SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
bigslick
Posts: 101
Joined: October 4th, 2005, 2:01 pm

model for yield curve

February 28th, 2006, 9:09 pm

rcohen,the concept is great... the problem now is that economists are wrong more often then they are right (heh). However, there could be massive benefits in correctly pricing curves once the data is released. it just gets real complicated when you start factoring the free flow of capital between international markets (as Sanfran pointed out... different markets have different drivers and sometimes even conflicting). i'll be looking for more of your work on this in the future. good luck.
 
User avatar
cosmologist
Posts: 640
Joined: January 24th, 2005, 8:08 am

model for yield curve

March 1st, 2006, 10:28 am

ruben wroteYou’ve got a problem with this?! I challenge you, or any other economist for that matter, where ever you think you sit on the intellectual ladder, to come up with an original, publishable idea in physics – and I don’t mean “econo-physics”!my commentHahhahahahh. Ruben, let me be a little more frank. Don't you think at this stage of advancement 99% of us cann't go back to physics. I think, baringly few applied fields and string stuffs,most of the pure branches are almost , i said almost, done with. anyways,that was a clever challenge. The odds are mightily loaded in your favor. I have not gone through your paper. hence didn't write a word about it. Shall go through on the week-end and see how much i can digest.Well, I personally have been benefitted from your work,published in Wilmott. Good job.cheers
 
User avatar
SanFranCA2002
Posts: 649
Joined: October 3rd, 2002, 5:05 pm

model for yield curve

March 1st, 2006, 3:35 pm

ruben wroteYou’ve got a problem with this?! I challenge you, or any other economist for that matter, where ever you think you sit on the intellectual ladder, to come up with an original, publishable idea in physics – and I don’t mean “econo-physics”!Sorry. I was out yesterday (and will be out the rest of the week also) so I missed the whining. An unfortunate byproduct of a physics education is that they teach the students extreme arrogance. I've seen it over and over and over. So if you know physics, hey theres math in economics so all you have to do is read a few papers and believe you are cutting edge and away you go. I have a relative who builds bridges, I would never begin to tell him how to do that, but because he reads the business section of the paper he regularly tells me all about the economy. And he get immediately upset if I mention another way to look at the data because he is an expert at building bridges.
 
User avatar
rcohen
Topic Author
Posts: 252
Joined: November 15th, 2001, 12:06 pm

model for yield curve

March 1st, 2006, 6:54 pm

QuoteOriginally posted by: bigslickthe concept is great... the problem now is that economists are wrong more often then they are right (heh). I know, I know..., but what can you do. Every major bank seems to have an economist or two, whose projections go into the bank's models to produce this, that and the other. Wrong or right (wrong, as it always turns out to be), these forecasts in most cases follow some consensus, plus or minus a microscopic amount. I guess if you come up with a model these days, it has to be such that it uses these forecasts or expectations as an input. The yield curve model I've proposed does just that. Whether it's good or garbage, only time can tell.
 
User avatar
rcohen
Topic Author
Posts: 252
Joined: November 15th, 2001, 12:06 pm

model for yield curve

March 1st, 2006, 7:02 pm

QuoteOriginally posted by: cosmologistDon't you think at this stage of advancement 99% of us cann't go back to physics. I think, baringly few applied fields and string stuffs,most of the pure branches are almost , i said almost, done with. anyways,that was a clever challenge. The odds are mightily loaded in your favor. I have not gone through your paper. hence didn't write a word about it. Shall go through on the week-end and see how much i can digest.Well, I personally have been benefitted from your work,published in Wilmott. Good job.cheersCouldn't agree more with you. I'd leave this stuff to physicists and tend to my own business.Looking forward to seeing your comments.
 
User avatar
saimqn
Posts: 122
Joined: December 6th, 2005, 6:37 am

model for yield curve

March 1st, 2006, 9:20 pm

My comments:1. Empirical- You take just a few arbitrary data points in time and show high R^2 values but this goodness of fit may be due to other reasons. Some more serious empirical evidence would take into account the entire time series of the curve.- The way the model is presented is for zero-coupon yields but you fit it to data from par or near-par coupon yields from the Fed site- I would stress-test your findings by looking at a longer time series and comparing the fit from your model to that of, say, a two-factor Vasicek or CIR. If you do better consistently, then you may be onto something. 2. Theoretical- I am not at all clear why higher or lower expected returns on assets (the market) would correlate with booms (upward sloping curves) or recessions (downward sloping curves). It is intuitive in a way but must be justified empirically or theoretically.- The model does not satisfy no-arbitrage, which a good model should. I recommend you read the work of Monica Piazzesi, she has written extensively on non-arbitrage term structure models.- Nelson and Siegel have something similar, yes. Their yields load on the factors with constant coefficients on time and 0 intercept. Something like [1 - exp(-kT)]/kT, where k is a constant and T is time to maturity. This would hold only in a Vasicek-type model with constant market price of risk. Not only constant but with specific values But the data seems to reject this constant market price of risk (the expectations hypothesis), pointing to the fact that the MPR is a function of the factors that drive the short rate.
 
User avatar
mensa0
Posts: 191
Joined: January 20th, 2004, 8:56 am

model for yield curve

March 2nd, 2006, 5:38 am

I haven't read the paper, but I've downloaded it and I will.On the physicist/economist issue, in my opinion, all the math skills in the world will not necessarily lead one to a better understanding of the markets.Consider the risk/expected return relationship. Many investors expect a HIGHER return on low risk (low volatility) stocks. They think they'll earn a better return on say, WalMart, Walgreens, or P&G, etc., than on any small, volatile, risky stock, especially in the long run. So, for them, their expected return is inversely related to risk. I see this all the time in the buying habits of many equity investors. Are they just concerned with half of the return distribution? Now, where's the financial model where risk and expected return are inversely related? A model of a market (or a yield curve for that matter) must account for behavior which is probably outside most finance/economic theory, at least in the traditional approaches.Mike
 
User avatar
rcohen
Topic Author
Posts: 252
Joined: November 15th, 2001, 12:06 pm

model for yield curve

March 4th, 2006, 7:28 pm

Thanks, saimqn.Very useful feedback, indeed. As discussed in the PM, I will incorporate most of them (especiallly those that are within my time constraints) as I modify the paper.
Last edited by rcohen on March 4th, 2006, 11:00 pm, edited 1 time in total.
 
User avatar
farmer
Posts: 13477
Joined: December 16th, 2002, 7:09 am

model for yield curve

March 6th, 2006, 12:22 pm

I'm having a moment of doubt as to whether the yield curve calls for a model. Models are useful when you have lots of people dealing with local versions of the same phenomenon over and over. The higher frequency and more standardized and numerous something is, the more it calls for automation. But there is only one US yield curve. And the factors influence it at such a slow pace, you don't lose much by assigning an actual person to spend his whole day following it. And what the benefit of making a model which he can transmit to other people is, and teach to the masses, is not clear to me.So much of science is great because it equips people to deal with situations, with the benefit of the experience, and the trial and error of others. But suppose you play the saxophone. Is the best song you can compose, a song which every music student at every university can be taught to play? It seems like you might throw away an opportunity, when your composition must take the form of a training manual. If you limit yourself to inventing things in a teachable form, you limit your invention to things that can be taught.Maybe there should be a longer time between invention, and reduction to a reproducible system. In other words, coming up with the idea, and then being able to codify the idea, might be years apart. An understanding of the yield curve could precede an ability to put it into math by years. Many people are at the tops of their professions for years, and still wouldn't be able to write a book telling how to do what they do. Getting it in teachable form is less important than getting it right, unless creating it in teachable form is the best way to work on it.
Last edited by farmer on March 5th, 2006, 11:00 pm, edited 1 time in total.
 
User avatar
farmer
Posts: 13477
Joined: December 16th, 2002, 7:09 am

model for yield curve

March 6th, 2006, 1:07 pm

Here's an example, quoting from your paper:Quoteit is accepted that two independent schools of thought lie behind what shapes the curve. These are the (a) market segmentation theory and (b) expectation theory (Fabozzi, 1996 & 1999). Whereas in the former the shape of the yield curve is determined by the supply and demand for securities within each maturity sector, assuming that neither investors nor borrowers are willing to shift from one maturity sector to another, in the latter, broadly speaking, it is presumed that the underlying forward rates signal the market’s expectations of future actual rates.A model might take circumstances as an input, and turn them into an output assuming participants behave in a certain way. More useful it would seem to me, would be actually taking a poll, or otherwise gathering information on why people are choosing to postpone borrowing at every unique moment in history.A model might assume political or economic conditions translate into a certain supply of US government debt at various maturities. But as to whether the US treasury actually decides to lock in certain maturities, might vary from year to year for the "same" set of conditions. So then you're "tuning" your model, when you could simply be figuring out whether Uncle Sam is going to be selling or not, or something.What shapes the yield curve is not static, but is discoverable for each subsequent moment.
Last edited by farmer on March 5th, 2006, 11:00 pm, edited 1 time in total.
 
User avatar
rcohen
Topic Author
Posts: 252
Joined: November 15th, 2001, 12:06 pm

model for yield curve

March 7th, 2006, 6:59 pm

QuoteOriginally posted by: farmerAnd what the benefit of making a model which he can transmit to other people is, and teach to the masses, is not clear to me. Check this out:http://www.wilmott.com/messageview.cfm? ... SGDBTABLE=
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On