- 变量筛选-我们在prosper和lending club的案例分析中列举了不同的变量，有些是直接相关的（例如调查），另一些是不明显而需要计算的（例如信贷的稳健性和偿付能力）。
- 变量筛选剔除了大量那些不满足投资者需求的贷款。例如，如果一个投资者分析借款人的历史借款信息，它可能表明, 以“婚礼”为筹款目的的借款人目的不符合利率、违约率、或者组合要求的收益率。在这个例子中，投资者将排除所有目的为“婚礼”贷款,而不管这笔贷款的其他属性。
There are several ways to build a high performing portfolio of Online Lending loans. We have written blog posts on some of the factors that are involved in making a portfolio successful. They include:
- Proper diversification – Selecting the correct investment amount per loan in order to invest in a large enough number of loans to mitigate risk of loss. The more loans that are in a portfolio, the lower the variance in default rate, which makes the likelihood of a negative return less likely as well.
- Variable Filters – we’ve done many posts on different variables on both Lending Club and Prosper. Some are straightforward (e.g. inquiries), others are less obvious and necessitate some calculation (e.g. credit robustness or ability to pay).
The 3 most common investment strategies for Online Lending are index-style, variable filters, and statistical models. Each of these strategies results in a loan either passing or failing to meet a specified set of investment criteria.
- Index-Style investing results in all loans passing. This strategy tends to be popular among investors with large amounts of capital, or those who do not have the time or ability to maintain a detailed credit strategy.
- Filters remove swaths of loans that do not meet an investor’s requirements. For example, if an investor analyzes historical loans by loan purpose, it may show that loans to borrowers stating “wedding” as the purpose do not meet the interest rate, default rate or return requirements for the portfolio. In this case, the investor would exclude all loans with “wedding” as the loan purpose, regardless of the other attributes of the loan.
- A statistical model scores or ranks each loan individually, then the investor creates a strategy to determine the levels of these scores/rankings that are acceptable.
Each investment selection strategy has merits and drawbacks. Index style investing results in the largest volume and is easiest to implement. However, it is completely reliant on the origination platform’s underwriting capabilities. Filters are easy to develop and implement, but they can cut out a lot of “good” loans, especially if a filter contains a large number of variables. Statistical models are the most precise way to invest, but implementation is difficult without a programmatic connection to the originator. Models also require a knowledge of statistics and need to be maintained with periodic testing and validation.
Regardless of the selection strategy, the investor must decide which specific loans to invest in at any given time. The investor may choose to invest in every loan that qualifies under the selected strategy, regardless of the distribution of interest rates. If there are more loans available than the investor has cash to invest, then the loans would be randomly selected for investment. The other option for the investor is to prioritize certain loans over others for investment. Investing in every qualifying loan is the easiest option, as the investor simply sets the strategy and then invests in everything or randomly selects if more loans are available than cash. The issue with this option is that the investor will be relying heavily on the borrower mix of the originators. The prioritization option enables the investor to more closely manage portfolio returns.
To illustrate this, assume an investor uses the passive-style, index approach. If the investor simply invested in every Prosper fractional loan that became available, the resulting portfolio (numbers below are from 2012, adjusted slightly to today’s interest rate environment) would have the following returns and distribution by credit grade
This would result in an overall return, weighted by the distribution of credit grades, of 6.4%.
If the investor were to adjust the portfolio based on the credit ratings, and prioritize lower credit grade loans, the return could be different:
Using the distribution in the above table, an investor would achieve a 7.3% return. While the default rates are higher as well, it should be offset by the increased interest rates. This distribution would make sense for an investor with a higher risk appetite. Given the distribution, fewer loans would qualify than the first approach, so the investor would need to compare the likely future inventory available with the anticipated amount of cash to invest and determine how much to invest in each loan to reach a relatively diversified portfolio.
In conclusion, there are several different approaches to manage an online lending portfolio. Regardless of which approach is right for an investor, prioritizing investments is helpful in making the strategy even more precise. This is especially useful when there are more loans available than an investor can actually buy, or in the current situation, when many loans become fully-invested quickly, and prioritization is required. The difficulty an investor will encounter is that it is very difficult to actually implement such a strategy using graphical interfaces. In order to effectively and efficiently manage this portion of an investment strategy, a programmatic interface with the origination platform is preferable.