P2P Lending / NFT Lending Forum

https://forum.lendacademy.com/index.php?topic=1560.msg23102#msg88888888Quote"> from: Randawl on September 14, 2014, 01:31:56 PM

You don't provide the username/password, you provide your authorization token.

The only change I made recently was to use the API vs. the CSV, at which time I also learned to pull only the _new_ loans. Are you sure you weren't just investing twice or even three times in the same repeated picks? Over the weekend I could check to see how the volume of picks has changed across the last few weeks.

True story. My system didn't even run due to lack of a worthwhile loan drop.

Or I could simply run the algorithm on historical originations (not applications--too cumbersome) and provide some aggregate numbers. The algorithm has changed a number of times over the prior year. The most-recent iteration might be best.

Hot off the press, LendingClub filed their S-1 today. You can read the document at the SEC EDGAR tool. Update this thread with any interesting tidbits!

http://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001409970&owner=include&count=40" class="bbc_link" target="_blank">http://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001409970&owner=include&count=40

Our philosophy is to base picks off statistical evidence. Just as there are no picks above 25000, 60 month terms, etc., it will be 3 long years before we model DTI greater than 35. To be fair this change may need us to review our script because 35 was previously a hard cap so we didn't need a line to remove things in excess. I'll have to check.

I appreciate the intent, but I don't think this was the right approach. Nathan, I think we should gather up the 3rd-party players, have a conference call, and pound out a revision and pass it by them. If everyone groups together, then we have some leverage in that if nobody agrees to use the API, there is no third-party ecosystem.

Logging into LendingClub.com [trigger]...
[Thu Aug 21 10:00:28 PDT 2014]: Loans on platform...
[Thu Aug 21 10:00:28 PDT 2014]: 684
[Thu Aug 21 10:01:01 PDT 2014]: 606
[Thu Aug 21 10:02:00 PDT 2014]: 598
[Thu Aug 21 10:02:49 PDT 2014]: No rise detected. Aborting main script.

I probably missed the rise because of starting looking at 28 seconds after the hour. By then perhaps it was already loaded. I've changed the script to start at 1 second after the hour. However, you can see that people are buying up paper quite quickly between 28s after the hour and 1 minute after the hour (the first moment loans are seeable in the CSV file). See my "asymmetrical timing" thread for more details.

Dear Subscriber,

P2P-Picks has maintained a beefy web server to deliver our LendingClub picks since our launch about 18 months ago. However, in recent times, such delivery has been inefficient. The picks are often fully invested within a few seconds of release, leading to stale picks (clicking the "Go" button with no investment). At the same time, people have asked for more automation solutions because they do not have time to log into our site multiple times a day.

So, we have deprecated the publications from our website and all picks will now be delivered via API or through one of our partners. To continue using P2P-Picks, you have a number of options:

1)   Sign up with one of our delivery partners, such as BlueVestment (write to Nathan, as we have a grandfather clause to allow signups with his system). We are working to bring on another delivery partner within the coming two weeks. Stay tuned.

2)   If you are a programmer and comfortable taking things into your own hands, write to us for a DIY API kit. We will provide our API specs and you can interface with our systems and obtain our publications directly.

3)   Invest with LendAcademy Investments, a registered investment advisor, for which we serve as a credit modeler.

Yours Sincerely,

Bryce Mason
Principal

The problem is that so many offers require a 4 point fee for the cash advance. If I had cheap debt I'd probably give it a go.

You can make any extra filter that you like on BV, if these bother you.

I have tested whether months since last major derogatory makes any difference above and beyond my standard correlates and it has not.

I cannot test current delinquencies as that data is not provided historically. In fact, LC only started showing it recently. I have to presume that the distribution of people we see today with these characteristics is similar to what it was in the past when we were blind to it. I do not let it bother me, even if there is some face validity to the idea that they might be higher risks. When the data comes in, I will adjust if necessary.

I noticed the 10am load was like 2 seconds after the hour. I'm going to have to make an effort to observe what's going on this weekend and adjust things.

Nice, WTG. I should have an announcement soon about a partner for retail.

I use Top 5% on PMax, D/E only. Sometimes I go through a drought where I might get 1 note all day. However, in recent days, volume has been very good.