New Fund Launching That Uses Artificial Intelligence to Pick Loans

Ranger Capital

Gary Melara has been working with artificial intelligence (AI) programs for decades. So, when he first considered investing in Prosper and Lending Club back in 2009 he wondered if he could apply that technology to his loan selections. With the wealth of data that Prosper and Lending Club make available he was able to do just that. He has been investing using his AI program since early 2010 with some impressive returns.

Scott Canon is the president of Ranger Capital, a $3 billion investment management company based in Dallas offering both traditional and alternative investment products. He became interested in peer-to-peer lending last year and began exploring the space. Like most of us he was impressed with the high yields and low volatility of this new asset class.

To create a successful fund Scott knew he needed a top-notch portfolio manager who could create a model to generate the best ROI. He interviewed several people for this position including Gary. It was very clear to him that Gary’s approach was by far the best and his external consultants agreed. Gary was also the best fit culturally for Ranger Capital. So, they began laying the groundwork for putting together a fund.

I chatted with both Gary and Scott last week as they told me about their new fund, Ranger Specialty Income Fund, which will be launching shortly. The fund will be investing in Lending Club and Prosper notes initially but down the road they may expand to other high yield opportunities including student loans, small business and international lending platforms.

The Artificial Intelligence Algorithm

I was curious about Gary’s AI algorithm and how it worked in picking loans. He calls his system TruSight Technology. While he wouldn’t give away any details he did say this. His algorithm does not just pick the same popular loans that most of the other credit models also pick. It goes much further. Because his algorithm takes a more holistic approach he will often choose loans that others are avoiding. Here is what Gary said when I asked him why an AI system is better than traditional statistical analysis:

AI enables more profitable loans to be selected because diverse and complex patterns can be recognized unlike “tunnel vision” from filters or statistical analysis.

Here is a brief summary as to how the AI algorithm works:

  • Takes the 35-45 most significant data fields from Lending Club and Prosper.
  • TruSight evaluates each field in complex combinations with other fields.
  • Only the fields and their relationships to other fields that actually impact loan ROI are used.
  • Combined confidence from different algorithms is required to select a loan.
  • Isolates loans with largest spread between borrower rate and charge-off risk.

They said they would have no problem deploying capital into Prosper and Lending Club because they don’t have to compete with everyone else for the so-called best loans. There are always loans available with their algorithm so they don’t have to focus on the mad rush for the new loans. Although, Gary did say he will also try and secure his share of the most popular loans assuming they meet his requirements.

The other unique feature of having an AI algorithm is in its adaptability. If underwriting changes at Lending Club or Prosper, or the macroeconomic environment changes the algorithm can quickly adapt with it. Also, the team has augmented the AI process with a committee of consumer credit experts to help the fund.

Like every other fund in this space The Ranger Specialty Income Fund will be available for accredited investors in all 50 states. It will also be available for international investors. The minimum investment is $250,000 and they are going to start raising money for their fund shortly. If you are interested in learning more you can contact Bill Kassul at

Disclosure: This article is for information purposes only and is neither an endorsement nor a recommendation to invest in this fund. Also, I have received no compensation whatsoever for writing this article.


Notify of
Newest Most Voted
Inline Feedbacks
View all comments
Andrew Johansen
Andrew Johansen
Sep. 3, 2013 10:12 am

Any word about whether they will be funding exclusively whole loans or competing for the fractional pool as well? I would presume both.

Out of curiousity, did they mention a range for their management fee?

Dan B
Dan B
Sep. 3, 2013 2:12 pm

” He has been investing using his AI program since early 2010 with some impressive returns.”

How impressive have these returns been? What are the numbers?

“To create a successful fund Scott knew he needed a top-notch portfolio manager who could create a model to generate the best ROI. He interviewed several people for this position including Gary. ”

I thought that the AI program had been created by Scott back in 2010 & it was the model that in fact generated the best ROI? How does a top- notch portfolio aid in that process going forward?

“The other unique feature of having an AI algorithm is in its adaptability. If underwriting changes at Lending Club or Prosper, or the macroeconomic environment changes the algorithm can quickly adapt with it. Also, the team has augmented the AI process with a committee of consumer credit experts to help the fund.”

The notion that a “committee” of anything can actually agree on anything in order to help anyone else, much less an AI, is rather quaint, but setting that aside,,,,,,,,,,,,,, Since I assume that the AI program has no ego & doesn’t require a pat on the back or other forms of moral support, isn’t the committee really there to try to correct mistakes made by the AI? Does their presence & potential interference not in effect challenge the notion of the AI itself? How would the committee identify the mistakes in the first place?

And what is all this going to cost? What sort of expense ratio & fees (upfront or otherwise) will this fund have……….. 2%, 3%, 4% of portfolio? Or a percentage of portfolio plus a percentage of profits?

I apologize if my responses & questions seem like I’m trying to point out shortcomings in this article or in Ranger’s approach. I’m only here to try to “augment” 🙂

Sep. 3, 2013 2:36 pm
Reply to  Dan B

Somewhere, Marvin Minsky’s lacrimal glands just activated for no apparent reason.

Dan B
Dan B
Sep. 3, 2013 11:52 pm
Reply to  nonattender

Nonattender………I thought you’d be happy. Isn’t this what you were looking for when you recently said something to the effect of…………’d wait a year until some fund became available that made you 2% less than people you called “super users” made but required absolutely zero of your precious time? Well here you go. 🙂

Personally, I’ll remind myself to become impressed with AI & all its relatives when any of them can accomplish something as simple as pick out movies I’d enjoy watching on Netflix. Performance to date can be charitably described as mediocre, despite my 6 years of Netflix ratings & history, to say nothing of the millions of dollars spent by Netflix to develop such a system.

Rob L
Rob L
Sep. 3, 2013 2:45 pm

I wouldn’t want to pick on AI. One of those “it’s only 10 years away technologies” for the past 40 years or so. I’ll just second Dan B’s final question about expense ratios. How many bp’s do I pay for the alpha AI selection provides.

And while I was typing Peter provided the answer. Thank you,

Andrew Johansen (Randawl)
Andrew Johansen (Randawl)
Sep. 3, 2013 5:07 pm

Will the fund be exclusively investing in whole loans or will they be competing for the fractional pool as well?

Will someone be individually screening each loan or is it “set it and forget it” with intermittent evaluation post selection and investment?

I see that they are using the 1 and 10 management fee model. Will Ranger Capital Group be negotiating payment processing fees with LC like is seen with LCA and certain SMA investors? Likewise, what about a CSA (credit support agreement) similar to what LCA investors enjoy?

Michael Fox
Michael Fox
Sep. 3, 2013 7:20 pm

Andrew, quite curious to hear more about the 2 points you raised, namely the payment processing fees negotiation and the CSA. Could you elaborate a bit more or point me to some additional reading on these two issues, very interested in learning more.

Andrew Johansen (Randawl)
Andrew Johansen (Randawl)
Sep. 3, 2013 8:38 pm
Reply to  Michael Fox

From the most recent 10-Q regarding the CSA, page 23:

Credit Support Agreement
The Company is subject to a Credit Support Agreement with a Certificate investor. The Credit Support Agreement requires the Company to pledge and restrict cash in support of its contingent obligation to reimburse the investor for credit losses on Member Loans underlying the investor’s Certificate, that are in excess of a specified, aggregate loss threshold. The Company is contingently obligated to pledge cash, not to exceed $5.0 million, to support this contingent obligation and which cash balance is premised upon the investor’s Certificate purchase volume. As of June 30, 2013, approximately $3.0 million was pledged and restricted to support this contingent obligation.
As of June 30, 2013, the credit losses pertaining to the investor’s Certificate have not exceeded the specified threshold, nor are future credit losses expected to exceed the specified threshold, and thus no expense or liability has been recorded. The Company currently does not anticipate recording losses resulting from its contingent obligation under this Credit Support Agreement. If losses related to the Credit Support Agreement are later determined to be likely to occur and are estimable, results of operations could be affected in the period in which such losses are recorded.

Likewise regarding management fees, page 12:

Management Fees
LCA acts as the general partner for certain private funds (the “Funds”) in which it has made no capital contributions and does not receive any allocation of the Funds’ income, expenses, gains, losses nor any carried interest. Each Fund invests in a Certificate issued by the Trust pursuant to a set investment strategy. LCA charges limited partners in the Funds a monthly management fee, payable monthly in arrears, based on a limited partner’s capital account.
LCA also earns management fees paid by separately managed account (“SMA”) investors, paid monthly in arrears, based on the month-end balances in the SMA accounts.
These management fees are classified as a component of non-interest revenue in the consolidated statements of operations and are recorded as earned. Management fees can be, and have been, modified or waived at the discretion of LCA.

A Lend Academy forum topic on Certificate Investors:

A link to the most recent 10-Q:

Sep. 4, 2013 2:44 am

Machine learning is not new; this is simply applying data science / smart data analytical solutions to p2p lending. I do believe this will be a trend moving forward – expect to see more of these plays in the future. Or for Prosper / Lending Club to offer greater level of analytics of their platform data as a paid service.

Ranger Capital Group
Sep. 5, 2013 9:21 am

Peter, thank you for the article on our new fund. We appreciate all that you are doing to promote and educate P2P lending.

We wanted to respond to a couple of the questions your readers had about our AI technology and how it is being used in selecting loans.

Your reader is absolutely correct in that you do not want to place any filters or limitations on an AI system. Our AI system has over 3 ½ years of successfully selecting loans that potentially could yield the highest possible ROI. But we did not want to have a loan selection process that was singly based on technology. Our Credit Committee has a diverse and extensive consumer credit experience that we wanted to leverage. Some of their responsibilities include monitoring the fund performance, using their knowledge of current credit conditions to tune the AI selection program and to provide the AI system additional credit data that supplements what is available on the P2P platforms. The Credit Committee involvement will be used to enhance the excellent returns that have been achieved thus far as opposed to being a drag on it as some have suggested. We feel that blending technology with an experienced credit committee is the best approach for this market.