Regression charts?

SanMiguel

Experienced member
Messages
1,136
Likes
25
Does anyone know how to get up regression charts other than by using a Bloomberg terminal?
Would I have to somehow export chart data to Excel?
 
Yes, unless whatever you're using to chart stuff has regression functionality in it.
 
Yes, unless whatever you're using to chart stuff has regression functionality in it.

Know any software that allows you to export the data?
Let's say for example, I wanted to plot BLand against Land securities. I can do this on a chart but it doesn't give me the price of both shares rebased to 1.
 
Do you mean regression lines on charts along with upper and lower standard deviation bands based on confidence levels? If so, then eSignal has this functionality.
 
Do you mean regression lines on charts along with upper and lower standard deviation bands based on confidence levels? If so, then eSignal has this functionality.

No, I mean a bit like a ratio chart.
If something were correlated then the ratio would be up and down between 1. I guess a ratio chart is easier than regression.
With regression, you rebase the charts to 1 somehow and plot their prices.
 
hopefully i understand what your trying to do...whenever i have to do something like this i don't rebase the data...all you do is just plot the percentage returns into a scatter graph and plot a trendline with the trendline showing you how far y would move for a 1% in move x (0.8% in this case). i've attached an example, is that what your looking for?

btw if you wanted to do a pair trade type situation you could just divide one price into another and then you have a ratio like any currency pair and i believe you take the STDEV of the 50day correl to trade off or something similiar.
 

Attachments

  • bland.xlsx
    35.1 KB · Views: 210
Just had a look at the spreadsheet. How do you know what's what or rather, where we are now, without a time axis?
 
errr...i don't get it...the only way time is relevant is the period that your looking at...so what the spreadsheet says is that between 29Mar10 and 25Mar11 for every +1% daily move in BLND (x-axis) over the period, LAND (y-axis) returned +0.8% (or so is my understanding, no1s taught me the maths behind it, i only know it through having to do R-squared stuff for my degree).
 
hopefully i understand what your trying to do...whenever i have to do something like this i don't rebase the data...all you do is just plot the percentage returns into a scatter graph and plot a trendline with the trendline showing you how far y would move for a 1% in move x (0.8% in this case). i've attached an example, is that what your looking for?

btw if you wanted to do a pair trade type situation you could just divide one price into another and then you have a ratio like any currency pair and i believe you take the STDEV of the 50day correl to trade off or something similiar.

That's pretty much what I'm on about - thanks.
What are your last 2 columns? Edit: oh I see ratio of today's price to yesterday and then shouldn't I do column D / column E?
Let's say for example, you bought 1000 shares in BLAND and wanted to sell Land securities, I was under the impression, you would sell the same monetary amount or should you follow the ratio guides? With pairs trades, you're always looking for the ratio to return to 1:1.
 
Last edited:
the last two columns are the daily returns...so you were saying something about rebasing but if you think about it looking at the daily percentage return is the same thing as rebasing(i suppose)...either way, the scatter graph is looking at the relationship between the returns of BLAND/LAND and the daily return is the thing you want to compare.

so if you look at the r-squared it is 0.8 (or whatever) which means that x explains 80% of the moves in y (r-squared isn't the same as correlation although someone else should explain the difference tbh as my grasp is weak at best). i can't remember exactly how you do the ratios, i've never done pair trading seriously, there is an ig index pdf somewhere that explains everything in detail though. (i would hazard a guess that you just do the same monetary amount, as the volatility is controlled by looking at daily percentage returns, although im probablly wrong).
 
Thanks.
On an aside, how do market neutral strategies manage risk?
I guess the very fact that some hedge funds run long short portfolios means that even in a market crash, the portfolio probably survives fairly well? Volatility must be risky for leveraged accounts.
 
There's many approaches... Most of the time, it involves multi-factor models and portfolios that are constructed to be as factor-neutral as possible. Sort of a next generation refinement of CAPM beta.
 
Sorry I meant more how the hell you can be long/short and factor neutral and get a return. Sometimes I end up posting what I think out loud... before I've actually thought about it lol.
 
Sorry I meant more how the hell you can be long/short and factor neutral and get a return. Sometimes I end up posting what I think out loud... before I've actually thought about it lol.

You're playing a spread really, I think he meant "market neutral as possible".
It's market neutral in the sense that if all things remain the same with company earnings, the stock market remains fairly stable then you should see the spread return to its normal amount. However, let's say the market crashed 10% in one day - yes things should return to normal as they have done before and by being market neutral you should be + an - in the pair trade still but the volatility could wipe out accounts that are leveraged too much.
 
Sorry I meant more how the hell you can be long/short and factor neutral and get a return. Sometimes I end up posting what I think out loud... before I've actually thought about it lol.
Well, this is relative value trading, innit? These strategies are, essentially, a form of liquidity provision and your return is probably going to be similar to that of a mkt-maker.

And yes, volatility kills, but death by volatility is everywhere and always a function of size, more than anything else.
 
I was thinking more along the lines of net income streams/total return as I (probably incorrectly)assumed they would be holding.
 
And yes, volatility kills, but death by volatility is everywhere and always a function of size, more than anything else.

Presumably it would be a bad idea to have 100% of an account in pairs trades exposed to equities? Having said that, the market neutral factor should help balance the volatility?
 
Top