Clicky

  • Welcome to P2P Lending / NFT Lending Forum.
 

ETH.LOAN

News:

This was the original Lend Academy peer-to-peer lending forum, since forensically restored by deBanked and now reintroduced to eth.loan.

To restore access to your user account, email [email protected]. We apologize for errors you may experience during the recovery.

Main Menu
NEW LOANS:   | muk.eth 1.500 Ξ | torkoal.eth 0.200 Ξ | tropius.eth 0.200 Ξ | ALL

P2P-Picks.com Beta is Live

Started by Peter, February 13, 2013, 11:00:00 PM

Previous topic - Next topic

brycemason

Hi forum goers! As colleagues in the P2P lending space, I wanted to share that my website is up and currently in beta. For the last year I have been creating statistical models to drive my LC investments ($38k so far) and working on a nice interface to give subscribers access to my research. I invite you to visit my site and ask questions or provide any feedback here. I'm eager to hear the thoughts of people. Also, if you would like to be a beta user (while things are still free to use), just drop me a line (sending me a pm here will do) and tell me your name and email address.

Bryce Mason
http://www.p2p-picks.com" class="bbc_link" target="_blank">www.p2p-picks.com

Keltset

Bryce, I have followed your posts and loved your insight. I hope to be able to check out your site and a big congrats on getting it up and going!!

Peter

I have been following Bryce's progress for the last year and have been really impressed by what he has done. I have been investing in his picks through one of own accounts for several months now. I think it is the most rigorous statistical analysis that has been done in this space, at least from a retail investor standpoint.

Congrats Bryce in getting it off the ground.

Peter
Publisher of the Lend Academy blog

See my returns here: http://www.lendacademy.com/returns

rawraw

Bryce, is there any way to figure out what impact the old underwriting criteria had on items?  It seems like there may be relationships among variables that may change due to the change in underwriting.  And cool site, do you plan on adding a ability to track portfolio returns?

TravelingPennies

Rawraw! Easy answers.

The underwriting changes don't affect the modeling at all, because I use factors that don't change based on LCs credit risk model. A person's credit score is what it is, for example. The upside of is is that if LC lowered their standards tomorrow, the model just wouldn't pick anything.

No need for portfolio tracking as (1) users can do that themselves in LC if interested an (2) there is no reason to expect any differences among subsets of my picks because they are all essentially the same in terms if expected return.

HTH,

Bryce


TravelingPennies

Bryce,

I love the site so far and after looking at a few of the picks was very happy with the decisions the model is making. One thing that I particularly love about your site is that your pay structure would be on a per pick basis. I do understand this means there is a limit of information that you would want to provide to avoid scaling picks without driving them through your payment method. I would love to see some more basic information provided though. It would be great to have some more basic info like loan interest and expected return rate prior to moving onto the payment 'go' button. I understand the site is in it's earlier stages of development and you may have this kind of thing already planned out.

I did make a suggestion for your site already here to have it's own forum section. I do believe that you have modeling that fits into how I would like to fully implement it myself and believe that your site will be making an impact on my own modeling and purchasing procedures.

TravelingPennies

If you have specific fields that would help your decision making, please let me know. I thought a great deal about how much is really necessary, but if people can make a great argument I'll consider it. For example, maturity is something that might be important to the user. Same goes for the broad loan grade, as some lenders might further subdivide the picks based on their appetite for the likelihood of default (i.e., some users might wish to employ a mini max strategy where they buy into the idea of the profit maximization but, given that, also want to minimize their risk of defaults by choosing only the higher graded notes).

% funded might not be necessary any longer, as the picks are getting posted so frequently. And I might just remove the "Top" as it's causing confusion; the picks essentially all perform roughly the same in expectation and there's a lot of variance in the 1% notes, so people would probably be better off just randomly choosing among them all. Plus, it would take a long time to build a portfolio of 1% notes. This would give room to put other fields.

Of course, changing the upload file structure is a sort of big deal. Hehe. Lots of scripts to alter, from the statistical script, to the uploader on the web server, to the database tables.

TravelingPennies

I assumed you were being vague on loan details so someone couldn't figure out what is being picked and then do it without paying.  Is this correct?

TravelingPennies

Rawraw, there's an element of that, yes. But really once a person has bought into the method, there are few reasons to discriminate further on the types of loans chosen. I tried to think of reasons unrelated to the picking process that might make sense to display. I'm open to others if there's a reason that's functional beyond the scope of trying to improve results.

Useful Field and Potential Reasons

Maturity: "I'd like my money back faster"
Loan Grade: "Perhaps I don't intend to carry many notes and would like to reduce the variance around my results."

TravelingPennies

For any new discussion on P2P-Picks there is now a special section for them on forum. New topics can be started here:
http://www.lendacademy.com/forum/index.php?board=12.0" class="bbc_link" target="_blank">http://www.lendacademy.com/forum/index.php?board=12.0

breitenm

can you give a few statistical performance measures of your model? For example, area under curve?

TravelingPennies

Yeah, I'll be happy to post some ROC stats. Will be Tuesday probably.

viking

What parameters are you using in your model?

For example, Employer is not available in LCs history file, but I never invest in loans where Employer=N/A.

What about months since major derogatory ? Data is only available since Aug 2012, so there is no significant history to analyze. I avoid loans with a "recent" major derogatory.

TravelingPennies

Hi Viking, sorry I'm not willing to post the factors that are included in the model.


NEW LOANS:   | muk.eth 1.500 Ξ | torkoal.eth 0.200 Ξ | tropius.eth 0.200 Ξ | ALL