Author Topic: LC vs Prosper  (Read 3422 times)

rawraw

  • Hero Member
  • *****
  • Posts: 2671
    • View Profile
Re: LC vs Prosper
« Reply #15 on: July 17, 2017, 04:30:37 PM »


What do you have in mind as far as a test that would prove your hypothesis incorrect?  If you give me something I may be able to take a look at it.
Here is an example of how Bryce did one: http://www.lendacademy.com/comparison-of-different-p2p-lending-investment-strategies/

Well, since I assume the pricing is much more efficient, I think it depends on your goal.  Your goal with the score seems to be loss minimization?  Then I'd show how that would have worked in 2015 vintages and if the impact is large enough to justify the fee a customer would have to pay. 

RomanLegend

  • Newbie
  • *
  • Posts: 30
    • View Profile
Re: LC vs Prosper
« Reply #16 on: July 17, 2017, 06:25:54 PM »
Yea loss minimization first, maximizing interest rate thereafter.  Price point for the app would be very low.  I'm thinking <$35 annually or pay as you go for picks.  They would have really inexpensive packages of 'x' amount of picks to purchase.  Or some variation of that model.

We did do back testing with a good volume of loans.  I personally don't mind sharing but let me double check and get back with you.  The short story is the back testing did revise my approach a bit.  It really eliminated using loans from A1 to B3 based on the results.  This handicaps my ability to offset approaching charge offs but also reduces a large portion of the early pay off notes as well.  Delicate opportunity costing...

-Chris



What do you have in mind as far as a test that would prove your hypothesis incorrect?  If you give me something I may be able to take a look at it.
Here is an example of how Bryce did one: http://www.lendacademy.com/comparison-of-different-p2p-lending-investment-strategies/

Well, since I assume the pricing is much more efficient, I think it depends on your goal.  Your goal with the score seems to be loss minimization?  Then I'd show how that would have worked in 2015 vintages and if the impact is large enough to justify the fee a customer would have to pay.

nonattender

  • Hero Member
  • *****
  • Posts: 685
  • I am not here.
    • View Profile
Re: LC vs Prosper
« Reply #17 on: July 17, 2017, 07:51:23 PM »
Delicate opportunity costing...

What's a bottle of Ritalin going for?  It's nice that you're fascinated with your new-found toy, and all, but...  control it, will ya?  Thanks.
A little nonsense now and then is relished by the wisest men.

Rob L

  • Hero Member
  • *****
  • Posts: 1714
    • View Profile
Re: LC vs Prosper
« Reply #18 on: July 17, 2017, 08:27:40 PM »
Generic methylphenidate is covered by most Medicare and insurance plans, but some pharmacy coupons or cash prices may be lower.
The lowest GoodRx price for the most common version of methylphenidate is around $25.74, 70% off the average retail price of $87.80.

Meanwhile if you want to invest with LC, really, just download the big file, use "R" to run a logit regression, and go with it.
There are plenty of secret sauce clues in prior posts by others (one in particular) that suggest the best independent variables of value.
Otherwise find your own.

AnilG

  • Hero Member
  • *****
  • Posts: 1038
    • View Profile
    • PeerCube
Re: LC vs Prosper
« Reply #19 on: July 18, 2017, 02:03:11 AM »
+1 I can't upvote this enough, the most sensible advise I ever read on LA forum.

If you are using more than 4-6 important variables to select loans you are doing it wrong. Also, please stop backtesting credit variables for returns directly and trying to squeeze out 0.1% extra return in your backtesting. Model risk (not return) with credit variables. There is a reason risk-adjusted return term exist and not return-adjusted-risk or return irrespective of risk terms. Also evaluate whether the extra return in backtesting is statistically significant or not.

Meanwhile if you want to invest with LC, really, just download the big file, use "R" to run a logit regression, and go with it.
There are plenty of secret sauce clues in prior posts by others (one in particular) that suggest the best independent variables of value.

Otherwise find your own.
---
Anil Gupta
PeerCube Thoughts blog https://www.peercube.com/blog
PeerCube https://www.peercube.com

RomanLegend

  • Newbie
  • *
  • Posts: 30
    • View Profile
Re: LC vs Prosper
« Reply #20 on: July 18, 2017, 01:08:22 PM »
Hahaha, this from the guy with the tag line "A little nonsense now and then is relished by the wisest men" ......curmudgeon.

Delicate opportunity costing...

What's a bottle of Ritalin going for?  It's nice that you're fascinated with your new-found toy, and all, but...  control it, will ya?  Thanks.

nonattender

  • Hero Member
  • *****
  • Posts: 685
  • I am not here.
    • View Profile
Re: LC vs Prosper
« Reply #21 on: July 18, 2017, 08:34:48 PM »
A little nonsense now and then is relished by the wisest men.

RomanLegend

  • Newbie
  • *
  • Posts: 30
    • View Profile
Re: LC vs Prosper
« Reply #22 on: July 18, 2017, 09:27:14 PM »
haha