'% day losers' strategy?

[DDH]

Hi everyone,

This is my second post on the forum - if my question is silly or already been asked... please be kind!!

Is there such a thing as a short-term trading strategy that involves investing in the ASX's daily biggest % losers (e.g. investing in say MCX which lost 20% today next Monday and selling when it regains value)? I know this is quite a 'lazy' trading strategy, but provided I undertake some technical analysis (I am nearly finished my M. Applied Finance degree) and so long as I stay away from the really low value stocks e.g. share prices less than $0.50, is this a viable strategy?

Would love some opinions!

Thanks everyone!

 

[CanOz]

Hi everyone,

This is my second post on the forum - if my question is silly or already been asked... please be kind!!

Is there such a thing as a short-term trading strategy that involves investing in the ASX's daily biggest % losers (e.g. investing in say MCX which lost 20% today next Monday and selling when it regains value)? I know this is quite a 'lazy' trading strategy, but provided I undertake some technical analysis (I am nearly finished my M. Applied Finance degree) and so long as I stay away from the really low value stocks e.g. share prices less than $0.50, is this a viable strategy?

Would love some opinions!

Thanks everyone!

Sounds like a mean reversion strategy. You're looking to buy oversold with a view to selling on the retracement. I reckon a guy like Sinner would have a great answer and maybe could run a backtest on something like this...I think he's a bit busy lately though.

If the general market wasn't selling off you may get a decent win rate, but I'd reckon you'd want some hint that there was some buying in the blow off low...Its like catching a falling knife really...ouch.

CanOz

 

[barney]

Hi everyone,

This is my second post on the forum - if my question is silly or already been asked... please be kind!!

Is there such a thing as a short-term trading strategy that involves investing in the ASX's daily biggest % losers (e.g. investing in say MCX which lost 20% today next Monday and selling when it regains value)? I know this is quite a 'lazy' trading strategy, but provided I undertake some technical analysis (I am nearly finished my M. Applied Finance degree) and so long as I stay away from the really low value stocks e.g. share prices less than $0.50, is this a viable strategy?

Would love some opinions!

Thanks everyone!

Howdy DDH, and welcome to ASF.

Irrespective of whether that particular strategy has merit or not (personally I wouldn't go there .... or at least have some stringent filters to weed out some of the inherant problems it might create)

ie. As CanOz has mentioned ... the "falling knife" will most likely punish you eventually, unless you have a sound strategy to offset the potential downside.

That aside, the fact that you are seeing potential in buying "value" is a good place to start ... irrespective of what you decide to trade ...

Good luck with it.

 

[PinguPingu]

You could explore the mirrored version of % day winners to buy into if you expect the strength/(or if in) up trend to continue. With that you have the risks of pullbacks instead of catching falling knives.

 

[Superboot]

Hi DDH,

I have just finished trading my first couple of months on a mean reversion system. I use the RSI indicator to determine when a share was oversold. I also use an index filter to ensure the overall market was bullish. Back testing has given positive results as did the first couple of trading months (except for QBE with its 10-15% drop in one day = falling knives).

After trading the system, I find it hard to not have a predefined exit strategy. The current system design uses the RSI to move back into overbought territory. A share could keep drifting lower and lower waiting for a bounce. How long could you handle this? In system testing I tried adding in trailing stops etc, each of which significantly hurt the returns.

I am still tinkering with this system.

Cheers

 

[saroq]

Hello DDH, I trade the US market so the scan I will give you is one I'm tinkering with on stockcharts.com

The scan is looking for high volume stocks that are oversold but starting to head out of oversold territory. I use a slow stochastic < 20 and an RSI > 30. I usually run the scan on the weekend and then add the stock charts I like to my watch list.

There's more to it than just running the scan. I'm looking for a cross in the stochastics, the stock has to have a history of bouncing out of oversold, you have to consider support/resistance levels and any news that's affecting the stock. I don't know if this will help but maybe it will.

[type = stock] and [country = us] and [daily sma(30,daily volume) > 2000000] and [country is US] and [Slow Stoch %K (14,3) < 20.0] and [Slow Stoch %D (14,3) < 20.0] and [Close > 2.00] and [Close < 99.9] and [RSI (14) > 30.0]

 

[tech/a]

A mean reversion system should be designed around stocks trending higher which have
Reversed below their mean in a short sharp time period and show reason to believe they will
Return to that mean price.

Simply looking for over sold in an indicator wont cut it.
You'll be decimated!

Get a hold of Bandy's book!!!
Search blue owl press.

 

[sinner]

Hi everyone,

This is my second post on the forum - if my question is silly or already been asked... please be kind!!

Is there such a thing as a short-term trading strategy that involves investing in the ASX's daily biggest % losers (e.g. investing in say MCX which lost 20% today next Monday and selling when it regains value)? I know this is quite a 'lazy' trading strategy, but provided I undertake some technical analysis (I am nearly finished my M. Applied Finance degree) and so long as I stay away from the really low value stocks e.g. share prices less than $0.50, is this a viable strategy?

Would love some opinions!

Thanks everyone!

It's a common hedge fund strategy, AFAIK, but for single names they generally do it on the weekly. i.e. on friday close, long the top 5% of fallers in the universe and short the top 5% of risers in the universe from last 5 days. Even monthly. For example if you're working with the ASX20, then on Friday close you'd long the stock which fell the most over the last week and short the stock which rose the most.

You can't expect high frequency (<5 days) mean reversion in single names the same way you can on indices. Some stocks and bonds _do_ have a mean reversion factor <5 days but it's not as common and can be a fleeting condition as rolling cointegration falls off.

This is a portfolio strategy. You can't run it without the size required to avoid transaction costs eating into profits and losses. Express with volatility sizing, or better, Mean Variance Portfolio for best results. Returns have a high correlation to the index.

There is an adaptation of the weekly mean reversion strategy which has better results based on using unexpected changes in earnings forecasts as a fundamental filter (e.g. long the top 5% fallers which had an analysts earning forecast change upwards within the last 3 months). Unfortunately I can't find the info on it in my notes.

 

[So_Cynical]

Is there such a thing as a short-term trading strategy that involves investing in the ASX's daily biggest % losers (e.g. investing in say MCX which lost 20% today next Monday and selling when it regains value)?

If you can pick carefully this strategy could work, you would also need a fairly open time frame...the trick would be avoiding the real dogs and avoiding buying against sentiment when a sector is turning.

Take the Mining services sector in late 2012 for example...if you jumped in to early, bought the early falls you would of had almost all your trades go against you, however that was not the case for every stock in the sector but was true for more than half of them.

Stock Selection and Timing would be key.

 

[AlterEgo]

A mean reversion system should be designed around stocks trending higher which have
Reversed below their mean in a short sharp time period and show reason to believe they will
Return to that mean price.

Simply looking for over sold in an indicator wont cut it.
You'll be decimated!

Get a hold of Bandy's book!!!
Search blue owl press.

Tech, so how do you explain the results of this mean reversion example on Bandy's site?: http://www.blueowlpress.com/WordPress/trading-systems/mean-reversion-based-on-rsi/
That system does exactly what you say "won't cut it" and "you'll be decimated". The results however, are far from being decimated!

 

[tech/a]

Tech, so how do you explain the results of this mean reversion example on Bandy's site?: http://www.blueowlpress.com/WordPress/trading-systems/mean-reversion-based-on-rsi/
That system does exactly what you say "won't cut it" and "you'll be decimated". The results however, are far from being decimated!

There are a number of reasons.

(1)

The beginning and ending dates are chosen intentionally so that the period has very little net price change.

SPY closed at 126.31 on 1/28/1999, and closed at 126.66 on 10/31/2011.
While there was little net price change over that period, there were five large moves within it.
Rise from 126 to 155 in March, 2000.
Drop from 155 to 77 in October 2002.
Rise from 77 to 156 in October 2007.
Drop from 156 to 67 in March 2009.
Rise from 67 to 126 in October 2011.

Just let me know when you have 10 yrs advanced notice that this is going to happen.

(2) There is only one chart chosen.
Try doing the same with a portfolio of randomly chosen stocks
The OP 's criteria was/is

that involves investing in the ASX's daily biggest % losers

Attempting to trade a basket of stocks with the only filtering criteria being biggest % losers and perhaps throwing in the RSI ---Will see you decimated.
MCX which was delivered up as an example I will return---as an example.

(3) His example is trading BOTH long and Short.

Going in 1 direction would be far more difficult.

But Howard does point out something which I stand corrected on.

mean reversion systems enter long positions after a period of weakness in anticipation of return to average.

They dont need to be trending--in fact as the example---which is particularly suited shows.
Non trending is preferable. After all this is the type of market the RSI excels in.

Howards timing of his work I dont think is random. (Mean Reversion systems)

I think the market in general will not see longer term trending for a long long time.
So long term trend following systems will not fair well.

I hope this answers your question and goes some way toward determining good criteria for any given method.

 

[AlterEgo]

Hi Tech,

I was not addressing the OP's strategy, I was just addressing your comment below, and pointing out that many mean reversion strategies do not do this and are still profitable.

A mean reversion system should be designed around stocks trending higher which have
Reversed below their mean in a short sharp time period and show reason to believe they will
Return to that mean price.

 

[tech/a]

Hi Tech,

I was not addressing the OP's strategy, I was just addressing your comment below, and pointing out that many mean reversion strategies do not do this and are still profitable.

You were??

Tech, so how do you explain the results of this mean reversion example on Bandy's site?: http://www.blueowlpress.com/WordPres...-based-on-rsi/
That system does exactly what you say "won't cut it" and "you'll be decimated". The results however, are far from being decimated!

Your query seemed squarely pointed at the results of Howards test------as in proving my comments totally without foundation.---

How do you explain

So I explained.

that many mean reversion strategies do not do this and are still profitable

In reply to this then.

The only other setup I know of is ranging markets/stocks-etc.
If you could list me the "many mean reversion stratagies" Ive missed
perhaps I could learn something.

 

[AlterEgo]

Tech,

Your previous statement (unless I have misunderstood you) indicates that a mean reversion system should buy in the direction of the trend, ie. find oversold security, and only enter once it starts rising back up. That's how I read ".....and show reason to believe they will return to that mean price". Is that not what you meant?

Also, you said "Simply looking for over sold in an indicator wont cut it.", however the example on Bandy's site, only does exactly that! And obviously Bandy's example does cut it, therefore completely disproving what you said.

 

[tech/a]

Tech,

Your previous statement (unless I have misunderstood you) indicates that a mean reversion system should buy in the direction of the trend, ie. find oversold security, and only enter once it starts rising back up. That's how I read ".....and show reason to believe they will return to that mean price". Is that not what you meant?

Also, you said "Simply looking for over sold in an indicator wont cut it.", however the example on Bandy's site, only does exactly that! And obviously Bandy's example does cut it, therefore completely disproving what you said.

I give up.
(1) First reply I stood corrected.
(2) First reply explains why other factors make Howards example excellent.

Simply looking for an oversold indicator WONT CUT IT.

Go ahead develop a system and try it.

You know at times this place gets so so tedious!

 

[AlterEgo]

and as to the "many mean reversion stratagies", check out any of the many systems in Larry Connors ETF trading books. All of them buy on an oversold indicator only - no other criteria used (except for above 200MA). And I'm sure there will be many more examples like this in Bandy's upcoming book too (my copy is in the mail).

 

[AlterEgo]

Simply looking for an oversold indicator WONT CUT IT.

Go ahead develop a system and try it.

You know at times this place gets so so tedious!

I have developed one. I have done extensive back and forward testing on it. And I have traded it with real money. It does work. Don't believe me, as I said above check out Larry Connors ETF trading books.

 

[sinner]

and as to the "many mean reversion stratagies", check out any of the many systems in Larry Connors ETF trading books. All of them buy on an oversold indicator only - no other criteria used (except for above 200MA). And I'm sure there will be many more examples like this in Bandy's upcoming book too (my copy is in the mail).

A mean reversion system should be designed around stocks trending higher which have
Reversed below their mean in a short sharp time period and show reason to believe they will
Return to that mean price.

lulz.

Larry Connors systems have been broken down on the internet many times, check out Bulkowskis review of the Double 7s. Check out marketscis Larry Connors reviews, etc.

http://marketsci.wordpress.com/2009...-the-geeky-of-“high-probability-etf-trading”/

Here is what Connors himself says:

Go long at the close when the index: (a) is trending up in the long-term, but (b) has shown multiple days of weakness (in other words, a pullback in an uptrend). Reverse entry rules for going short.

None of this has much to do with the OP question, AlterEgo, I am just pointing out your inconsistencies.

 

[AlterEgo]

Oh geeeeez, your definition of uptrend/downtrend really depends on what timeframe you're looking at. Yes, you'll almost always find some timeframe that anything is trending up, and another timeframe that the same thing is trending down. All these setups are buying into a falling market, ie. the day you are buying is below the previous day(s). If that isn't a downtrend, I don't know what is! Conventional wisdom is to buy on strength, mean reversion does the opposite, that's what I'm saying.

And just a few other comments:

1. Funny how Bandy rates Connors work then, if they don't work.

2. Do some backtesting yourself before automatically rejecting anything.

3. I'm not using one of Connors systems, but one I developed myself, using similar "mean reversion" principles, based on my own findings.

4. Agree, this has little to do with the OP's original question, just got sidetracked by Tech's comments.


To the OP, all I can say is get a good backtesting program and test your strategy for yourself, try different variations of it, add additional filters or parameters, etc, etc, and then you'll quickly find out for yourself if it works (or can be made to work with some tweaking) or not.

 

[tech/a]

A/E

Do you use a portfolio based mean reversion long only method with a single Over sold Indicator or something different. (What timeframe?). IE whats your universe and indicators in what timeframe.