Market Neutral Trading

Niall

Junior member
Messages
28
Likes
1
Any other traders using market neutral for portfolio trading? I do not mean intra day pairs, but holding multiple pairs in a portfolio for days or weeks or months, and dollar neutral with sector based pairs..

If not sure what I mean there are other threads here concerning market neutral.. do a search... the general explanations of the strategy are within those threads and links to other sites also..

All those threads however tend to be quickly dominated by those who are only really interested in intra day pairs, which is not of any interest to me any more.. so wondered if there were any others like me trying this approach..

I trade on the usa markets only and previously traded for a couple of years intra day.. I am finding this method so much less stressful and more rewarding and altogether more relaxing..
 
Niall, I would have thought the timeframe for holding a market neutral pair trade was immaterial.
 
TheBramble said:
Niall, I would have thought the timeframe for holding a market neutral pair trade was immaterial.

Can't say I agree with that at all bramble... there sure is a difference between 1hr and 1 week.. in many ways..

Also the point is how do you hold say 10 pairs over a few hours? Why would you want to you might ask.. but that again is the point in a market neutral strategy of the type I am referring to.. it spreads the risk.. so all your eggs are not in one basket (so to speak).. of just one pair. Impractable to find and act upon a portfolio of stocks within an intra day time frame.. for me at least... way too frantic to even attempt such a thing..

Please note I am NOT criticising in any way people who intra day trade pairs.. I am just wanting to discuss the longer term portfolio based method.. as has often been stated on this forum and elsewhere, there is no one way to trade.. each person has to find a style that suits their own personality..
 
Trading a pair over the course of several days or weeks is something I've looked into periodically and it seems very interesting. I got to the point of manually back-testing (with all the bias that entails, sadly) a couple of interesting opportunities with the SBRY/TSCO and ANL/HSBA pairs.

I had hoped to develop a program to find highly correlated stocks where the spread was mean-reverting, perhaps over the past 90 days and then look for divergences over the weeks following and trade them. I got as far as correlations for the whole FTSE but came to the realisation that if you placed your bets based on just that, you'd be losing money very quickly indeed.

Short answer is that I think it's a great strategy and I'm looking forward to having more time to explore it. I'd be very interested to hear how you get on.
 
gbr128 said:
Trading a pair over the course of several days or weeks is something I've looked into periodically and it seems very interesting. I got to the point of manually back-testing (with all the bias that entails, sadly) a couple of interesting opportunities with the SBRY/TSCO and ANL/HSBA pairs

Short answer is that I think it's a great strategy and I'm looking forward to having more time to explore it. I'd be very interested to hear how you get on.

The way I am looking at it is really not possible to backtest.. way too complex and amorphous.. I looked into the corellation method of finding pairs.. and can see how there may be value in it.. but for now I am relying on basic ta to select stocks.. in my case pullbacks to trend support/resistance... simply aiming to find strong stocks and weak stocks within the same sector..

I think it is a potentially good strategy.. but very much relates to your expectations of trading.. the aim of market neutral portfolio based trading would be to produce a steady equity curve.. outperforming the markets on down years and underperforming them in good years... nothing sensational but steady gains.. thats one aim anyway, as practised by market neutral hedge funds..

That in itself puts it way out of the interest of most daytraders in my opinion.. when I started daytrading a couple of years ago I had the classic disaster start.. I made money easily in the first month... I really did believe.. for a brief time.. that trading was easy and I was a natural! haha.. well yes.. seems silly now to say the least... I spent the following months losing and trying to figure out why the damn level 2 was not working for me anymore.

After a while I settled down and spent time learning and watching and as I have a reasonable sense of discipline stabilized my trading but never quite found a style that worked for me for any extended period. I came to realise that aiming for the gains I had believed were possible in trading was not likely to happen for me and gradually adjusted my expectations to the point that a nice steady gain would do nicely for me.. forget the dreams of wealth.. hence my shift from intraday towards swing trading last year..

I spent a few months trying to find a method I liked and found myself naturally inclining towards a market neutral style before I had even heard of it as such.. and on researching discovered the strategy as described by others.. as with any method I would say there are a zillion ways to operate the same basic concept, but so far this year for me it has worked better than my intra day trading did.. as I said in my first post I find it more relaxing even tho I have now maintained a portfolio of between 5 and 7 pairs constantly since jan 2nd.. remarkable lack of tension when the market is doing its usual mad demonic dance... still trying to understand the dynamics of the the porfolio action on different days.. seems very complex but stable.. depending of course on my stock picks...

This is getting too long.. shall stop now... may write more if not boring people too much
 
Niall, my point about timeframe being immaterial to market neutral pairs relates to my understanding and use of the system.

I presume you look for specific 'general market' conditions to exist before considering a trade - such as an oscillating market (rather than trending). These general market conditions can be considered to exist looking at a monthly, weekly, daily or 5min chart.

For instance, the DJ may look like it's in a trading range on a monthly chart, but trending like crazy on a 5min. Hence, the decision on which timeframe to use determines what you see - hence, immaterial to the overall process of market neutral pair-trading.

In your initial post you mentioned sector pairs, this immediately struck a chord as this has been mentioned elsewhere as being the most advantageous long-term approach to pair trading - using a pair from the same sector, which is by no means initially such an obvious choice.

As for managing more than a couple of pairs at any one time, regardless of timeframe (and ignoring the increasing information load as you shorten the timeframe) there's also an issue of just how market neutral can you be with more than one pair? You'd need also to be sector-neutral which complicates things just a little too much for my tastes.

As other posters have said, the selection of potential pairs is difficult enough to achieve for even one pair. If you narrow that down to a pair from the same sector, even more difficult.

Managing half a dozen is not that likely to be a problem too often.
 
I agree with Bramble. A market is a market is a market, sure there are a few differences (such as the opening 15 mins of volatility on an intraday chart) but generally the statistical correlations of one bars character against the next should be the same across all time frames. Human fear and greed (which iswhat the chart shows) is the same on a 1 min chart as it is on a 1 day chart.

As for longer term pairs though, have you thought of sector rotation? eg 5 financials vs 5 utility pairs? Stay away from Tech sectors though as there is too much 'speculative money' in these for sector rotation to be significant.
 
Bramble.. you make some perfectly valid points about time frame, but for me its my 'perception' that matters.. it may well be all of a muchness in reality.. but my perception of the shorter time frames is that the noise level is higher.. such as, it may well be perfectly fine to look at ma's on a 1min chart or a 5 min chart.. as I have done many times.. but to look at the ma on a daily chart seems to be something taken notice of by far more traders.. relative to how many people care if qcom has just broken its 50ma on a 2 min chart for example... so many time frames but the most watched I would believe is the daily.. and as people move stocks that seems more powerful to me.. but again that is my opinion and my perception.. you can have an opposite one and thats fine.. thats trading..

As it happens I do not look at the overall market for general conditions before placing a trade.. the point in this method for me is to remain fully committed at all times, regardless of market conditions.. thats because the stock pairs and the overall effect of the portfolio is market neutral.. it goes its own way whatever the market does.. some days the market moves hard and fast.. and my port barely shifts.. pairs within the port can move all over the place but the combined moves are balanced out... other days the market is flat.. and I am rapidly moving one way or another.. still trying to understand this, but the market mood def does not seem to matter to me..

Also I think it is perfectly possible to be market neutral however many pairs one has.. my definition of market neutral is one in which you have an equal value of shares long as short.. being sector neutral makes it tougher I agree.. and I have to say hard work too.. but with decent screening software and gradually evolving ideas, this is becoming slowly easier for me.. as I said above I am running 7 pairs now and for some time and will soon be sitting with 10.. not easy but doable

BBB.. thanks for that idea.. sector rotation is not a bad concept at all.. I do tend to drill down as deep as possible in my screening within industry groups within sectors to get as close as I can.. but sometimes do end up backing away to the wider sector definitions.. the trouble is that once a pair is matched and trade opened.. anything can happen.. some have closed within 24hrs.. some have taken 2 months.. so rotating is a bit tricky like that and also implies that I might actually understand what is going on out there in the market! I mostly regard the market as slightly mad and impossible to predict or follow..

All in all I am finding this very interesting and as what I am doing is working thus far am mostly concerned with polishing and improving my methods.. the management of an active portfolio is fairly complex I think.. at least setting up my own combination of software and control systems.. and also of course, its all for nothing without good stock selection.. as with any trading..
 
Maybe you should think of futures spreads. The relationships and rational are more obvious. Far more leverage while keeping risk under control to boot.

Read some Howard Abell
 
This is a really interesting topic - thanks for sharing your thoughts with us Niall.

How do you go about creating your pairs? Do you have your own software or do you use a proprietary system. I believe that AIQ Matchmaker can be useful in this context.

http://www.aiqsystems.com/docs/mm_ovrvw.pdf

Do you just trade the US where it is easy to go short? What about the UK using CFD's?
 
Thanks BBB, I shall certainy keep that in mind... while I am pretty happy with progress thus far and familiar with dealing in us stocks I am always interested in other approaches which may work better or could be switched to at a later date if this stall at some point.. so grateful for the input..

Thanks also Roger, I appreciate that.. and shall have a look at that link.. My method of selecting pairs is fairly basic at the moment as broadly described above, I have looked into the more complex methods such as correlations but avoiding that for now.. keeping it simple for now.. software is inexpensive and I will describe it later for anyone interested..
 
Balancing your portfolio with pairs requires your choice of stock-pair to have one above its 'value' (whatever you are using to decide on 'value') and one below its 'value.

The rationale being that as the market moves in whatever direction it moves - (you neither know nor care which direction) - one of your pair will react more deliberately to that momentum, the other less, but one will be losing - one will be profitable.

For instance. Stock ABC is well above its 'value' and 'XYZ' is well below its 'value'.

You short ABC and to an equal value of stock you go long XYZ.

The market moves up (for argument's sake).

ABC (which you are short) is likely to move up only a little (as it's already over-priced according to your value calculation) and will incur a small loss, while XYZ (which you are long) should move relatively more strongly - bringing in a greater profit than your loss on ABC.

You close the trade at a pre-determined target level for XYZ.

Now, I may be trading pairs differently or 'incorrectly', but the above system I imagine will work as well in any timeframe.

I'm don't have an opposite opinion as to timeframe Niall. I don't care which chart people use to make their trading decisions. If Daily is the most used - that's fine by me. But all those who buy/sell their positions based on a Daily chart - do so the next day - at all different times of the day.

Just like a Mandelbrot set, you get the same patterns in all timeframes.
 
You might also want to look at the betas of the stocks in the pair to see how they move with the market. Once I have my data sorted out I am interested at looking at the historical chart of various pairs.

Stew
 
Niall

I have had a good look at this approach while overseas for an extended period a couple of years ago. I traded a few pairs over 2-20 days based on a daily/weekly total of best bid/best ask sizes for the top 40 Nasdaq 100 shares. It worked out quite well, and as you say, it was a lovely relaxed way of trading.

However, one definite conclusion that I arrived at months later when I went over the results was that my dollar neutral approach wasn't as effective as using volatility neutral position sizing (for my method at least). I may have been dollar neutral at the outset, but sometimes with 1 volatile and 1 docile stock. Then the whole result hinged on the volatile stock.

If I had based the sizing on each side on the 10 day ATR of each stock, then results were (theoretically) improved dramatically. I still find the method very interesting though, although I havent played with it since. You've encouraged me to take another look !

rog1111

Niall said:
Any other traders using market neutral for portfolio trading? I do not mean intra day pairs, but holding multiple pairs in a portfolio for days or weeks or months, and dollar neutral with sector based pairs.......

.......I trade on the usa markets only and previously traded for a couple of years intra day.. I am finding this method so much less stressful and more rewarding and altogether more relaxing..
 
TheBramble,

Another way of looking at pairs is to go long the stock which is strong relative to the market and short the weak one.

The idea is that if the market moves up the strong stock will continue to outperform while the weak one will remain so and underperform.

LII
 
Re: Re: Market Neutral Trading

Rog, couple of interesting points I'd appreciate your comments on.

[...]I traded a few pairs over 2-20 days based on a daily/weekly total of best bid/best ask sizes for the top 40 Nasdaq 100 shares.


Can you explain exactly how that worked? I understand the top 40 (by market value I assume), but I don't understand what you mean by "daily/weekly total of best bid/ask". How did you pick pairs based on that data and was it daily or weekly?

If I had based the sizing on each side on the 10 day ATR of each stock, then results were (theoretically) improved dramatically.

Position size in pairs aims to have the same value on each side (e.g. 100 X $50 stock and 75 X $67). How would you have applied the 10 ATR to this standard' pair sizing algorithm?
 
LevII said:
TheBramble,

Another way of looking at pairs is to go long the stock which is strong relative to the market and short the weak one.

The idea is that if the market moves up the strong stock will continue to outperform while the weak one will remain so and underperform.

LII

That's an interesting point too LevII.

It would be an interesting exercise to do a correlation of the two approaches.

My own gut feeling is that there's less 'to go for' in the same direction if there's already been a sustained trend in that direction. Could be wrong though. (Frequently am!).
 
Re: Re: Re: Market Neutral Trading

The Bramble

To clarify, I used 2 approaches. Daily totals were used as a basis for the shorter term positions (2 - 5 days) and weekly totals were used independently for longer trades holding for up to 20 days.

The totals themselves were collected via an Excel VBA routine written specifically for the purpose, which would run all day during market hours. At the top of the bid and ask the prices/sizes would constantly change. For all Nasdaq 40 stocks (in order of Index weighting), the sizes were summed continuously on both sides, with a couple of conditions, eventually arriving at a ratio of all bids:all asks. From memory I can tell you that the ratios usually varied from approx 0.5 to 2 - the higher the ratio the higher the relative (temporary) strength of the stock. So the 2 extreme stocks out of the 40 were chosen as candidates for short/long. I could have done this for all 100 stocks, but PC resources were under pressure with all the live Excel links.

I realise that equal dollar amounts on each side are commonly used. However, the 10 day ATR approach could be used as follows. First calculate the 10 day ATR of both stocks, as a % of their last closing price. Suppose the Long stock was SBUX with an 10dATR of 2%, and the Short stock was MSFT with a 10dATR of 1%. Regardless of the $ price of each, I would want to be holding twice as many dollars worth of MSFT in this case, in order to attempt to balance out my risk eg Short 400 MSFT@25 = $10,000, Long 200 SBUX @25 = $5000

Suppose then on the next day, both stocks went up by a price movement equiv. to their 10dATRs, the position would have been :
MSFT -1% x 10,000 = -$100
SBUX +2% x 5,000 = +100
total = $0

If I had gone $ neutral with $7500 on each side, and the same pricre movement occurred, we now have :
MSFT - 1% x 7,500 = -$75
SBUX +2 % x $7,500 = +$150
total = $75

OK the latter model gives a better PL in this case, but the point is that I am trying to balance things more effectively and reduce the effect of the more volatile stock on the overall position. If the above moves had been the other way round then the second model would have been more negative than the first. If you just go $ neutral without considering price volatility (ATR), then one stock may be way more volatile, and will dominate the whole position almost as though you were just trading the one side.

I hope this makes sense and that I have got my figures correct - it's getting late & my brain is tired !

rog1111

TheBramble said:
Rog, couple of interesting points I'd appreciate your comments on.

Position size in pairs aims to have the same value on each side (e.g. 100 X $50 stock and 75 X $67). How would you have applied the 10 ATR to this standard' pair sizing algorithm?
 
Top