Hi,I try to build an inflation curve in order to estimate the inflation forward, but I don't know how to start ? Should I use the inflation bond to stip an inflation curve ?Could I get some reference on this subject ?Regards,

Yes... using inflation bond to strip an inflation curve is a good way to go. You can find a very brief explanation on "Inflation Indexed Securities - Second Edition" by Mark Deacon, Andrew Derry and Dariush Mirfendereski at pages 272/274. Same as with yield curves, you need to make assumptions about the shape of the inflation curve. In the book they suggest to interpolate on the year-on-year inflation rate... Unfortunately the book is more about bonds than derivatives, so it could be a bit expensive if you only need those three pages... but at least it's a place to start...gc P.S. The book by Anderson, Breedon, Deacon, Derry and Murphy called "Estimating and interpreting the yield curve" also gives some ideas on how to build inflation curves (the approach there is to use McCulloch). Unfortunately the book is out of print, and I had to go to the library of a local university (I went to the LSE in London) to find it.

Things to bear in mind(1) correct modelling of seasonality(2) correct interpolation between only a handful of bonds(3) some assumption of term structure of bond-swap spread

You could use the zero inflation swap curve. Is probably easier tha using bonds.

point (1) still stands

If using an inflation bond to strip the inflation curve does it matter if using a bond or an index as Barclays Inflation-Linked Bond Index ?

Barclays index measures a basket of bonds, totally irrelevant in curve building. Two simply ways to build an breakeven curve:(1) Using Govie Linkers: using all govie linkers available to generate as many par yield points as possible. Interpolate them with care. Bootstrap to get zero curve. Overlay seasonality model. This gives a real yield curve in sovereign credit space. Add credit spread to convert to real swap space. Breakeven (or inflation) can be obtained using the norminal swap curve and fisher equation.(2) Using inflation swaps: using all tradable tenors. Interpolate them with care. Bootstrap to get zero breakeven curve. Overlay seasonality model.

Would you happen to know a good way to estimate hedge ratios (betas) between linkers and nominal bonds? Is there any evidence of betas being dependent on rate level? Anything fancy beside rolling linear regression that you have heard of? Thanks for sharing!

Three ways in practice(1) use whole sample period mean beta(2) rolling beta with period depending on your investment horizon(3) model the beta vs. norminal rate dynamics (beta tends to be higher in low rate environment)