Australian (ASX) Stock Market Forum

Deviation from the regression channel - Forward Testing

Jump to new
Gringotts Bank

Gringotts Bank

Joined
30 June 2007
Posts
7,200
Reactions
1,226
This is an idea that I can't backtest, so I'll paper trade it here. I'll start with 20 trades and see how I go.

System looks back to the bar where the greatest deviation from the regression channel occurred on the previous up/down leg, and makes that the trigger for long or short.

B: KAR so long as it closes above 6.65
B: ALL 3.81
B: TTS if closes above 3.18
B: SGT 2.78

Sh: HZN .427
Sh: GPT 3.86
 
Re: 20 trades, maybe more

Gringotts Bank said:
This is an idea that I can't backtest, so I'll paper trade it here. I'll start with 20 trades and see how I go.

System looks back to the bar where the greatest deviation from the regression channel occurred on the previous up/down leg, and makes that the trigger for long or short.

B: KAR so long as it closes above 6.65
B: ALL 3.81
B: TTS if closes above 3.18
B: SGT 2.78

Sh: HZN .427
Sh: GPT 3.86
Click to expand...

HZN index shenanigans tomorrow, might want to leave that alone as you won't know flow size/direction.
 
Re: 20 trades, maybe more

skyQuake said:
HZN index shenanigans tomorrow, might want to leave that alone as you won't know flow size/direction.
Click to expand...

ok thanks for that sky. Since I probably wouldn't be watching that sort of info on a regular basis, I'd better keep it in.

An exit strategy would be useful. Maybe just a parabolic cross or 5% SL from entry price. Mainly I want to see if the entries are any good.

My idea is that volatile days are probably meaningful and their highs/lows can be used as pivot point entries , when they get swung past in the opposite direction.
 
Re: 20 trades, maybe more

Gringotts Bank said:
ok thanks for that sky. Since I probably wouldn't be watching that sort of info on a regular basis, I'd better keep it in.

An exit strategy would be useful. Maybe just a parabolic cross or 5% SL from entry price. Mainly I want to see if the entries are any good.

My idea is that volatile days are probably meaningful and their highs/lows can be used as pivot point entries , when they get swung past in the opposite direction.
Click to expand...

Interesting GB, usually the entries don't prove to be as valuable as the exits.

I'm just guessing but this is a mean reversion strategy or are the channel breaks on an uptrend a long signal?

CanOz
 
Re: 20 trades, maybe more

Gringotts Bank said:
This is an idea that I can't backtest, so I'll paper trade it here. I'll start with 20 trades and see how I go.

System looks back to the bar where the greatest deviation from the regression channel occurred on the previous up/down leg, and makes that the trigger for long or short.

B: KAR so long as it closes above 6.65
B: ALL 3.81
B: TTS if closes above 3.18
B: SGT 2.78

Sh: HZN .427
Sh: GPT 3.86
Click to expand...

Can you illustrate on a chart on of these set ups? Just so ppl know what the regression channel looks like and which bar you are referring to.

What about stop loss?
 
Re: 20 trades, maybe more

Chart upload doesn't work on my laptop, but I put it here: http://i50.tinypic.com/ae9x5g.jpg

The low of the most volatile bar on the previous up leg is the trigger for a short... and vice versa.

I guess I'd call it swing trading canoz... not sure really! It may have no advantage at all, I just liked the look of it and it sort of makes sense.

If you're thinking that the top pivot isn't known until it happens, that's correct, but the regression just gets drawn from the low pivot (which is known) upwards, and you take the most volatile bar, wherever that occurs.
 
Re: 20 trades, maybe more

There's an easier way to do this and the regression channels prob aren't necessary. Maybe finding the highest 1 day ATR since the last pivot would be easier. So I might change it,,, need to work that out.
 
GB, it is interesting that of the 4, 3 meet my current criteria to await an entry, and the 4th (TTS) was dropped off the list 26 Feb.

The triggers for mine will be......

KAR 685
ALL 378
SGT 279
PLUS positive market sentiment,


......for very different reasons to yours. I gave up on the involved technicals and use very simple market indicators in conjunction with the market sentiment for the particular stock and this has worked very well for over a decade. The only issue is the interference I get is when the bots are active it requires adjustments my algorithm.

Cheers
Country Lad
 
Country Lad said:
GB, it is interesting that of the 4, 3 meet my current criteria to await an entry, and the 4th (TTS) was dropped off the list 26 Feb.

The triggers for mine will be......

KAR 685
ALL 378
SGT 279
PLUS positive market sentiment,


......for very different reasons to yours. I gave up on the involved technicals and use very simple market indicators in conjunction with the market sentiment for the particular stock and this has worked very well for over a decade. The only issue is the interference I get is when the bots are active it requires adjustments my algorithm.

Cheers
Country Lad
Click to expand...

Hi CL, interesting, what is your sentiment indicator, if you don't mind?

CanOz
 
CanOz said:
Hi CL, interesting, what is your sentiment indicator, if you don't mind?

CanOz
Click to expand...

An algorithm taking into account buy versus sell pressures in the stack and the buy and sell pattern of the trades.

Candidates are generally found with simple charts such as (but not restricted to) Siroc and OBV diversion and P&F patterns.

Cheers
Country Lad
 
Country Lad said:
An algorithm taking into account buy versus sell pressures in the stack and the buy and sell pattern of the trades.

Candidates are generally found with simple charts such as (but not restricted to) Siroc and OBV diversion and P&F patterns.

Cheers
Country Lad
Click to expand...

So basically you structure the market with your
Siroc and OBV diversion and P&F patterns.
Click to expand...
to find levels where a turning or continuation could occur....

& then determine if you should enter based on the "stack" or depth being bullish or bearish according the algo?


Apologies to GB for a slight digression...:eek:
CanOz
 
Yep, but add fundamentals to that as well which I assumed would taken as a given - the stock has to have something going for it fundamentally and I have mentioned a few here previously. The method is too simple for most who think that a good system has to be very technical or involved.

That has worked satisfactorily when applied to reasonable stocks and not spekkies. I tend not to trade spekkies because of lifestyle dictating that I may not have internet access for days on end or may be too busy fishing, hiking, travelling. So it is mainly stocks with something behind them other than a wish and a prayer.

Better get this back to topic, sorry GB. I will look forward to any of your candidates to see if they coincide with mine.

Cheers
Country Lad
 
Country Lad said:
Yep, but add fundamentals to that as well which I assumed would taken as a given - the stock has to have something going for it fundamentally and I have mentioned a few here previously. The method is too simple for most who think that a good system has to be very technical or involved.

That has worked satisfactorily when applied to reasonable stocks and not spekkies. I tend not to trade spekkies because of lifestyle dictating that I may not have internet access for days on end or may be too busy fishing, hiking, travelling. So it is mainly stocks with something behind them other than a wish and a prayer.

Better get this back to topic, sorry GB. I will look forward to any of your candidates to see if they coincide with mine.

Cheers
Country Lad
Click to expand...

Thanks for the taking the time to explain CL, simple is brilliant, especially in this case!:xyxthumbs

GB, that's one thing that i learned while system testing too was that for all the things you look for its usually the simplest things that seem to stand the test of time. Nick pretty much hammers that home as well. Also, its not usually something you'll just find on a chart, but think beyond the chart.

Cheers,


CanOz
 
no probs, all comments welcome.

I've realized that there's no need for a drawn-out forward test. For the last few hours I have been using a bar-by-bar replay on historical data (without referencing future data) on a number of different stocks. A conclusion is impossible to draw without knowing definite stops. And there's no point creating definite stops because if I set something and it doesn't work, going back over the trades and trying to find better stops.... you see where I'm going. It's far too time consuming.

So... if anyone has the ability to code the suggestion, perhaps post it up here. The idea is pretty basic and is demonstrated in that link I gave above: http://i50.tinypic.com/ae9x5g.jpg and explained in the first few posts. As an alternative, the volatile bar could be identified as "the bar with the highest 1 day ATR of the previous swing in the opposite direction".

So I'm cutting it short, sorry.

AB programmers :1zhelp:
 
Well you could slap a tight MA on it or use Parabolic Sars and eyeball the stops...

CanOz
 
CanOz said:
Well you could slap a tight MA on it or use Parabolic Sars and eyeball the stops...

CanOz
Click to expand...

Yes I did that, but you know how it is when you get a few whipsaws in a row. Only a couple of them can make a huge difference to a system that otherwise looks good. I'm trying to code the simpler version myself in the mean time.
 
Top