nothing lent on ISA side on 25th

yes - same for me several days of no substantial change - hopefully some quality borrowers will come along soon 

Mr Misotu's lending slowed to a trickle way back in August and has never recovered. I haven't been able to add significant new money for over six weeks. At one point last week, he was lending so slowly that repayments pushed his funds on offer up over the £2000 threshhold.

At a time when average queue/match in Core has consistently been under 72 hours (according to the weekly statements), he's had new money unlent for weeks and weeks. I am currently in detailed discussions with them about this.

Most of the replies I've had so far a) contradict each other, b) make no sense anyway and c) answer a question I've not asked. But the last response I had (while still not addressing the question I asked!) was much better quality and provided some interesting information.

My Core ISA was 0.3% below the projected rate of return. Because of this, apparently, the algorithm prioritised my lending with a view to improving my overall rate of return.

What actually happened was I lent out the remaining £9,000 or so of my allowance extremely rapidly - up to £2000 per day and could easily have lent more. But here's the curious thing. I was matched largely to low-yield secondary market loans and now my whole ISA allowance is lent out with a projected yield of 3.4% - *still* 0.3% below average! Hard to see how Zopa plans to address this now, given that no further new funds can be added, but the algorithm (and Zopa) appears broadly to be confusing speed of lending with yield. I would have thought they'd need to deliver a basket of loans with a projected return of around 4.1-4.2% to lift my overall projected return to about average. And I'd have thought this would in fact slow down my lending speed, since fewer loans would be eligible for matching to my funds. But according to Zopa, lending almost half my ISA allowance at lightning speed and at below projected is the algorithm "working correctly".

Since Mr Misotu's Core ISA has a projected return of 0.2% above the average, I wonder if his account, conversely, has been "deprioritised"?  Zopa haven't actually addressed this yet, but his lending speed is really ridiculously slow. And oddly, even though there have been a zillion low-priced secondary market loans around, he continues to match to higher-rate loans and, unlike me, hardly ever to a loan returning <2%. So his projected return remains stubbornly above-average. 

Our discussions continue. They did actually make this comment: "That being said, we have taken your experience into consideration and have already started making adjustments to the matching system to limit the impact of this behaviour in the future and I would like to thank you for bringing this situation to our attention." 

So they seem to be acknowledging that there's a problem. But since they didn't indicate which bit of my "experience" was relevant, nor to which ISA account they referred, I have absolutely no idea what they consider the problem to be.

.

As with any other investment, more diversification (smaller chunks, in Z's case) just reduces the deviation from whatever the mean turns out to be .. the FC site used to have some nice graphs showing the effect in detail. Doesn't affect the mean value, but reduces the chance you will be luckier/unluckier than average. You can even work out how much it reduces the chance (some math required).

I always lend in £10 chunks and have done so since 2011 - that's one of the reasons why I was so exasperated when Mr Misotu's repayments pushed his funds on offer over £2000 (he was lending out more slowly than repayments were coming back). 

I lent £9,000 in around a week by topping up daily. Way more than 900 matches, because around 75% were secondary market loans, which gives you a feel for how fast my funds were shooting out of the door. I am pretty sure I could have lent even faster if I'd topped up more frequently than once a day. 

You might take the view that "whether the Zopa algorithm is correct remains to be seen". My view is that it demonstrably is not functioning correctly when it takes action specifically designed to *uplift rate* but which delivers no improvement in rate whatsoever on a largish sum lent out in the smallest chunks possible. That, quite frankly, approaches my definition of "broken".

My ISA loanbook is £20,128 diversified over 2381 loans. Sufficient, I would have thought, to minimise deviation from the norm pretty effectively. 

No worries aju 

Where Zopa and I differ on all of this is I do accept that accounts will behave a little differently, but I think more than a certain deviation from average isn't acceptable unless the algorithm identifies it and sorts it out effectively over a reasonable timeframe.

Most people lending will be fine because statistically they will be getting around the average return. But the algorithm needs to look for the exceptions and deal with them *effectively*. Right now, Zopa's line seems to be "well, you are 0.3% below projected average for Core, so the algorithm has prioritised your account and this means it is working correctly". So far so good - but all the algorithm actually did carry on lending out all of the rest of my funds at 0.3% below projected ... but very fast! So now I'm stuck with my whole ISA allowance returning below projected and the algorithm continues to lend my repayments super-fast but at low rates. 

They seem perplexed by the fact that I don't think this is what should be happening    I would have been ok with a slightly lower than average speed of lending if it had hiked my overall projected return up to around the average. But that is absolutely not what happened.

ok misotu I feel your pain a little more see wow - 89 loans - £887 lent in <12 hours

the curious thing for me was many were £10 loans but were from SG relends, they are mostly SG's but its still a bit odd to say the least.

I'm hoping that now I have invested all the large sums and only 1000 or so still in queue, it will settle down into a better pattern of slow relending time will tell I guess.