Harmoney

[myles]

Be careful simply adding total tax paid back to determine XIRR - the result is not the same as having not paid tax - time value of money and all that...depends on what you really want the final figure to be...

[Wsp]

https://www.cio.co.nz/article/632369...s-credit-risk/

[leesal]

https://www.cio.co.nz/article/632369...s-credit-risk/

What the machine will learn when credit conditions change

[leesal]

well when I stated I have 166 E's, that was when I had the max invested, ie my outstanding principle was the highest.

I have now had 207 E's over the course of my investing in harmoney. Some of those are still active, some have been paid off already, and about 10.14% have been written off (I downloaded the csv as per yesterday E writes offs increased from 8.8% - the 9% i quoted, to 10.14%).

So 10.14% of all the E's I have invested in (not current, have invested) have been written off.

Of all the loans I have that have been written off, E's make up 48.84%, despite being 11.81% of what I have invested in.

How much interest have you earned off E's? How does this compare to $ value of E Charge offs?

On 1.0 as long as you're 50% ($ Charge / $ Int ) or less then you are doing fine.

[myles]

Good to see that the early claimed 50% error has waned to a more realistic value and that interest is forming part of the discussion

The biggest impact on return, for those in the higher risk grades, comes from defaults that are somewhat unpredictable. Using a more accurate calculation of returns early is pointless (in my opinion) when you start to factor in the 'hazard curve'. With only the data Harmoney provide (which we know is likely to be a bit off) a guesstimate of actual return is the best you can expect - and this is going to be influenced by what's happening out in the real world... Comparing RAR or XIRR (corrected or not), of one set of loans to another that are of a different age, is never going to be a fair comparison... Comparing two mature sets of loans would be 'more fair'...

I'd be interested, leesal, on how you will factor this into your wager - the full impact of a default isn't known until close to 1 year after it actually starts? Since those in your wager have started at different times, any comparison would be somewhat distorted as default rates vary over time and are influenced by the real world? The only accurate comparison would come when Outstanding Principal reaches zero and XIRR (or similar) is used on all in's and out's, but starting at different times has an impact even if the term is set the same?

[leesal]

Good to see that the early claimed 50% error has waned to a more realistic value and that interest is forming part of the discussion

The biggest impact on return, for those in the higher risk grades, comes from defaults that are somewhat unpredictable. Using a more accurate calculation of returns early is pointless (in my opinion) when you start to factor in the 'hazard curve'. With only the data Harmoney provide (which we know is likely to be a bit off) a guesstimate of actual return is the best you can expect - and this is going to be influenced by what's happening out in the real world... Comparing RAR or XIRR (corrected or not), of one set of loans to another that are of a different age, is never going to be a fair comparison... Comparing two mature sets of loans would be 'more fair'...

I'd be interested, leesal, on how you will factor this into your wager - the full impact of a default isn't known until close to 1 year after it actually starts? Since those in your wager have started at different times, any comparison would be somewhat distorted as default rates vary over time and are influenced by the real world? The only accurate comparison would come when Outstanding Principal reaches zero and XIRR (or similar) is used on all in's and out's, but starting at different times has an impact even if the term is set the same?

For mature portfolio's yes

Its a bit of fun. No-one went too hard out early, besides those first guys think they have the advantage on platform 1. But we do have groundrules over arrears, everything over 60 days gets charged etc.

Regarding survival. Longitudinally 2nd year is on average 30% (iirc) worse then 1st, so I'll extrapolate my position. 14% return now can get screwed in many ways. So certainly not going on a spending spree anytime soon!

Also am dollar cost averaging ... After 18 months, portfolio will be 9 months through the hazard giving higher reported returns. Same thing if you invest mainly in 36 month loans. Some comparability issues there. But take your point, earlier is going to be higher Stn Dev.

[myles]

For mature portfolio's yes

Not at all, can happen from day one and does, especially if you buy into the higher risk loans...this has a much greater impact on a younger portfolio's than mature ones? Losses early that blowout over time.

These are also variable across the year, appears to be more significant than most think.

Sounds like you haven't yet factored in the big ticket items that really affect returns.

[leesal]

At the risk of being the odd couple, rewind and start again....

Good to see that the early claimed 50% error has waned to a more realistic value and that interest is forming part of the discussion

For mature portfolio's yes **(my return on PP = 50%) We're talking about XIRR here and no-one wants that so lets move on **

I'd be interested, leesal, on how you will factor this into your wager - the full impact of a default isn't known until close to 1 year after it actually starts? Since those in your wager have started at different times, any comparison would be somewhat distorted as default rates vary over time and are influenced by the real world? The only accurate comparison would come when Outstanding Principal reaches zero and XIRR (or similar) is used on all in's and out's, but starting at different times has an impact even if the term is set the same?

Its a bit of fun. No-one went too hard out early, besides those first guys (**ie in our wager**) think they have the advantage on platform 1. But we do have groundrules over arrears, everything over 60 days gets charged etc.

**(ie now we're on the whole relatively cool with everything)**

Good to see that the early claimed 50% error has waned to a more realistic value and that interest is forming part of the discussion

The biggest impact on return, for those in the higher risk grades, comes from defaults that are somewhat unpredictable. Using a more accurate calculation of returns early is pointless (in my opinion) when you start to factor in the 'hazard curve'. With only the data Harmoney provide (which we know is likely to be a bit off) a guesstimate of actual return is the best you can expect - and this is going to be influenced by what's happening out in the real world... Comparing RAR or XIRR (corrected or not), of one set of loans to another that are of a different age, is never going to be a fair comparison... Comparing two mature sets of loans would be 'more fair'...

Regarding survival (**hazard**). Longitudinally 2nd year is on average 30% (iirc) worse then 1st, so I'll extrapolate my position (**ie take my return after one year, and apply a factor against the defaults, to determine what the return would be. Years 2 and 3 are on average much worse for defaults**). 14% return now can get screwed in many ways. So certainly not going on a spending spree anytime soon!

Also am dollar cost averaging ... After 18 months, portfolio will be 9 months through the hazard giving higher reported returns. Same thing if you invest mainly in 36 month loans. Some comparability issues there. But take your point, earlier is going to be higher Stn Dev.

(** agreeing with you here, that higher risk grades increases variability; length invested will likely suppress return via the hazard curve; pointing out different investment grow early vs grow late even if the same age would be at different stages along the hazard.... while a portfolio of 3 year loans vs portfolio of 5 year loans at 18 months likewise would on average produced very different returns).

can happen from day one and does, especially if you buy into the higher risk loans...this has a much greater impact on a younger portfolio's than mature ones? Losses early that blowout over time.

These are also variable across the year, appears to be more significant than most think.

we agree here, earlier portfolio's will be higher standard deviation. As are DEFs. To reiterate my greatest fear if a recession hits some lower grade portfolio may not produce returns.

[leesal]

This is how tracking charge offs. There are 5 loans in arrears (1 to 30 days), so they get a 20% probability load against them.

	units paid	unit arrears	chargeoff %	harm est chg%
A	27	0	0.0%	0.1%
B	220	0	0.0%	0.2%
C	126	0	0.0%	0.6%
D	134	1	1.7%	1.7%
E	71	1	3.4%	4.4%
F	15	3	47.2%	7.6%
Tot	593	5	2.0%	1.3%

Really early days. Though not good signs on F grade.

[permutation]

This is how tracking charge offs. There are 5 loans in arrears (1 to 30 days), so they get a 20% probability load against them.

	units paid	unit arrears	chargeoff %	harm est chg%
A	27	0	0.0%	0.1%
B	220	0	0.0%	0.2%
C	126	0	0.0%	0.6%
D	134	1	1.7%	1.7%
E	71	1	3.4%	4.4%
F	15	3	47.2%	7.6%
Tot	593	5	2.0%	1.3%

Really early days. Though not good signs on F grade.

Here are some long term stats from my 34 months in the Harmoney "Universe";

• E Grades taken 71- Charged off 11 = 15.49%
• F Grades taken 30- Charged off 5 = 16.67%

• Total E&F Grades to date 101- Charged off 16= 15.84%
• Total A>D Grades to date 1560 - Charged off 14= 0.8974%
• Note: This default rate is across 34 months. Harmoney's forecast defaults are p.a.

Attachment 9423