This is a mobile optimized page that loads fast, if you want to load the real page, click this text.

qldfrog weekly Skate inspired system

Jump to new
More journal like thoughts:
daily volatility systems will be left alone..wait and see

The weekly side:
Our new Sector influenced QFSec:
of the 8 purchases last monday, we lost 13% average during the week, so without opportunity to get out yet
a real disaster as we also lost open profit on top...
It is an issue I have no easy answer as the fall was brutal and BT matches real trade..so what next, on this on, I think I should carry on

I understand BT means sxxt but look at the difference of smothness lately:

here as well down trend since early 21...unsure what to do there, reduce amount per entry until when???

Lastly, QFDuc the old boy running various version since 05/2020, even the current version goes nowhere since2021 and especially since mid april.


I suspect I can improve the results by removing some of the version past "improvements..."
This is really underperforming under the current conditions
 
And for QFDuc, one tested version was more stable and better on the last period, so still good if we have a rerun of crash boom, but handling current situation;
Let's use it; it is not new and has been tested previously..and still a qfduc
and let's stick with it
 
Ouch, seems like a tough period for your systems. They all go through that.

You mentioned a Guppy system. Is that one of Daryl Guppy's systems? Haven't heard that name in a very long time. What's he up to these days? For a very long time he was always talking up his "Guppy Indicator"--which was a whole bunch of different period MAs. Are you trading his Guppy Indicator?
 
@MovingAverage
FYI and there's a few more posts on the subject matter.

 
Not exactly his system, got my tweaks etc but the basic idea was provided by Skate post as indicated above by @frugal.rock ??
My variation on a daily version But still trend and Guppy concepts so different from other systemsaleays this diversification attempt which did not work obviously last week
 
So what were the decision and where do I stand by this morning?
the 2 weekly
QFDuc reverted to version 21 (-5 )which seems more suited to current market, some buy and sell today
QFSec: I will persevere with the current code, going nowhere in 2021 but positive in backtests since 2016..but there was a nasty period early 2016 and the current version if I had played it from start would have been OK
we carry on unchanged.no buy on this week (filter) but sells

The weekly:
Both Guppy and the Flying Bat had proper MC, decent but not flash long term returns on tests we are still going on, but I will reduce the amount invested while we are in a negative run on DL guppy..DL guppy has been losing in the last 3 months..But I still believe it has potential..but not in that market
 
Today was ok, 47 trades so half a grant in brokerage, and ahead.
But another day with more crash following SPPa few last week, and today POSin the investment side, Grange( GRR) trigered my trailing SL with 15% fall in the last few days
So quite a lot of up and down even on good shares
 
Today was ok, 47 trades so half a grant in brokerage, and ahead.
But another day with more crash following SPP
a few last week, and today POS
in the investment side, Grange( GRR) trigered my trailing SL with 15% fall in the last few days
So quite a lot of up and down even on good shares
 
OK week system wise, not so good with currency loses on the NY portfolio and on my investment side but happy with the systems picking up:
XNT up 1.9% so that is our reference:
Daily:
FB up 0.6 % or 0.3k, now fully invested
Guppy: packet reduced by half until I can see a positive trend after last week disaster
; But this cost $ this week as Guppy was up:
+2.9% or +$1.3k 50% cash due to reduced packets
volatility ASX:
+2.15% or +$1.1k ,full cash tonight
volatility US:
-1.4% or -$0.2k ,full cash tonight

weekly: a bit disappointing
QFSec: -1.5% or -$1.1k only 23% invested
QFDuc:+0.8$ or +$0.7k

overall for the week:
+$2.1k or +0.6% with 50% cash as of tonight
obviously we lost nearly 1.5k by reducing risk on Guppy but I will wait another positive week before reverting to full size packets
I missed a few good entries LKE etc as they went past my buy order so will try to model that in my backtest as this happens often
 
Hi Frog, you've reduced your exposure on guppy--what's your general approach to adjusting your exposure?
 
MovingAverage said:
Hi Frog, you've reduced your exposure on guppy--what's your general approach to adjusting your exposure?
Click to expand...
No general approach, it is a first and so not exact science
Guppy has had 3 months of ongoing not insignificant losses with the current condition/realm.Both Backtests/real trades confirmed
It is a decent system : was and will be again,especially with this running version;
So I try to stop the bleeding/reduce $ losses;
if I get 3 weeks (arbitrary) of above xnt perf, i will reinstall full size parcel.
Not real science but the current underperformance is real, and real trade is the only way I can realistically check any condition change;
with 2k parcels, I am down to lower exposure..This week was nearly 3% up, well above XNT
Another 2 over performances and i will revert
 
AB help required:
In real life, some shooting start buys are missed as my:
last close+3% (for example) buy values at open are overshot;
I would like to model that in backtests to ensure backtests are as near real results as possible:
so a Buy if O<ref(C,-1)*1.03
I can do that but in cases like that we want to ensure this buy is missed not just replaced by the next one based on the positionscore
so need to reduce on that day the total number of buys by one
And obviously this should not interfere with explore results;
If someone has already coded this , I would be grateful, otherwise i will do the yard yakka...
And share with you on demand
 
inspired by the LKE missed last week which had a 47c to 66c jump in a week and that my backtest will show as taken...which it was sadly not...
 
Hi @qldfrog,

Any code suggested in this post is in general terms only and not specifically AB code.

Firstly your buy code should be along the lines of O<=Ref(C,-1)*1.03 rounded down to comply with the following ASX requirements:-

Up to 10c minimum increment = 0.1c
10c up to $2.00 minimum increment = 0.5c
$2.00 up to $99,999,990.00 minimum increment = 1c.

Unless I am missing something critical, the following should be the case, as there is no "replacement stock" to be considered.

Given that you are using "positionscore" in your code and that you can calculate the number of system positions
available, then only that number of stocks would be allowed to participate in the opening auction for the next trading day.

Cheers, Rob
 
Thanks Rob,
Ok with rounding, not a real issue.
Let's say there are 10 free packets to buy on day D
If a packet shots up at open, the condition will remove it from the backtest buy..with future leak obviously but all good
If i do nothing else, another packet will replace that missed one so position score 11 will be selected and added in the backtest, i need to "dynamically" decrease the number of packets purchased but only on that day..
It is an area i do not often code so not so sure...
I could try that with a few tests so not impossible, but not That simple and wanted to leverage any experience in that area.
It is not very simple to google that issue as a bit difficult to phrase the issue...
 

Just saw this.

One interesting method I have used a while ago is to trade the equity curve.

(1) Shut down or adjust your positions as you wish
(2) In the background paper trade full positions as you would normally. Plot your curve.
(3) Revert to normal trading when the Curve exhibits bullish-ness has returned.

There is a number of ways of determining a change of sentiment with the curve.
 
Thanks Tech/a, actually doing that the lazy way by looking at the backtest run using normal size position vs actual.
Obviously, it then becomes important that backtest match real world so, human error excepted, that should do the trick.
This triggered one of the following posts on how to have BT matching real world better to take into account the , not that infrequent cases where i miss entries as open shot up past my buy...
.There is still a lot of human intervention when deciding that conditions are back bullish for a system but i can decide something arbitrary.
Thanks for your input.and might definitively use it.
 
not that infrequent cases where i miss entries as open shot up past my buy...

Proper Montecarlo Testing should help the curve.
 
tech/a said:
not that infrequent cases where i miss entries as open shot up past my buy...

Proper Montecarlo Testing should help the curve.
Click to expand...
true and MC actually done on the actual system under reduced sails if I remember well, not AB MC but real not taking 20% of of Buy using fake optimised on 1000 runs.
And true I could MC on the latest shorter period.Thanks for the alternate thinking..always hard when on your screen alone to take the step back
 
I have recently switched from Amibroker to Realtest, partly because of the problem you have pointed out.
Realtest allows you to specify the maximum number of setups that can be considered for entry on any one day so if one packet is above the entry limit it will not be replaced by another with a lower position rank.
 
You must log in or register to reply here.

Similar threads

Trading systems/methods guidance
Replies
14
Views
6K
rnr
R
Emergency Exit for Weekly System
2
Replies
27
Views
5K
MovingAverage
Swing System
Replies
19
Views
5K
MovingAverage
Weekly Breakout Strategy For QQQ ETF
Replies
17
Views
5K
KevinBB
K
System optimisation: IS training vs. OOS testing, and Paper Trading
Replies
10
Views
2K
Garpal Gumnut
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more...