Jack Lee Macdowell
Analyst
Yes, sure. It's a good question. And obviously, that is a function of several things, market conditions being one; two, the legacy portfolio naturally will run off over time. But we're certainly focused on sort of what we have in-house that we're looking at, which is kind of our model portfolio. And one of our primary goals there is to have a diversified portfolio across complementary sectors or product types that would provide stability and durability across different economic housing and interest rate environments. The legacy portfolio, primarily made up of the reperforming loans, super seasoned loans. They've got a ton of equity in them. In many cases and for many reasons, we like those assets. At the same time, they were securitized in a different period of the market. And so we're looking and they've delevered. And as they continue to delever, that has an impact on our earnings power. So going forward, I would say -- MSRs, we did our first MSR transaction. It's now less than 2% of our overall capital allocation. As we think about where we want to be from an equity duration perspective, again, depending on the types of MSRs that we're buying, are they at the money? Are they more out of the money like the trade that we did in July? I see those being anywhere from 15% to 25%, something like that, give or take. Agencies, we've mentioned, is a very important component of our overall portfolio allocation, but that's going to be a function of our liquidity needs, where we're seeing relative value, some of the opportunistic -- relative value opportunities we're seeing in other products and sectors. So it's hard to nail down like what is our specific long-term portfolio allocation. But what I can say is agencies has a permanent role there, MSRs have a permanent role there, and being opportunistic around other product sectors is something that we're always going to do just because we've got the capabilities to allocate and manage assets across the entire spectrum of residential credit products. And so we don't want to limit ourselves to any just 1 or 2. I don't know if that helps, but hopefully, it gives you some idea of what we're thinking.