|
|
Post by GSV3MIaC on Nov 5, 2014 21:14:32 GMT
It does seem to look for the highest interest rates, just an adequate interest rate. I think there is a 'not' missing from that sentence? Apart from that I agree with your assessment of the black box .. it probably hits the primary market first, IMO, but for a newly arrived lender it will not be able to dispose of all funds quickly unless it taps the SM too. For recycled (repaid) funds I don't know what the PM/SM split might be .. I see SM sales (at par) nightly when the repayment run has probably happened, but whether that is because there is no primary auction which the funds can legitimately be applied to, or whether there is actually a balance (rather than 'PM first, SM second') I can't guess. We do know that it seems to shun the primary market for loans which are already pretty much filled, it doesn't seem willing to go underbid people very much!! I continue to regard it as a) stupid and b) misnamed (should be 'autoinvest' .. buying on the SM is NOT 'bidding').
|
|
blender
Member of DD Central
Posts: 5,719
Likes: 4,272
|
Post by blender on Nov 5, 2014 22:18:14 GMT
It does seem to look for the highest interest rates, just an adequate interest rate. I think there is a 'not' missing from that sentence? ... Thanks GSV. There was a not and I changed the wording to be less definite, but forgot the not. Have corrected it. I think the basic point is that it does not choose the best offers first and then allocate to lenders, It seems to step through the lenders and find the first part which fits the requirements. It definitely could do better for the buyer, but it is also there to sell portfolios, even rubbish portfolios, and so it has to be judged against both purposes.
|
|
|
|
Post by thesnoop on Nov 5, 2014 22:50:26 GMT
Oh thanks guys.
Filling loans early would make sense for FC, I guess the manual army would rather avoid tying up cash for the duration of an auction.
A tendancy towards the mid/lower end of the user set scale could possibly keep average rates attractive for borrowers...
It makes sense to hit the SM only when there aren't enough new loans needing filling and also, perhaps, to monitor SM activity and play with the weighting to ensure liquidity.
I get why it only buys at par too , that rules out all sorts of SM mischief.
Interesting to know it can only hit each loan once per lender... and that it is busy at night. I guess makes sense from diversification and also for spreading load on their computational resources.
I'd be very interested to see sell_rate vs loan_size graphs for loan_parts offered at par. Something tells me smaller loans might have an edge, just a guess though.
|
|
|
|
Post by GSV3MIaC on Nov 5, 2014 23:01:00 GMT
Not sure what you meant by 'user set scale' .. At auction autobid will put in exactly one bid, as large as allowed, at the rate the user sets, but if it constrained by lack of cash (usually the case) it will first look at auctions with low % funding. On the SM it'll buy at par the biggest part it can find which meets the users set rate/ exposure in any given loan .. Even if it could have bought the same exact part (eg fixed rate property loan parts) at a discount.
Yes, autobid users keep average rates at auction lower than they would otherwise be.
|
|
fasty
Member of DD Central
Posts: 1,038 
Likes: 388
|
Post by fasty on Nov 5, 2014 23:15:42 GMT
I've noticed what seems to be another peculiar characteristic of autobid buying on the SM: It seems to have a habit of buying parts from the same loan in sequence. For example, today I sold 2 parts of property loan 7220 one immediately after the other, but to different people. Later I sold 2 parts of property loan 6215, one immediately after the other, again to different people. Later I sold 2 parts of property loan 6248, one immediately after the other, again to different people.
This seems to happen too regularly to be coincidence. Odd.
|
|
blender
Member of DD Central
Posts: 5,719
Likes: 4,272
|
Post by blender on Nov 6, 2014 0:09:47 GMT
My experience is that the autobid lender purchases are grouped. For example 5th Nov just gone, am having a clear out and 52 loan parts sold. In that there was one sequence of five sales to the same buyer, different loans all at par all in £40-£60. And close to that another group of three from a different buyer. On 4th, twos, threes, fours, a five. In one case a single autobidder took four than a few other sales and then another two different from the buyer of the four. Autobid seems to do scheduled runs and works by buyer first, loan second, at least for me. Not that it matters. (Nothing known to be wrong with these loans, BTW, and in 2013 they would have been at a good premium.)
I wish they would time stamp the sales.
|
|
blender
Member of DD Central
Posts: 5,719
Likes: 4,272
|
Post by blender on Nov 6, 2014 0:10:30 GMT
Sorry, repeated
Re-using this space to reply to GSV below : that's well beyond my capabilities but thanks for info.
|
|
|
|
Post by GSV3MIaC on Nov 6, 2014 8:38:59 GMT
They do time stamp the sales, and that data is sent back in the .json data object, to the webpage java script, which chooses not to show it. Ditto the percentages which are sent to much higher accuracy than displayed. You can extract the data if you really want it, but it is not trivial. The sales are displayed in correct time order, even though you can't see the detail. ----------------------------- PS OK, instructions if you want to try. To get the iframe directly (assuming you are logged in, and have already done the necessary 'sort by' and 'x parts per page'), go to www.fundingcircle.com/loan_parts_sold/At the very bottom of this page (as inspected by 'view page source', in whatever browser you favor) you'll see the .json data object, of the form "appData['loan_part_sales'] =". In there you'll find the times as well as the dates (some magic decoder ring work may be required). Now this is FC, so it could all change / revert / die horribly at any time, but that works right now. I guess the time is not being displayed simply owing to lack of columns in your assumed browser, although on my widescreen monitor there is acres of room, there isn't on a 7" tablet or smart phone.. or maybe they just thought nobody would want to know. |
|
