Home migration is used to tackle the home assignment problem in home-based software distributed shared memory systems. We propose an adaptive home migration protocol to optimize single-writer pattern. It is unique in that it uses a per-object threshold that is continuously adjusted to facilitate home migration decision. This adaptive threshold is monotone decreasing against the likelihood that a particular object presents lasting single-writer pattern. And it is tuned according to the feedback of previous home migration decisions at runtime. We implement this adaptive home migration protocol in a distributed Java Virtual Machine that supports transparent truly-parallel execution of multi-threaded Java applications on clusters, where a global object space virtualizes a single object heap for distributed Java threads. Both the analysis and experiments show that our new home migration protocol demonstrates both sensitivity to the lasting single-writer pattern and robusticity to the transient single-writer pattern. In the later case, home migration is prohibited to reduce the heavy-weighted home redirection overhead.