Guaranteed Investment Bonds

meester

caliston wrote: »

I don't quite understand this bit. Take some example days:

Day we09/01/87(1107) close 1752
Testing y1(11/01/88) vy1=1760.2 diff 8.2
Annualised return 1.16 after 1 year

Day 12/01/87(1108) close 1755.2
Testing y1(12/01/88) vy1=1739.2 diff -16
Testing y2(11/01/89) vy2=1834.1 diff 78.9
Annualised return 1.14891253 after 2 years

Day 11/05/87(1227) close 2163.3
Testing y1(10/05/88) vy1=1792.6 diff -370.7
Testing y2(10/05/89) vy2=2117 diff -46.3
Testing y3(10/05/90) vy3=2157 diff -6.3
Testing y4(10/05/91) vy4=2524.3 diff 361
Annualised return 1.13164696 after 4 years

Day 27/04/99(5596) close 6593.6
Testing y1(26/04/00) vy1=6256.5 diff -337.1
Testing y2(26/04/01) vy2=5868.3 diff -725.3
Testing y3(26/04/02) vy3=5159 diff -1434.6
Testing y4(we25/04/03) vy4=3870.2 diff -2723.4
Testing y5(we23/04/04) vy5=4570 diff -2023.6
Testing y6(25/04/05) vy6=4864.9 diff -1728.7
After 6 years no win, dates <50%:
12/03/03 3277.5
13/03/03 3287
Annual return 0.948112237 after 6 years

I've got a table that looks like this:
Day 1107, annualised return 1.16
Day 1108, annualised return 1.14891253 ie 1.32^(1/2)
Day 1227, annualised return 1.13164696 ie 1.64^(1/4)
Day 5596, annualised return 0.948112237 ie (1+(4864.9-6593.6)/4864.9))^(1/6)

Do I just multiply up these N numbers and take the Nth root? You seemed to be suggesting weighting the longer terms in the geometric mean, but that would be just the same as not taking the yearth root initially.

(1.16 * 1.1489 * 1.1489 * 1.13 * 1.13 * 1.13 * 1.13 * 0.948 * 0.948 * 0.948 * 0.948 * 0.948 * 0.948) ^ (1/13)

Or of course you could just do (1.16 * 1.32 * 1.64 * 0.737) ^ (1/13) which is exactly the same. It's the product of the returns to the power of 1/
(Ʃyears)

caliston

OK, I do that and I get a gain of 11.4% over the full dataset

(but I have to do it by taking logs, because 1.1^8000 is a very big number. So I compute:

lnsum=LN(1.16)*1+LN(1.1489)*2+LN(1.13)*4+LN(0.948)*6
geommean=EXP(lnsum/13)

Also note that the standard deviation I quoted previously was incorrect)

I'm not making the distinction between trading days and non-trading days because it gets messy when you have to worry about whether Christmas was on a Saturday and so you gain a trading day etc etc. Overall my day count is inflated by about (7.08/5) if 8 bank holidays a year (7.08=7+(8/104)). That does mean that the number of days where there was a fall is exaggerated if it stretched over a weekend or Bank Holiday and so I overcount.

I think you might be double counting anyway - I make only 110 days in the list I posted. Excluding those that begin on weekends, that's 71. In your 5000 days (why did you pick 2003 as endpoint?) that's 1.42% ie a reduction in your quoted profit figures (which ones do you mean?) of (0.15*0.0142)=0.21%

meester

caliston wrote: »

OK, I do that and I get a gain of 11.4% over the full dataset

Sounds like a reasonable return for the level of volatility taken on.

I'm not making the distinction between trading days and non-trading days because it gets messy when you have to worry about whether Christmas was on a Saturday and so you gain a trading day etc etc. Overall my day count is inflated by about (7.08/5) if 8 bank holidays a year (7.08=7+(8/104)). That does mean that the number of days where there was a fall is exaggerated if it stretched over a weekend or Bank Holiday and so I overcount.

Not sure what you mean. In the dataset you should only have FTSE prices for trading days. At least that's the case for the data I saw.

I think you might be double counting anyway - I make only 110 days in the list I posted. Excluding those that begin on weekends, that's 71. In your 5000 days (why did you pick 2003 as endpoint?) that's 1.42% ie a reduction in your quoted profit figures (which ones do you mean?) of (0.15*0.0142)=0.21%

I quoted a profit figure earlier in this thread, based on a simple April-April calculation. I did not look at your dates, but essentially would state that it's consistent to count only trading days, so if there are 71 days in 5,000, then that is indeed 1.42%. I chose 2003 (as an end point for start dates for the bond) to exclude bonds that might not have matured yet (the maximum length is of course 6 years).

caliston

Today's closing price: 6056.5

Some useless stats: here are the results from previous years of my hypothetical bond when taken out +/-50 points from that level:

Close after 1 year: 255 days (opened in the period from 19/03/98-15/02/00 and 10/03/06-06/12/06)
Close after 2 years: 1 day (15/02/99)
Close after 3/4 years: no days
Close after 5 years: 6 days (27/04/01 to 10/05/01)
Close after 6 years with interest: 43 days (17/04/00, 10/05/00, 22/12/00 to 08/06/01)
Close after 6 year with original stake: 11 days (22/02/99, 21-28/02/00, 18/04/00, 19/05/00-23/05/00)
Close after 6 years with a loss: 0 days

Date ranges are a guide only, and not all days in a range necessarily were within +/- 50 points. I ignore any days where the result is not yet known.

Of course this is all rather sensitive to the exact nature of market movements due to the dotcom bubble so it's quite meaningless. 'Just a bit of fun'

meester

caliston wrote: »

Today's closing price: 6056.5

Ouch.

Market seems quite inflated ATM.

ukdutypaid

Yes... I was rather hoping the RBS announcement would have kept it at sub the 6000 mark....

Ah well... we're off...Let the wheel spin.... I'll see you all again in a year! lol

cepheus

Yes I was wishing for a close below 6000 so as to provide a technical barrier if the worst comes to the worst at 3000. Market seems irrational even by its normal behaviour, there is also a lot of downside if all this intervention by goverment doesn't work, could it be a case of sell in May and go away for 5-6 years? then again the UK market is undervalued if it does work, so volatility seems likely.

SpinnerB

See you all back here on the 20th April 2009

Mikeyorks

SpinnerB wrote: »

See you all back here on the 20th April 2009

The 18th April 2011 or 2012 sounds a better time for a celebratory glass?

Is anyone who dipped a toe in PLE37 ..... going for PLE38 as well? As it's a bit like backing the same horse twice, in the same race!

meester

Mikeyorks wrote: »

The 18th April 2011 or 2012 sounds a better time for a celebratory glass?

Is anyone who dipped a toe in PLE37 ..... going for PLE38 as well? As it's a bit like backing the same horse twice, in the same race!

perhaps PLE39, maybe they will up the rate by then. 14.5% doesn't look like a good deal when they've just been offering 16%.