

qldfrog · 23 April 2020

jjbinks said:
You will still have access to your old premium data.
Maybe back up the most recent premium data updater before u switch

@jjbinks : is it worthwhile in your opinion, I understand the access to historical data but any other advantages?

jjbinks · 23 April 2020

For me main advantage was historical data.
Also norgate data can be linked witn Python. Still a work in progress for me with python

qldfrog · 24 April 2020

I completed major system rework yesterday, refining GTFO test of a pretty decent Duc inspired 3rd systems etc;
The result was that the 2 existing reworked system where still out after the crash.
If I want to follow a system, better follow it, and this work has definitively improved the current systems.
I decided to reset and sold all shares purchased under the older systems in the last 2/3 weeks
Phoenix is reset with a $70 loss inc brokerage;
While ZLextra is in profit of $1031 after brokerage.
This gives me an extra 2 days to make sure my current code is robust, and having it double checked as well;
Thanks you know who!!!;
So we are back to fully cash portfolios:
Phoenix @$89k
ZLextra: @$91.2k
As a timely reminder, were at 100 and 100.5 in February when we started last year so we lost a raw 20k (10%) from our capital in that crash ;
Not too bad even if we actually lost some of our paper profits as well.
Disclaimer, these are not my only investment but I think it is worthwhile putting things in perspective

qldfrog · 2 May 2020

May here we are, the month when we should just stay put.
System wise, I have 3 systems:
a QFDuc inspired by some of @ducati916 comments, it has nothing to do with @Skate efforts, I have had no insider knowledge or even bibliographic , theory etc base
was just triggered into trying to leverage volatility , volume and price into a system;
It is relatively highly responsive and has results quite divergents from my other systems, can be profitable when other are not etc.
ZL extra based on Ehler ZeroLag plus own sauce:interesting
And Phoenix, the grandson of @Skate Cam based idea after many generations and tweak
The trusted old truck with less than stellar performance
I had a talk with Mrs QldFrog and compared various behaviours as seen in backtest, and we reached a consensus:
I will only run 2 weekly systems and I have eliminated phoenix.
Exciting road ahead as I will run from now on both ZLextra and QFDuc
I am sure the code will be updated as time goes but these are the new seeds
Both starting from 90k portfolio and started 01/05/2020

Warr87 · 2 May 2020

good luck mate. i'll be watching intently.

qldfrog · 2 May 2020

what's on Monday?
QFDuc has 4 buys
ZLextra 3 buys
For info Phoenix has 7 buy: one shared with ZL, 2 with ZL
I do not feel especially good with the market so will check these buys against a daily system just to get a better feel

qldfrog · 4 May 2020

QFDuc purchased:
FPH 167 @$26.26
GSW 5844 @$0.76
SAR 1050 @$4.19
TNE 466 @$9.51

ZLextra bought:
AHY 4347 @$1.04
DHG 1724 @ $2.52
DMP 79 @ $55.05
nice surge on GSW after the purchase, a good start
my daily toddler system did not engage yet
Have a great week

qldfrog · 8 May 2020

End of a week which was OK:
Both overtaken by XAOA but these are just starting to be invested

ZLExtra started 1/05/2020
Total invested $90,000.00
Cash remaining $76,777.43
Current portfolio value $90,653.01
Profit $ 653.01
Nb of active positions 3
Nb max position 20
Return since start 0.73%
Annual return so far 37.83%
purchase value per position $4,532.65
Invested percentage 15%
On Monday-> buy 9
and
QFDuc started 1/05/2020
Total invested $ 90,009.85
Cash remaining $ 72,311.83
Current portfolio value $ 91,108.93
Profit $ 1,099.08
Nb of active positions 4
Nb max position 20
Return since start 1.22%
Annual return so far 63.67%
purchase value per position $4,555.45
Invested percentage 21%
dividends $ -
Average per week: 549.54
invested 18,797.10
On Monday-> sell 2, buy 12

qldfrog · 8 May 2020

now I want to put a small story about my daily system,
which has been a fail and back to the draw-board.
Luckily not a major financial hit but nevertheless a fail:
I worked for a while on the idea that in times like now, weekly systems are too slow to react to volatile conditions.
SO with time on my side, let's go daily;

I design, back-test, confirm an edge, then here I go:

this week: first trades for Buy on Tuesday,
Wednesday hum weird, that explore list does not seem to match yesterday one,
Thursday, yeap it is confirmed there is "repainting"
running replays in the past does not give the same signal than initially received..a sure sign of future looking code
Nothing wrong in itself, but uncheckable-> no backtest is meaningful
I did the code check before going live, have the shift by one day for buy/sell but definitively not right
today was the great unwinding:
6 positions still active, paper loss of $700 on these at market close and profit of $200 for liquidated positions, [including brokerage] so not a disaster but still a disappointment
I hopefully will find the flaw this week end to restart anew on Monday

qldfrog · 11 May 2020

today: 20 trades for @Skate's experiment
Then my systems:
QFDuc:
we sold
SAR@ $4.46: profit of $283
FPH@ $27.89profit of $272

Buy:
APT 109 @ $39.90
APX 145 @ $30.49
CDV 10344 @$0.43
ELD 474 @$9.28
FMG 362 @$12.30
KGN 502 @$8.70
MP1 309 @ $14.50
PNV 1717 @ $2.59
PPH 722 @$6.08
We end up with10 positions and 54% invested

ZLextra system:
no sell
we buy:
FPH 156 @ $27.89
PFP 1451@ $3.09
PRU 4128 @ $1.06
RMD 179@ $24.88

ducati916 · 12 May 2020

qldfrog said:
now I want to put a small story about my daily system,
which has been a fail and back to the draw-board.
Luckily not a major financial hit but nevertheless a fail:
I worked for a while on the idea that in times like now, weekly systems are too slow to react to volatile conditions.
SO with time on my side, let's go daily;

I design, back-test, confirm an edge, then here I go:

this week: first trades for Buy on Tuesday,
Wednesday hum weird, that explore list does not seem to match yesterday one,
Thursday, yeap it is confirmed there is "repainting"
running replays in the past does not give the same signal than initially received..a sure sign of future looking code
Nothing wrong in itself, but uncheckable-> no backtest is meaningful
I did the code check before going live, have the shift by one day for buy/sell but definitively not right
today was the great unwinding:
6 positions still active, paper loss of $700 on these at market close and profit of $200 for liquidated positions, [including brokerage] so not a disaster but still a disappointment
I hopefully will find the flaw this week end to restart anew on Monday

Mr Frog,

Possibly consider moving away from 'all systems at all times' to a more nuanced position of: appropriate system for market conditions.

So system (a) is daily; and system (b) is weekly. In high volatility environments, system (a) would (should) have better results with lower risk: faster entries/exits, constantly re-evaluating environment.

When the market returns to a trending environment, system (b) can take advantage.

Now it doesn't have to be either or (100% (a) or (b)), you might go 75%/25% and morph back/forward with the systems. That way, the system provides objective feedback in dollars, what the market conditions actually are, rather than a more subjective assessment that you might make through personal observation.

jog on
duc

qldfrog · 12 May 2020

ducati916 said:
Mr Frog,

Possibly consider moving away from 'all systems at all times' to a more nuanced position of: appropriate system for market conditions.

So system (a) is daily; and system (b) is weekly. In high volatility environments, system (a) would (should) have better results with lower risk: faster entries/exits, constantly re-evaluating environment.

When the market returns to a trending environment, system (b) can take advantage.

Now it doesn't have to be either or (100% (a) or (b)), you might go 75%/25% and morph back/forward with the systems. That way, the system provides objective feedback in dollars, what the market conditions actually are, rather than a more subjective assessment that you might make through personal observation.

jog on
duc

Spot on, so my desire to start the daily ASAP, but I was a bit too eager last week as there was an uncovered technical glitch and most probably a future looking code..my only explanation to what I saw, i rechecked the code with slight modification and am paper checking the daily system to ensure this is not happening again, and hopefully start next week.
I trust figures and code more than my own instincts in investing
I have to say that the timing of system entry is critical as I and many other have found and this is adding resistance in me taking on Mr Skate experiment: with already 2 weekly systems, mostly trend based in the month, adding 20k is not the wisest..
Then I could be lucky but let's not be fooled , this is dreadful risk management, on the other end, 20k is just 10% of the now 3 systems, is not that humongous and my own systems take time to ramp up so I jumped in....
Still keen to start the daily system venture ASAP

mostly

qldfrog · 12 May 2020

ducati916 said:
Mr Frog,

Possibly consider moving away from 'all systems at all times' to a more nuanced position of: appropriate system for market conditions.

So system (a) is daily; and system (b) is weekly. In high volatility environments, system (a) would (should) have better results with lower risk: faster entries/exits, constantly re-evaluating environment.

When the market returns to a trending environment, system (b) can take advantage.

Now it doesn't have to be either or (100% (a) or (b)), you might go 75%/25% and morph back/forward with the systems. That way, the system provides objective feedback in dollars, what the market conditions actually are, rather than a more subjective assessment that you might make through personal observation.

jog on
duc

And to be clear, by starting multiple systems (both weekly and different codes), I expect the $ returns to tell me quickly which kind is better for a given condition;
An interesting point would be: do you re-balance them, how and when (assuming systems are only at most loosely correlated)
System A gains 20% after a year,
System B gain 120% after a year,
System C loses 30% after a year,
What now? reduce A and B to re-float C?
=> what do you do? re-balance funds at a third after a year?
interesting point.Not there yet but will reach that stage..and when to act? yearly ? a smoother more continuous approach?
I will keep busy in the next 20y!!!

Warr87 · 12 May 2020

ducati916 said:
Mr Frog,

Possibly consider moving away from 'all systems at all times' to a more nuanced position of: appropriate system for market conditions.

So system (a) is daily; and system (b) is weekly. In high volatility environments, system (a) would (should) have better results with lower risk: faster entries/exits, constantly re-evaluating environment.

When the market returns to a trending environment, system (b) can take advantage.

Now it doesn't have to be either or (100% (a) or (b)), you might go 75%/25% and morph back/forward with the systems. That way, the system provides objective feedback in dollars, what the market conditions actually are, rather than a more subjective assessment that you might make through personal observation.

jog on
duc

This would be portfolio optimisation. Something I wish was possible in AB.

I would suspect you would alter the weights on the systems based on a filter. This would reduce position size during a down-market or alter the allocation to each system (the 75/25 split).

Not sure how we do this but I am very interested to know.

Lone Wolf · 12 May 2020

qldfrog said:
And to be clear, by starting multiple systems (both weekly and different codes), I expect the $ returns to tell me quickly which kind is better for a given condition;
An interesting point would be: do you re-balance them, how and when (assuming systems are only at most loosely correlated)
System A gains 20% after a year,
System B gain 120% after a year,
System C loses 30% after a year,
What now? reduce A and B to re-float C?
=> what do you do? re-balance funds at a third after a year?
interesting point.Not there yet but will reach that stage..and when to act? yearly ? a smoother more continuous approach?
I will keep busy in the next 20y!!!

One of the issues with directing funds towards the system that's currently performing better is that you don't know which system will perform better next. It's a bit like an index filter. It takes time to confirm the index is falling and switch the system off. Then it takes time to confirm the up trend has resumed. This can actually hurt system performance as sometimes you ride most of the way down and miss out on much of the recovery. Not saying it's a bad idea, just not easy.

There has been talk of using the equity curve as a filter. You might have a couple systems, you focus your capital on the system with the strongest equity curve. (you run both systems in sim mode in parallel to your real trading account so as not to affect their equity curves) The idea being, what's the best indicator to tell you whether market conditions are right for your system? The equity curve. However, the same issue remains. By the time you determine the conditions are not right for the system, conditions could change.

Lone Wolf · 13 May 2020

It's getting too late. My point was going to be that what you've discussed about altering capital allocation does go some way towards fixing the issue with the lag. I like what Peter2 does with changing position sizes based on market conditions rather than a hard on/off switch. The equity curve idea was first presented to me an an "equity curve switch" to be used on a single system. In that format I'm not sure it would do better than any other index filter.

qldfrog · 13 May 2020

Lone Wolf said:
It's getting too late. My point was going to be that what you've discussed about altering capital allocation does go some way towards fixing the issue with the lag. I like what Peter2 does with changing position sizes based on market conditions rather than a hard on/off switch. The equity curve idea was first presented to me an an "equity curve switch" to be used on a single system. In that format I'm not sure it would do better than any other index filter.

No worries, if there was an easy solution, outside the crystal ball, we would not even discuss it;
In a nutshell, to sort that issue, we have:
#capital reallocation between systems
[
and should this be encourage the winner or the looser?
Both choice could be a sensible choice in the absence of a crystal ball if we assume that all systems are not faulty just not optimum on the same market]
#position size changed based on a parameter such as market index, volatility, volume etc;
# using each system equity curve to guide one of the two choices above
Or a mix/mess of the above.
Mr Skate, as I understand him, is in the belief that a robust system will not get obsolete.I would tend to respectfully disagree , at the very least in term of efficiency/optimum return;
As mentioned already somewhere, some of my systems under development were systemically offering consistent out performance against other in a way closely matching test period age:
See backtest figures belowjust for actual $ figures after different start year of 100k, DD and annual returns similar .

Sadly and realistically, I think we can only detect an obsolete system too late, and might even close it just as it becomes relevant again.
Knowing our limits, should we base our benchmark on professional funds returns maybe...
This discussion much appreciated

qldfrog · 13 May 2020

Today was special in that i did my first system override: one
of my weekly buy AQG (alacer gold miner).
There was a trading halt and a merge proposal with a Canadian firm.i got burnt last year in such a similar issue and do not want to see my shares locked for weeks or worse listed in Canada i did exit.. with a profit..today

qldfrog weekly Skate inspired system

Similar threads

Privacy & Transparency

Privacy & Transparency