Shorting spikes in equities

[CanOz]

Moderator

Thread requested by SKC to discuss shorting spikes in Equities....

Cheers,


CanOz

Well that didn't work out too well. Closed at $3.21

I always trade badly when I think I can ride this for months all the way down to $2. May be I'd learn one day.

I had a similar problem when I started swing trading ASX stocks, thinking I could trail shorts the same as longs on a nice decline.

While you may see the occasional chart which looks like this (low volatility decline) generally I've found the most reliable short returns are only 1-2 days after an up (or at least consolidation) day when looking for downtrend swings.

Basically fade pops on the first sign of intraday weakness and cover or trail very tight into long SL levels.

I was reading today a paper on short interest in (all U.S) stocks and it indicated that the consistent monthly returns to short side generally only come from small (> micro < large) stocks with large short interest ratio, and it's only the highest 1% (~50-60 out of ~5000) of short interest weighted equally which can give this kind of performance. So it seems better (in general) to look for mean reversion to provide short returns in stocks with the trend as a bias/preference mechanism.

 

[skc]

Re: BLD - Boral Limited

I had a similar problem when I started swing trading ASX stocks, thinking I could trail shorts the same as longs on a nice decline.

While you may see the occasional chart which looks like this (low volatility decline) generally I've found the most reliable short returns are only 1-2 days after an up (or at least consolidation) day when looking for downtrend swings.

Basically fade pops on the first sign of intraday weakness and cover or trail very tight into long SL levels.

I was reading today a paper on short interest in (all U.S) stocks and it indicated that the consistent monthly returns to short side generally only come from small (> micro < large) stocks with large short interest ratio, and it's only the highest 1% (~50-60 out of ~5000) of short interest weighted equally which can give this kind of performance. So it seems better (in general) to look for mean reversion to provide short returns in stocks with the trend as a bias/preference mechanism.

Agree and that is pretty much what I see in pairs trading. The short makes money after a pop usually in the first couple of days. You hope your long side also went up or at least not fall, and there's my 3% wins.

The BLD was more like a "fundamental short"... but the US housing stats are making others think the worst is over. Short is probably still the correct longer term direction...

 

[sinner]

Re: BLD - Boral Limited

Ok I couldn't help myself and had to test it. I like prorealtime for stuff like this because once you know the (easy to learn) language you can churn out tests for simple things very quickly and they handily have free ASX EOD data. I have a few 'algos' which I just plug together for tests like this.

Here's a 2 day hold of shorts under the 200 day moving average of closing prices, with the trigger to short being closes above the upper quintile of 4 day h/l range (1 year of test). 100% short, commissions $6 each way.



Eyeball of the trades shows sticking a trailing stop on profitable trades would increase the profitability significantly.

 

[CanOz]

Re: BLD - Boral Limited

Sinner, is that sample size representative? Can you test on more data?

Cheers,


CanOz

 

[sinner]

Re: BLD - Boral Limited

Sinner, is that sample size representative? Can you test on more data?

Cheers,


CanOz

It's not representative of anything more than the idea I wanted to convey to skc about short returns because he complained about something that has affected me. It's not a curve fit if that's what you mean, like I said I just plugged a couple of algos together (<200SMA, WPR4>-20, MACD<0, 2 day hold) to make this test.

I can/have test on more data, but because it's a general setup to get a representative sample it'd have to be across more than BLD, and then we are off-topic discussing short term mean reversion short returns across the bourse

 

[CanOz]

Re: BLD - Boral Limited

It's not representative of anything more than the idea I wanted to convey to skc about short returns because he complained about something that has affected me. It's not a curve fit if that's what you mean, like I said I just plugged a couple of algos together (<200SMA, WPR4>-20, MACD<0, 2 day hold) to make this test.

I can/have test on more data, but because it's a general setup to get a representative sample it'd have to be across more than BLD, and then we are off-topic discussing short term mean reversion short returns across the bourse

Yeah yeah sure, i understand. I was just curious as to how much data you prefer to use to get comfortable that an idea has merit and warrants further testing...I wasn't inferring anything, just curious as to your thoughts.

Cheers,


CanOz

 

[skc]

Re: BLD - Boral Limited

It's not representative of anything more than the idea I wanted to convey to skc about short returns because he complained about something that has affected me. It's not a curve fit if that's what you mean, like I said I just plugged a couple of algos together (<200SMA, WPR4>-20, MACD<0, 2 day hold) to make this test.

I can/have test on more data, but because it's a general setup to get a representative sample it'd have to be across more than BLD, and then we are off-topic discussing short term mean reversion short returns across the bourse

Here's a live forward test. Short DLS @ $1.16. 2c trail. Target $1.08.

Sorry BLD followers

 

[skc]

Re: BLD - Boral Limited

Actually you have to say BLD short is probably a valid trdae (again). Although I tend not to chase trades that I've just got stopped out on. Just bad psychology for me.

Short BLD $3.3. 2% trailing stop.

 

[CanOz]

Re: BLD - Boral Limited

Once of those nice little 1-2-3 corrective moves there too...

 

[sinner]

Re: BLD - Boral Limited

Yeah yeah sure, i understand. I was just curious as to how much data you prefer to use to get comfortable that an idea has merit and warrants further testing...I wasn't inferring anything, just curious as to your thoughts.

Cheers,


CanOz

Basically I try to work with very general concepts: momentum, trend following, mean reversion and look at the distribution of returns (and what affects/modifies the distribution). If you get too caught up in the timeframe or amount of data you will always find a case or two which can make the equity curve seem not worthwhile. Backtest is to make sure the concept is robust! I want to see my concept capture the return distributions I'm looking to capture, not necessarily make money just by applying the general concept. The idea is to apply 'what works' to the right instrument, and know whether or not (given overall market conditions!!!) what you're using is good at 'predicting returns'. It's hugely about volatility, because volatility is (unlike the underlying) predictably cyclical and is the lever by which the market dictates returns. Try trading an iron condor or mean reversion strategy in a low vol to high vol regime shift and watch your equity evaporate in one or two trades. Does that mean a strategy isn't robust? No, it only means it's about managing risk and adapting between strategies (which you know are robust because you tested them for robustness) depending on where vol is and where it's going. I size down or don't trade mean reversion if vol is low and looks set to explode or if the return distribution for mean reversion tells me I can't expect returns from the current market conditions. Same as trend following or momentum, I'm not looking to buy strength in chop, and if strength turns to chop then out I go. e.g. right now I think mean reversion on ASX stocks is good, much better than trend following but also running a little momentum at the moment too.

 

[skc]

Re: BLD - Boral Limited

Once of those nice little 1-2-3 corrective moves there too...

Yes. Althought the DOM is not showing any weakness yet. Still plenty of buyers lining up.

BTW CanOz you are a mod these days?! Can you move the last few posts to a new thread rather than buried in the BLD thread?

May be something like "The thrills and spills of shorting a spike"

 

[CanOz]

Re: BLD - Boral Limited

Basically I try to work with very general concepts: momentum, trend following, mean reversion and look at the distribution of returns (and what affects/modifies the distribution). If you get too caught up in the timeframe or amount of data you will always find a case or two which can make the equity curve seem not worthwhile. Backtest is to make sure the concept is robust! I want to see my concept capture the return distributions I'm looking to capture, not necessarily make money just by applying the general concept. The idea is to apply 'what works' to the right instrument, and know whether or not (given overall market conditions!!!) what you're using is good at 'predicting returns'. It's hugely about volatility, because volatility is (unlike the underlying) predictably cyclical and is the lever by which the market dictates returns. Try trading an iron condor or mean reversion strategy in a low vol to high vol regime shift and watch your equity evaporate in one or two trades. Does that mean a strategy isn't robust? No, it only means it's about managing risk and adapting between strategies (which you know are robust because you tested them for robustness) depending on where vol is and where it's going. I size down or don't trade mean reversion if vol is low and looks set to explode or if the return distribution for mean reversion tells me I can't expect returns from the current market conditions. Same as trend following or momentum, I'm not looking to buy strength in chop, and if strength turns to chop then out I go. e.g. right now I think mean reversion on ASX stocks is good, much better than trend following but also running a little momentum at the moment too.

Thanks Sinner...

I'll leave the thread to Boral now...

 

[CanOz]

Well its a new thread guys, but it ain't too pretty....Still trying to perfect that one!