Is BondFutureOptionPremium supposed to represent an American option (upfront premium) on bond futures ?
Am I right that opengamma is using black method in order to price that American option? shouldn’t an American option approximation or a lattice method be used? I read on one of your papers that “…with up-front payment the interest rate plays a role in the early exercise. We are back to the standard American option early optimal exercise question, a nontrivial one.”
Yes, you are right, an American option approach would be preferable. At this stage we make the approximation of the American option by the European one. With rates close to zero, the difference between the two is very small. If rates were constant at 0, there would be no difference. The difference appears when the rates a non-zero or when there is a dividend before expiry; there is no dividend on futures.
We have some American option implementation in the code, for example the BjerksundStenslandModel. But for (short terms) options on futures, we have not seen a large impact. We have not done yet the plumbing between the American option approximation and the bond futures options.
Thanks Marc for your reply. Another question is that, aren’t the quoted volatilities calculated by inverting Black-Scholes formula? in which case one should use Black76 anyway (otherwise - using some American pricer - we won’t recover market prices) ?