So, in my filter, a check of the debt consolidation box would be a negative of almost 2% when compared to credit card.

...

@Fred93 - I have seen your other answers here and I know that you know the game but a 2% hit seems more substantial to me than your explanation below.

I don't see that disparity with my filters, and I don't see that disparity with all loans. Have never seen it, so I have no explanation for you. Must be something odd about your filters. Everyone's filters are different, so you are expected to see different numbers than I do. Your result does seem unusual tho.

Here are some thoughts to help you work thru whether this is a fluke or a discovery: You can filter on individual years, and learn whether this holds up year after year. You can filter on individual grades, and observe same. Etc. Be careful when you do this tho. If you filter down to a small # of loans in any category, you lose statistical meaning. I try to stick with result categories that have >1000 loans in them. You might need to widen your filters to learn from a year-by-year or grade-by-grade breakdown.

In the case of the stats you showed us, a category that has $1.6M of principal in it is only about 100 loans. I suspect that's the problem. Statistical random variation has added noise in the return numbers. Open up your filters until you get >1000 loans and I'll bet the result goes away.

It takes careful thought to make sense of back filtering. It is easy to be mislead. For example, loan volume has been exponentially growing during LC's life. That means that when you get a result that includes many years, it is HEAVILY weighted to the later years. That's ok if you understand, but otherwise you can fool yourself. NSR nicely shows you the breakdown by year, so you can observe what is going on over time if you want to. Similarly, I always exclude the most recent year, ie right now I set the final date at 09/30/2014. Gathering "statistics" on loans that haven't lived very long is meaningless. Loans have to be 5 months old before then can default at all, and similarly loans that are 6 months old have only had one "chance" to default, and so forth, so statistics for very recent loans are heavily skewed by this time shift (counting interest and defaults for very different lengths of time).