Jump to content


Photo

Gaming the new Zopa


  • Please log in to reply
7 replies to this topic

#1 Gricehead

Gricehead

    Zeus

  • Members lvl 1
  • PipPipPipPipPipPip
  • 727 posts
  • Gender:Male
  • Location:in a quiet corner of Yorkshire

Posted 18 September 2012 - 11:53

Note: this is highly theoretical at the moment, as rates are too low for me to enter the market.

Now that the new Zopa has tightened the range of rates that match in a market, effectively ruling out snowballing as a valid exercise for the moment, I got to thinking how else I could maximise matching speed.

Round robin matching means that initial queue position is no longer a valuable asset, so it becomes much more easier to add and remove offers from a market without being penalised.

With the current A* shorter market being effectively spanned by offers at 6.4% (it's not possible to tell how much further beyond the 20000 mark 6.4 extends), it must be possible to seed multiple offers into the round robin list at the same rate, keeping them far enough apart (by time of last match) to limit the risk of over-exposure through double matching. The wider the span of the 6.4 rate, and the higher your tolerance to double matching, the more offers you could seed into that market, doubling, trebling etc. your matching speed.

Of course, this would require some micro management via the Market Data report to ensure your offers weren't closing in on each other, but if they are you can use the principle of Formula 1 pits stops; pit stop one of your offers out of the market, and feed it back in in a suitable "gap".

Thoughts on the theory appreciated, and apologies if it's already been posted/discussed in one of the many threads I've not caught up with.

#2 sl75

sl75

    ZopaHolic

  • Members lvl 1
  • PipPipPipPipPipPipPip
  • 6,918 posts
  • Gender:Male

Posted 18 September 2012 - 12:33

Looking at the current MarketQuoteReport, at this present instant, only offers up to 6.3% get a look-in even for a maximum-sized 15k loan...

... so no matter how many offers you have in the 6.4% queue you're definitely not going to be included in the very next loan to be matched.


Also, I'd tend to think that most lenders don't have enough funds to keep even a single offer with 10 exposure permanently funded at within-ZOPA rates. Such offers which periodically run out of funds will "always" end up back at the front of the queue when they're next funded (whether by topping up or by a loan application to which they're matched being declined).


However, someone HAS to do a visualisation of the MarketQuoteReport that depicts the lending queue as a racetrack and highlights your own offers as "cars" on that racetrack! Maybe some kind of smartphone app that lets you edit offers using some kind of virtual racing game? Not sure what it would need to do about the non-visible portions of the racetrack though? ... or how it would discover exactly how long the racetrack is - are we looking just at the start-finish straight, or all but a few hundred metres of it?

#3 propman

propman

    Zeus

  • Members lvl 1
  • PipPipPipPipPipPip
  • 2,327 posts
  • Gender:Male
  • Location:Thames Valley

Posted 18 September 2012 - 13:55

I admit to doing something similar on RS.

Issues are:
1) Any offer is on 2 markets (24&36 or 48&60) and positions and speed of rotation vary so while a few offers might be someway apart in both, managing several would be likely to keep putting one or other close to another offer.
2) If the quotes are coming thick and fast, many offers may be depleted or if the market moves, a significant number of offers may change rate bringing offers closer together, so you really have to be prepared to match most of the offers at a given rate with a single lender of write off a lot of time refreshing the data each hour to compensate.

I would say that I am going to split offers of >10 per borrower into seperate offers to maximise the number of loans matched.

- PM

PS: I think sl75's point re keeping offers funded may be true for many lenders if teh front of the queue matches at a small loan rate, but where this is say 5k the point is much more relevant as it would be if trying to match a large amount of funds quickly (eg if the rates were to increase significantly again).

However, as always this is merely how to get 0.1% more for a lot of work!

Edited by propman, 18 September 2012 - 13:59.


#4 MikeS1531

MikeS1531

    ZopaHolic

  • Members lvl 1
  • PipPipPipPipPipPipPip
  • 5,149 posts
  • Gender:Not Telling
  • Location:Herefordshire

Posted 18 September 2012 - 16:22

I am going to split offers of >10 per borrower into seperate offers to maximise the number of loans matched.

This would seem to be a sensible solution for any lender who is willing to lend more than 10 per borrower. The worst case result would be multiple offers positioned together in the queue, which would mean matching fewer requests, nearly all of which of which would take the maximum amount the lender is willing to offer to a single borrower -- but that's the same result a lender would have if they had just one offer with a larger exposure. The best case result would be matches of smaller amounts to more borrowers but the total lent should be virtually the same. Increased numbers of borrowers for a given size of loan book generally is considered a positive thing, so I see no downside for this approach. Might I be missing something?

#5 momac

momac

    Zeus

  • Members lvl 1
  • PipPipPipPipPipPip
  • 729 posts
  • Gender:Not Telling
  • Location:North Yorkshire

Posted 18 September 2012 - 16:29

Silly question time!

If I have one offer to all markets at different rates for each market, then I am matched with, currently, a Y36, does that mean that my offer in A* also goes to the back of the queue?

Apologies if I have not understood all that I have read. :blush:

#6 momac

momac

    Zeus

  • Members lvl 1
  • PipPipPipPipPipPip
  • 729 posts
  • Gender:Not Telling
  • Location:North Yorkshire

Posted 18 September 2012 - 16:32

Sorry, everyone. I have re-read the stuff and my question is ridiculous! Please relax and forgive, M

#7 sl75

sl75

    ZopaHolic

  • Members lvl 1
  • PipPipPipPipPipPipPip
  • 6,918 posts
  • Gender:Male

Posted 18 September 2012 - 17:09

... Increased numbers of borrowers for a given size of loan book generally is considered a positive thing, so I see no downside for this approach. Might I be missing something?

Beyond a certain point, I personally see it as a negative - it makes things harder to mentally keep track of and/or analyse. I tended to consider anything below about 0.2% per borrower "fiddling small change" that I'd really rather not be bothered with, in order to keep the loan book manageable.

#8 MikeS1531

MikeS1531

    ZopaHolic

  • Members lvl 1
  • PipPipPipPipPipPipPip
  • 5,149 posts
  • Gender:Not Telling
  • Location:Herefordshire

Posted 18 September 2012 - 21:48

Beyond a certain point, I personally see it as a negative - it makes things harder to mentally keep track of and/or analyse. I tended to consider anything below about 0.2% per borrower "fiddling small change" that I'd really rather not be bothered with, in order to keep the loan book manageable.

I can understand that. But 0.2% per borrower would mean a loan book of just 500 borrowers -- or possibly double that once the book is mature and the average capital outstanding is half the original principal. That might be OK for a lender offering in a limited number of markets, but spread over 10 markets there wouldn't be enough loans in some markets to produce a high probability of the lender's bad debt experience being close to the Zopa average actual experience.

Of course, that might not bother some people. And I can speak from experience that having a large loan book does mean that it's very time-consuming to try to keep aware of what's happening with all of my 'non-performing' loans. But I'm not sure it's making much difference to my ability to analyse my account. That's done with a computer and, aside from the need for a few extra seconds of recalculation time and a bit more hard disk space to store the data, the size of the data base has little impact -- for me anyway.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users