if you want to calibrate the mean reversion parameter as well, then you have got two ways out here:
1) Continue with CalibrationType “Bootstrap” and add another calibration swaption; it has to reuse the expiry of one of the existing calibration swaptions and have a different term. Like this:
expiries: 1Y, 2Y, 3Y, 5Y, 7Y, 10Y
terms: 5Y, 5Y, 5Y, 5Y, 7Y, 10Y
expiries: 1Y, 2Y, 3Y, 5Y, 7Y, 10Y, 5Y
terms: 5Y, 5Y, 5Y, 5Y, 7Y, 10Y, 10Y
With CalibrationType “Bootstrap” the LGM builder will overwrite the time grid for the volatility parameter with the distinct swaption expiries. So the second choice there does not increase the number of volatility parameters, but adds the extra calibration instrument you need.
2) Switch to CalibrationType “BestFit”
In that case the time grid for the volatility parameter is not overwritten, but the LGM builder uses whats given in the TimeGrid tag.
If that lists less grid points than there are calibration swaptions, then you are fine and BestFit will do a global optimisation.
I hope that helps, and I am curious how you get on with these options. Technically this should work (I just double checked), and that’s what I have seen users do.
An alternative (and my preference) would be not to calibrate the mean reversion parameter to European Swaptions, but to keep it fixed here. I’d rather try to find a reversion level that is consistent with market Bermudan Swaption prices if you have visibility of those (set reversion speed manually, calibrate to your co-terminal European Swaptions, price associated Bermudan Swaption on that calibrated model, compare with market prices if available, change reversion level and try again etc.). We do not have such an optimiser in ORE (yet).