AC - QAA a staggering 3.75%

It may depend on whether AC have seen a large influx of funds into the QAA from outside sources.  If they did/do, and they want to retain that money within the platform even if it just goes into the QAA and just parks there, then a significantly increased QAA limit might come sooner rather than later -- and your finger in the air guesstimate might turn out to be very prescient.

sl75:  AIUI, the problem AC were trying to solve with the QAA was that lenders weren't leaving much idle funds in their accounts so that they could increase their holdings to their targets when parts became available -- either because other lenders selling or because underwriters releasing parts immediately after drawdown.  I expect many were doing that because idle funds reduce lenders returns and most 'active' investors try very hard to avoid that.  Since the QAA allows lenders to earn 3.75% on what would have been idle funds, AC might have expected more funds to be made available, which would help uptake of newly drawndown loans and Aftermarket liquidity.  As an extra bonus, AC could use a goodly proportion of the additional funds provided to sop up parts of their loans, and contribute to their bottom line the difference between the interest rate on the loans in the QAA and the amount paid out on QAA holdings plus any contributions to the Protection Fund.

The problem I -- and others -- are trying to fix with limit suggestions, is that the 3.75% QAA rate is sufficiently attractive to expect that some people will use it to park money they otherwise might have put in banks and building societies.  With the current overall QAA limit of £1M, all it would take is for 40 people to 'park' £25k each in the QAA for the new account to stop functioning as designed for active AC investors since their idle funds would become truly idle again, always queued for the QAA but never getting in.

If AC were to be able to increase the overall QAA limit to the point where it isn't actually limiting then, yes, the problem would go away.  And while a modest increase might achieve that right now, once the word spreads about the QAA and it starts showing up in the Best Buy tables of comparison websites I'd expect the punters to come flooding in and swamp the original purpose of the QAA.  IMHO, AC need to be planning now what they're going to do to deal with the deluge.

The average withdrawal number is now up to 0.01 seconds.

It's a simple average as that made more sense to me.  Doesn't matter how big or small the transaction is.  Either the account allows rapid withdrawal or it doesn't, and the circumstances in which it wouldn't are the same for large and small transactions alike beyond what would be a brief transitionary period.

What are lenders actually looking for in that figure?

We could look at things like excluding the 10% fastest and 10% slowest transactions to remove outliers and then produce a weighted average of those in the middle.

Does it make sense to average the last 7 days or should it be a smaller fixed sample size?

Does it matter to a new investor that there was a slow day 7 days ago which is skewing the figures but it's been performing perfectly ever since?

What about bugs, should they be allowed to heavily skew the figures even when they're resolved and cannot reoccur?

Let's see how it performs.

Whereupon, you stick £25k into your MLIA, set your targets (huh, those were the days!) to authorise a Total Net Instruction of +£50, and let the system sweep your idle.