qldfrog weekly Skate inspired system

[qldfrog]

I will reply later, with the rain plenty to do on the block
I will check the flipper parameters , execution etc
As to why slipper? I wanted to have an ultra simple and validated system in place for a small amount i had aside :80k
Trying to avoid spending too much time on it, as the cam related systems need refining,while checking on paper what an abcd/1234 code would produce
A kind of useful carpark ....

 

[jjbinks]

@Newt
I'll need to have another look at Radge's book but for me 2 reasons why flipper seems nice
1)Fairly simple concept
2)If i remember correctly Flipper had shortest periods of draw downs/quickest to positive returns. So from this perspective I felt it is a good starting point as a beginner. (I though I would lose confidence in my system if it had a fairly long period of draw down. )

 

[Newt]

@Newt
I'll need to have another look at Radge's book but for me 2 reasons why flipper seems nice
1)Fairly simple concept
2)If i remember correctly Flipper had shortest periods of draw downs/quickest to positive returns. So from this perspective I felt it is a good starting point as a beginner. (I though I would lose confidence in my system if it had a fairly long period of draw down. )

Thanks JJB
I'll have to dig out the "Holy Tome" and re-read

 

[qldfrog]

Now that the storm is over and internet back:

About the flipper: I started buying on 16/12 so started run from 13/12/2019

The backtest is as follow:




I deviated by not buying NVL [from memory a takeover target]

The backtest during that period has a loss of $2147



My own results are





So in term of following code, all good;

For info, running that same code last year produces: using $80k on calendar year




For comparison:

The xaoa went from 57954 to 72371

An increase of 24.8%..so basically flipper returned a bit above the xaoa, and is not too linked to the other systems..and I know if market goes down, flipper can go down..



So “kind of all good” but I agree no stellar performance expected

 

[rnr]

Hi @qldfrog,

From my understanding of the system, excluding the Index filter at this stage, there are only 2 trades that have signaled a weekly entry with scans commencing on 13/12/2019.

Cheers,
Rob

 

[qldfrog]

Tha

Hi @qldfrog,

From my understanding of the system, excluding the Index filter at this stage, there are only 2 trades that have signaled a weekly entry with scans commencing on 13/12/2019.

Cheers,
Rob

Thanks Rob, that is interesting and would mean my code does not match the index methodology.
I have just started rereading the unholy grail yesterday.
As mentioned a bit earlier, i used this third system as an easy simple alternative, and went straight into code i did years ago and some i pinched from the internet.
You could very well be right.better look at the code itself

 

[qldfrog]

Hi @qldfrog,

From my understanding of the system, excluding the Index filter at this stage, there are only 2 trades that have signaled a weekly entry with scans commencing on 13/12/2019.

Cheers,
Rob

@rnr Rob, I cleared the code conditions etc, (looked at >$2 as per Radge..which I do not filter on in my own realm) and with that I still got quite a few entries, more actually
IMF, CDA, MVP for example : I looked at the chart and yes 20% above is triggered for these 3
we must use a different meaning of Low/High? do use use Low of closes and high of closes?
Only if interested in this
That would easily explain, and obviously how far back you go to define the low periods for buys (13/12 or before)
Anyway, what I mean is that this flipper based system is not an initial winner in the last month..
would make sense when you think of it:
short term you jump on the good news/days..and it will take a while to either clear out the duds and carry on (we are a momentum system still) to get the gains,
Actually, the ideal time to get into a flipper is at a trend reversion, when going from bear to bull market, after, we just follow the bull move so probably not a great way to park cash

 

[rnr]

@rnr Rob, I cleared the code conditions etc, (looked at >$2 as per Radge..which I do not filter on in my own realm) and with that I still got quite a few entries, more actually
IMF, CDA, MVP for example : I looked at the chart and yes 20% above is triggered for these 3
we must use a different meaning of Low/High? do use use Low of closes and high of closes?
Only if interested in this
That would easily explain, and obviously how far back you go to define the low periods for buys (13/12 or before)
Anyway, what I mean is that this flipper based system is not an initial winner in the last month..
would make sense when you think of it:
short term you jump on the good news/days..and it will take a while to either clear out the duds and carry on (we are a momentum system still) to get the gains,
Actually, the ideal time to get into a flipper is at a trend reversion, when going from bear to bull market, after, we just follow the bull move so probably not a great way to park cash

Hi @qldfrog,

It's excellent to hear that you are getting some rain over there.

The systems published in Nick's book "Unholy Grails" were designed with daily parameters so when testing over on a weekly basis these parameters need to be adjusted accordingly.
When comparing these systems with other parties, the conversion to weekly parameters may well cause inconsistencies between results.

My interpretation of this system revolves around a 20% HiLo Peak-to-Valley ZigZag indicator which determines the entries and exits as per the pic below.



A blue signal bar (high has moved up a minimum of 20% from the previous zigzag low - i.e BUY condition met) indicating a BUY at next OPEN and a red signal bar (low has moved down a minimum of 20% from the previous zigzag high - i.e. SELL condition met) indicating a SELL at next OPEN.

Other filters are ignored in this example.

And for the record the BUY and SELL signals do not look forward and can therefore be used for back-testing.

Cheers,
Rob

 

[qldfrog]

a bit of activity today:
system1 bought 2766 CAT @$2, 1049 EML @$5.31, 2603 FWD @$2.15 reaching fully invested status
System2 bought 810 LYL @$6.2, 1760 NEU@$2.76, 5092 PSI@$3.3 and 409 ZIM@11.73..That last one was a battle as the SP shot up from open and I only got 4.8k and not $5.5 even after manually chasing a little bit.
System 2 sold HTA @13c and SHV @8.42 for a loss of $955 inc brokerage costs
This HTA is going to go into my blacklist ..it is below the 20c limit that I do not want to implement as it seems to be detrimental on the XAO realm, but I am really doomed with that share

System3 (the flipper is under a bit of a review as per previous posts), I had a buy order on MSB but it shot straight up and I did not get any
System 3 sold FLC @42c for a loss of $215
Once the flipper review is completed, I will either close it or carry on after adjustments

 

[qldfrog]

about 20c limit, I will from today use the 40c limit..seems more beneficial than negative and would avoid basic increments being too big in percentage, we do not want to deal with these anyway...

 

[qldfrog]

A few decisions today: i will implement the 40c SP limit on the xao world for all systems, i i ha a few below 40c entries and i i wi let these run their course
The flipper system will be liquidated in the coming days
Not so much as for it being a bad system, but was a too quick harsh and not thought enough code
Not enough backtests, checks etc
Never rush if there is a lesson
I will probably replace it with a zigzag,flipper and own sauce later on but only if i can spend enough time to do a decent work

 

[tech/a]

Warning

Do not use the Zig Zag indicator in systems.

It is important to remember that the ZigZag feature has no predictive power because it draws lines based on hindsight.
It cannot signal until the low or high % value is reached.
So a system will always be looking backward in some cases weeks if using weekly.
So you don't know that a high or low is going to hold to a given % in real time.

 

[qldfrog]

Thanks @tech/a

Warning

Do not use the Zig Zag indicator in systems.

It is important to remember that the ZigZag feature has no predictive power because it draws lines based on hindsight.
It cannot signal until the low or high % value is reached.
So a system will always be looking backward in some cases weeks if using weekly.
So you don't know that a high or low is going to hold to a given % in real time.

Indeed, can be a miracle in backtests
Too good to be true.....
Following some leads from @rnr , i dig into the AB Zig function yesterday

Then fully aware of the mostly future looking of the Zig function, i(it is highlighted in the AB pages ) i use one of the recommended tweak to reduce that risk, and i only usethe zig in conjunction with a flipper style % of high low
Previous high and low points determined by the tweaked zigzag instead of tested on an arbitrary past period in my initial code

I have to say that my current task is really :
Does my code future leak or not?
If i can be confident that it is not leaking, then i will proceed but for the time being i will paper run and liquidate existing flipper attempt That was just stupidity from my part.

I should know better at my age...
Your inputs much appreciated

 

[qldfrog]

A standalone post about buying at open for systems users;
I use bell direct(very happy) and did order a buy on open at 3.3, order sent in the week end
That order was rejected on Monday morning and I was given a 3.26 indicative pricing; ultimately open was at 3.29 from memory and i had to chase the price the price to have my order executed
I asked BellDirect what is their formula so that I can write it and ensure safe Monday morning open orders
This is the answer from BD: I find it interesting:
I can advise that the order was sent off for manual review and was rejected by a designated trading representative as the trade was going to move the market too much in one order. The rejection order saying ‘price discovery at $3.26’ means that to place the higher order you will first need to place the order at 3.26 and then slowly move up to 3.30, allowing time for the market to react to your orders.



Please be advised that we have to abide by the ASX rules and regulations when accepting and manually checking orders for the market.

What it means in a nutshell is that your week end generated orders may get rejected just before the open..and you need to be actually active and logged in at 9/10AM on Monday morning
Not what I wanted to hear, even if this does not happen too often

 

[rnr]

Warning

Do not use the Zig Zag indicator in systems.

It is important to remember that the ZigZag feature has no predictive power because it draws lines based on hindsight.
It cannot signal until the low or high % value is reached.
So a system will always be looking backward in some cases weeks if using weekly.
So you don't know that a high or low is going to hold to a given % in real time.

Hi @tech/a ,

Let's just qualify that statement to read

Do not use the Zig Zag indicator when back-testing a system UNLESS you link the Entries and Exits with a % move validation indicator" as the ZigZag indicator on its own effectively looks into the future.

See below for an example.

Cheers,
Rob



or

https://www.aussiestockforums.com/t...tfolio-hypothetical.34661/page-2#post-1030875

 

[tech/a]

You may find the 20% Flipper or ( Whatever % ) System maybe more appropriate
for a Universe of stocks which are not Volatile (generally). Larger caps.
Nothing under $1 and $3 even better.

Just saying.

 

[Skate]

@qldfrog - @captain black often spoke of the ZigZag function for use in the development of his systems & went to great lengths to explain the ZigZag function should never used in a trading strategy because the signals keep repainting when more data is received. (Football analogy - the goal posts keep shifting with additional data)

ZigZag
The ZigZag indicator can be used to reveal the profit potential for different stocks & in different time frames though. Using the ZigZag function gives you a sense of where the profits are & what happens when you reduce trade duration.

Important
The ZigZag function looks into the future & of course, should never be used for trading with real money. Use the ZigZag function in your strategy development phase - use it as a guide to how accurate your signal are in relationship to the pivot points.

Skate.

 

[qldfrog]

You may find the 20% Flipper or ( Whatever % ) System maybe more appropriate
for a Universe of stocks which are not Volatile (generally). Larger caps.
Nothing under $1 and $3 even better.

Just saying.

I noticed better results indeed just moving my price threshold to 40c from 20c, and targetting restricted xao
Plenty to try

 

[tech/a]

Trading Pennies is like catching Flies.
You need the skill of a Kung Fu Master.

Trading Blue chips is like catching an Elephant
Even I can move that fast!

 

[qldfrog]

This one is a gem @tech/a
About zigzag and future leaks
Indeed the Zi() function is a well known future looking fct, you can observe this as the data gets fill and you see your zig line changing potentially evenchanging trend way: up or down
There is an interesting article and some mitigation code in AB forums:
http://www.amibroker.com/members/traders/11-2003.html

For my own purpose and to ensure I do not have future leak in my code, i found this useful AB snipplet:
(not restricted to ZigZag)
http://www.amibroker.org/userkb/2007/06/01/detecting-look-ahead-by-nulling-bars/
basically, the code nullify all price data after the currently displayed range in the chart:
if your code has any future leak, you should see your Buy/Sell indicator vary as you scroll thru the chart time period..better explanation in the link....
A manual process but not that painful