ashtondav
Member of DD Central
Posts: 1,805
Likes: 1,087
|
Post by ashtondav on Jun 15, 2019 16:10:29 GMT
Classic stats error of assuming linearity.
very little in life, economy, finance or stats is linear....
|
|
|
Post by rweb on Jun 15, 2019 16:26:52 GMT
Classic stats error of assuming linearity. very little in life, economy, finance or stats is linear.... Except that the relationship between date and sell time *has* been linear, so no assumption has been made there, and it is reasonable to assume that the trend could continue. An R2 value of 0.98 is fairly clear to me....
I did say, but perhaps not clearly enough, that the projection assumes that no other factors change the trend and it is always important to remember that when using statistics - this to me is the classic stats error!
What happens going forward is in FCs control as well as a number of other variables including volume of sell requests, new investors adding money, level of bad debt etc all of which could affect the trend.
However, I'm mostly trying to demonstrate that it would appear to me that FC have made a significant unannounced change to their model at some point in January and at that time (with the understanding of the linear relationship I'm presenting) you could have accurately told people when they would sell their loan parts up to yesterday. I was also trying to show others who suggested that the increase was exponential and it is not. I think we would all be much more concerned if it were!
It will also be interesting to see if the next few sales continue on the trend, including my own.
|
|
|
Post by Ace on Jun 15, 2019 17:54:04 GMT
It could well be that the true R2 value is exactly 1, and is being deliberately controlled by FC. It seems much too closelly correlated over too long a period to be a coincidental accident of the many varying factors involved.
The measured deviation is likely explained by a few people misremembering the sale start date. This doesn't seem to be recorded anywhere, so unless everyone wires it down they'll have to make a best guess.
Well, that's my conspiracy theory for the day! Could be a load of horlicks, but looks far too linearly correlated to me.
|
|
Stonk
Stonking
Posts: 735
Likes: 658
|
Post by Stonk on Jun 15, 2019 20:10:05 GMT
Just doing a little more work on this and forecasting sell dates based on the linear projection.
The linear equation works out at y=0.5348x-23244 (based on date x values in excel and y being the number of days to sell). The R2 value for the trendline is 0.9852. A value of 1 would mean a perfect linear correlation coefficient, so at 0.98 we can be confident that so far sell times have been increasing at a very steady and predictable rate.
That means going forward, and assuming the rate of growth in sell times remains constant, we can expect:
- 25/4/19 (Ref. indexwrangler), next on the list in criston post #1 to be sold on or around 26th June after 63 days
- Someone selling yesterday 14th June could expect to sell on or around 12th September after 89 days
A date for your diaries: 5 May 2028: "The Big Crunch".
This is the day when the time expected for a sale to execute exceeds the length of the longest loan term that FC offers.
On this day, FC will be able to retire the "Sell" functionality, safe in the knowledge that it could never again be of use.
|
|
|
Post by kcrane on Jun 16, 2019 21:34:00 GMT
This is the day when the time expected for a sale to execute exceeds the length of the longest loan term that FC offers. On this day, FC will be able to retire the "Sell" functionality, safe in the knowledge that it could never again be of use.
Nice analysis... hope I am not following this thread by then!
|
|
ceejay
Posts: 971
Likes: 1,149
|
Post by ceejay on Jun 17, 2019 5:21:45 GMT
Just doing a little more work on this and forecasting sell dates based on the linear projection.
The linear equation works out at y=0.5348x-23244 (based on date x values in excel and y being the number of days to sell). The R2 value for the trendline is 0.9852. A value of 1 would mean a perfect linear correlation coefficient, so at 0.98 we can be confident that so far sell times have been increasing at a very steady and predictable rate.
That means going forward, and assuming the rate of growth in sell times remains constant, we can expect:
- 25/4/19 (Ref. indexwrangler), next on the list in criston post #1 to be sold on or around 26th June after 63 days
- Someone selling yesterday 14th June could expect to sell on or around 12th September after 89 days
A date for your diaries: 5 May 2028: "The Big Crunch".
This is the day when the time expected for a sale to execute exceeds the length of the longest loan term that FC offers.
On this day, FC will be able to retire the "Sell" functionality, safe in the knowledge that it could never again be of use.
Alternatively, expect an announcement today from FC of their new 90 day notice account, into which all funds have been automatically transferred, effective immediately.
|
|
Stonk
Stonking
Posts: 735
Likes: 658
|
Post by Stonk on Jun 17, 2019 20:34:01 GMT
I'm on day 59, and haven't sold yet ... are we about to enter the 60s?
|
|
|
Post by Mr Smith on Jun 18, 2019 7:19:36 GMT
I'm on day 59, and haven't sold yet ... are we about to enter the 60s? Blimey. Keep us posted. Will be good to know % sold too. This talk of a linear increase in selling times is nonsense if % sold is also dropping.
|
|
invester
P2P Blogger
Posts: 612
Likes: 618
|
Post by invester on Jun 18, 2019 7:25:44 GMT
My sale came through.
I also notice that checking my statement I made an error on the loan selling date (the last loan part purchase was 19/4/19, so I assume the sale was started on 20/4/19).
Also seems I had a decent result, 84% was sold.
The email starts 'congratulations' and it does seem apt.
|
|
|
Post by Mr Smith on Jun 18, 2019 7:42:13 GMT
My sale came through. I also notice that checking my statement I made an error on the loan selling date (the last loan part purchase was 19/4/19, so I assume the sale was started on 20/4/19). Also seems I had a decent result, 84% was sold. The email starts 'congratulations' and it does seem apt. Well done. % sales is excellent. No need to panic after all.
|
|
|
Post by rweb on Jun 18, 2019 8:43:58 GMT
I'm on day 59, and haven't sold yet ... are we about to enter the 60s? Are you positive you started the sell on the 19th? "Invester" realised that he/she actually put theirs up for sale on the 20th April and it's come through today, which is in-line with expectations based on a sell date of 20th. I'd have expected yours yesterday or possibly today.
This would be my guess for the next few:
Investor | Sell Instruction date | Forecast days to sell
| Forecast sell date
| Actual Sell date
| stonk | 20/4/2019
| 60
| 18/6/2019
| 18/6/2019
| invester | 20/4/2019
| 60
| 18/6/2019
| 18/6/2019
| indexwrangler
| 25/4/2019
| 63
| 26/6/2019
| 20/6/2019
| Ace
| 26/4/2019
| 63
| 28/6/2019
| 21/6/2019
| rweb
| 27/4/2019
| 64
| 29/6/2019
|
| djay
| 29/4/2019
| 65
| 02/7/2019
|
| krane
| 30/4/2019
| 65
| 04/7/2019
|
|
I'd expect the forecast to be +/- 1 due to rounding. The model for example suggests djay's sell will take 64.72 days, so that could be 02/7/2019 or 03/7/2019 according to the model.
|
|
Stonk
Stonking
Posts: 735
Likes: 658
|
Post by Stonk on Jun 18, 2019 9:27:07 GMT
I'm on day 59, and haven't sold yet ... are we about to enter the 60s? Are you positive you started the sell on the 19th?
I was pretty sure ... but I suppose it could have been a day later. I'm certain it wasn't later than that. It's just a single loan part which went from downgraded to sellable on 18 April, and I remember noticing quickly that this had happened, but rebuked myself for not spotting it the same day.
If my sale happens today, then, perhaps we'll say I issued the request on the 20th to keep the list in order.
|
|
Stonk
Stonking
Posts: 735
Likes: 658
|
Post by Stonk on Jun 18, 2019 17:50:54 GMT
Excitement! My sale just went through, at 6:30pm today. It was for a single loan part, and succeeded.
As mentioned, although I thought the sale was initiated on 19 April, if the evidence being gathered here suggests that it was acually 20 April then I'm prepared to believe I was mistaken.
Thanks for this thread.
|
|
corto
Member of DD Central
one-syllabistic
Posts: 851
Likes: 356
|
Post by corto on Jun 19, 2019 10:57:13 GMT
To come back to an earlier point: FC has turned my ISA into a standard account within three days after asking by email. No problems at all. I'd expect this to have a tax advantage with many of the left over loans late and potentially defaulting sooner or later. I'll let the ongoing sale as per page 1 go on for now, although 1 of the 3 loans is already back to "live"
|
|
criston
Member of DD Central
Posts: 1,204
Likes: 628
|
Post by criston on Jun 19, 2019 14:15:39 GMT
corto:
Apart from any future defaulted loans to set against gains, are you able to ascertain where defaulted loans since April 6th, up to the transfer date, end up.
Are you still able to see the ISA account, in which case do existing defaulted loans show there. If not, do they show in the Tax Statement of the new Classic account.
|
|