Print

[ladeeda]

It would be nice if she shared how the calculations were done along with what assumptions were built into the calculations.

I mentioned this in another thread, but here's the calculation description as described in the email:

"Returns include all payments received net of principal repayment, credit losses, and servicing costs for Notes not bought or sold on Folio. Returns are divided by the average daily amount of principal outstanding to get a simple rate of return. This rate is annualized by dividing by the dollar weighted average Note age of your portfolio in days and multiplying by 365."

So...

R_total = (Payments Excl Prin - Principal Losses) / (Average Daily Balance)
R_annualized = (R_total / Average Note Age) * 365

The calculation is relatively straight-forward, but if you have any other questions, feel free to post them here.

[onthefence]

if you have any other questions, feel free to post them here.

I do have one more question, but I know you can't provide the information directly to me.  So we will have to do this the round-about way.
http://www.prospers.org/forum/will_researchpro_leave_or_admit_he_was_wrong-t11350.0.html;msg397330#msg397330

Edit: Doing some back of the excel spreadsheet calcuations on what ResearcPro has available for data, it seems like Prosper's quick & dirty calculation may be within the ballpark of ResearchPro's XIRR.

[Fred93]

This rate is annualized by dividing by the dollar weighted average Note age of your portfolio in days and multiplying by 365.

How exactly is this "average note age" computed?  How are loans that are completed counted in the "age" calculation?

[ladeeda]

How exactly is this "average note age" computed?  How are loans that are completed counted in the "age" calculation?

As I understand it, we only include observations where a loan is not charged off, revoked, or paid in full, and age is just the number of days between the observation date and the origination date.

[Fred93]

As I understand it, we only include observations where a loan is not charged off, revoked, or paid in full, and age is just the number of days between the observation date and the origination date.

Wow.  Only active loans eh?  Well do you then also only use active loans in the rest of the calculation, ie only count the interest from active loans?

This is not anything like an ROI calculation.  This is nonsense.

[ira01]

How exactly is this "average note age" computed?  How are loans that are completed counted in the "age" calculation?

As I understand it, we only include observations where a loan is not charged off, revoked, or paid in full, and age is just the number of days between the observation date and the origination date.

That makes no sense, as Fred points out.  What good is an ROI that doesn't include paid and charged-off loans?  Are you sure about what you said here?

[ladeeda]

How exactly is this "average note age" computed?  How are loans that are completed counted in the "age" calculation?

As I understand it, we only include observations where a loan is not charged off, revoked, or paid in full, and age is just the number of days between the observation date and the origination date.

Fred93 asked me about how completed loans are handled in the age calculation - that's what my response was referring to.

Obviously we include principal losses in the equation, as I said earlier:

I mentioned this in another thread, but here's the calculation description as described in the email:

"Returns include all payments received net of principal repayment, credit losses, and servicing costs for Notes not bought or sold on Folio. Returns are divided by the average daily amount of principal outstanding to get a simple rate of return. This rate is annualized by dividing by the dollar weighted average Note age of your portfolio in days and multiplying by 365."

So...

R_total = (Payments Excl Prin - Principal Losses) / (Average Daily Balance)
R_annualized = (R_total / Average Note Age) * 365

(Emphasis mine)

If you want to have a dialogue, let's do that, but throwing around remarks like "This is not anything like an ROI calculation.  This is nonsense." is not helpful. Besides, how does the idea that it's nonsense square with this?

Prosper's number may well be correct, or at least pretty close.

[ira01]

How exactly is this "average note age" computed?  How are loans that are completed counted in the "age" calculation?

As I understand it, we only include observations where a loan is not charged off, revoked, or paid in full, and age is just the number of days between the observation date and the origination date.

Fred93 asked me about how completed loans are handled in the age calculation - that's what my response was referring to.

Obviously we include principal losses in the equation, as I said earlier:

I mentioned this in another thread, but here's the calculation description as described in the email:

"Returns include all payments received net of principal repayment, credit losses, and servicing costs for Notes not bought or sold on Folio. Returns are divided by the average daily amount of principal outstanding to get a simple rate of return. This rate is annualized by dividing by the dollar weighted average Note age of your portfolio in days and multiplying by 365."

So...

R_total = (Payments Excl Prin - Principal Losses) / (Average Daily Balance)
R_annualized = (R_total / Average Note Age) * 365

(Emphasis mine)

I don't think you understand the issue.  I don't mean that as a slam, I mean that as a factual statement.  The age calculation is part of the ROI calculation.  Your explanation -- that the age calculation excludes completed loans -- makes no sense when that number is then used in the ROI calculation.  If, in fact, Prosper was really including all the losses/gains from completed loans (as it should), but using an age calculation that only considered active loans (as you state), then the resulting "ROI" number would (as Fred stated) be meaningless. 

You should also note (since it was before your time, but not before many of our times) that Prosper has a history of doing such meaningless number crunching in ways that falsely make it look better -- Prosper used to calculate (and Larsen widely disseminated to any news media that would listen and uncritically regurgitate) a "default rate" that it calculated by dividing "defaults" by total loans -- although Prosper erroneously (some would say fraudulently) understated that number by constantly making two "errors" (which were repeatedly pointed out to Prosper by many of us here, not that Prosper cared enough about accuracy to stop what it was doing).  First, Prosper erroneously decreased the numerator by only counting loans that it chose to affirmatively "default" -- i.e., sell to a JDB (back when Prosper actually used to have JDB sales as required by the LRA).  At the time, there were many loans that had been in Late 4-month+ status for a LONG time (some loans were close to a YEAR delinquent), yet Prosper didn't include those loans as "defaults."  Second, Prosper erroneously increased the denominator by including ALL loans ever originated, EVEN those loans that because they were less than five months old, could not have possibly "defaulted" yet even if they never made a SINGLE payment (a loan can't be 4 months delinquent until it is at least 5 months old).  By understating the numerator and overstating the denominator, Prosper massively understated its "default rate."  Your explanation makes it appear that Prosper is up to its old tricks (although, in this case, I suspect that you are wrong).

If you want to have a dialogue, let's do that, but throwing around remarks like "This is not anything like an ROI calculation.  This is nonsense." is not helpful. Besides, how does the idea that it's nonsense square with this?

Prosper's number may well be correct, or at least pretty close.

I think the most likely way to square the two ideas is that Prosper's ROI calculation may be pretty decent, but that your explanation of it is erroneous. 

As an aside, it seems to me that what Prosper really ought to be doing is calculating everyone's ROI with XIRR based on capital transfers in and out of Prosper.  Clearly Prosper could do that, as it (unlike the third-party sites) has access to the cash transfer history of each lender, and that would be the most accurate ROI calculation. 

[Fred93]

If you want to have a dialogue, let's do that, but throwing around remarks like "This is not anything like an ROI calculation.  This is nonsense." is not helpful. Besides, how does the idea that it's nonsense square with this?

Well I used the word nonsense, because what Prosper is giving us here looks like nonsense to me.

Its certainly not based on any textbook definition of ROI.  It doesn't seem to be based on industry standards for return calculations.  Prosper made it up.  And what is most difficult about this discussion is that it is so hard to find out what the calculation actually is.  You don't provide proper equations, and I feel like I have to play 20 questions to find out anything. 

The first question I asked got an answer that raised immediate red flags.  You said you don't count finished loans in the calculation of average age.  The whole idea of using an average loan age to normalize return is questionable (and I'm being polite in my choice of words here).  So when my first question about what the heck your age calculation is exactly gets such a goofy answer, it simply reinforces my impression that this is not a meaningful calculation.  But it is still possible that you're doing something somewhat reasonable and just not explaining it well, so I asked a 2nd question.  You didn' respond to my 2nd question at all.

If you can provide details which show how an age which does not contain finished loans is used to normalize in a meaningful way a number which is calculated from all loans (including those which are finished), then the dialog can continue. 

There ought to be some transparency.

My position, which I have held from the very beginning, ie 2006, is that the only proper return calculation is an IRR.  This is easily calculated.  Prosper has all the data required.

I do understand that there may be some hesitancy to use IRR based on the idea that people won't understand what it is, but then people don't understand what the thing you have already calculated is either.  Most people don't understand the mathematics of loans.  It is unfortunately a bit more complicated than most people's understanding.  (I will note that there is one place where the government forces you to do an IRR calculation.  That's inside the calculation of what the government calls APR.  You do it that one place, because the government has forced you to.)

If you read my early writings about prosper, you will find that from the very beginning there has been a problem with statements prosper has made about investors returns.  There has been plenty of time to get it right, so do it.

[ladeeda]

How exactly is this "average note age" computed?  How are loans that are completed counted in the "age" calculation?

As I understand it, we only include observations where a loan is not charged off, revoked, or paid in full, and age is just the number of days between the observation date and the origination date.

You said you don't count finished loans in the calculation of average age.

If you can provide details which show how an age which does not contain finished loans is used to normalize in a meaningful way a number which is calculated from all loans (including those which are finished), then the dialog can continue. 

I didn't say that. I said that we exclude those observations from the age calculation. After pay-off the loan is (by definition) "completed". If we attributed increasing age to the loan after it was completed, then, holding all else constant, the annualized return for the portfolio would fall as time passed, as no further interest would be paid on it.

Additionally, I would think that the use of IRR as I've seen it described here at .org would penalize portfolios where people let cash sit idly in their accounts. I don't really understand the rationale there - we're trying to report the returns you can earn by investing your money in a portfolio of loans, not the value of letting it sit on the platform.

All that said, I'm neither the person responsible for determining which calculation to use, nor am I the person responsible for its calculation. And to Ira's point, I may be wrong on the details or just not communicating them very well. But, given where things stand, I'm not sure what more I can be doing here.

[Xenon481]

Additionally, I would think that the use of IRR as I've seen it described here at .org would penalize portfolios where people let cash sit idly in their accounts. I don't really understand the rationale there - we're trying to report the returns you can earn by investing your money in a portfolio of loans, not the value of letting it sit on the platform.

No, an investor is investing in Prosper, not a loan. Even when purchasing a Note, it is an investment in Prosper.

If there were 0 lead time to take the money in your Prosper account to an active Prosper Note then I could agree that it would be negligible, but with significant amounts of money, there is no way to keep from having a significant amount of money tied up idle for significant periods of time.

The only true rate of return is the rate of return for the platform as a whole, not just the notes.

[Fred93]

If you can provide details which show how an age which does not contain finished loans is used to normalize in a meaningful way a number which is calculated from all loans (including those which are finished), then the dialog can continue.  

I didn't say that. I said that we exclude those observations from the age calculation.

I'm sorry, but I apparently don't understand your use of the word "observation" here.  Perhaps you could explain the calculation you do again, explaining what "observation" means in the calculation.  I really would like to understand what calculation you are doing.

Additionally, I would think that the use of IRR as I've seen it described here at .org would penalize portfolios where people let cash sit idly in their accounts. I don't really understand the rationale there - we're trying to report the returns you can earn by investing your money in a portfolio of loans, not the value of letting it sit on the platform.

Well that's a deep subject.  You can calculate an IRR either including the effects of idle cash, or not including the effects of idle cash.  I'd be happy with you taking either approach.

I specifically chose to include the effects of idle cash.  I would be happy to explain the reason.  Back when I was investing heavily in Prosper, I found that I had a very large cash balance at all times.  The reason was structural.  (In other words, it was not my choice, but came about  because of the way Prosper works.)  First there's the 4 business day delay associated with ACH transfers.  Then cash has to wait until suitable loans appear.  Some of it I would bid on the 1st day, but some waited 1 day, some waited 2 days, etc.  I was doing transfers into my prosper account every few days.  Lots and lots of transfers.  After I bid, my cash sat idle while waiting for the bidding to finish.  This could be up to 10 days.  I tried to bid mostly on loans where the bidding was nearly completed, but they weren't always available in sufficient number.  After the bidding ended, there was a period of several days while the loan was finalized.  In those days this was sometimes over a week.  When you add this stuff up, it is substantial.  Therefore, I chose to calculate my return including the effects of idle cash.  

However, as I said before, there is nothing inherent in the IRR concept that has anything to do with idle cash.  You just draw a line somewhere, and track the cash flows across that line.  I drew the line between me and my prosper account.  You could draw the line between the prosper account and the notes, thus calculating based on investment in and payments from the notes.

The basic thing that IRR does correctly is account for the time value of money.

[onthefence]

I didn't say that. I said that we exclude those observations from the age calculation. After pay-off the loan is (by definition) "completed". If we attributed increasing age to the loan after it was completed, then, holding all else constant, the annualized return for the portfolio would fall as time passed, as no further interest would be paid on it.

I think there is some confusion as to how exactly the data is measured.  A loan that pays until completion would count as approximately 1095 days.

Would it be fair to say that a loan that is get's charged off after 270 days would be factored into the average at 270 days, or is it no longer included in the equation since it it charged off?  (I think there is a communication issue in understanding this so far.)

I think part of the concern with Prosper's calculation of ROI is it appears to be very unique & people suspect it may be prone to failure when dealing with extremes & end points.  People may have been more satisfied if Prosper had used more familiar measurement of cash in/cash out.  Now it's possible that your equation accounts for this, but if it does, it isn't immediately obvious without crunching the equations.  Something that isn't quite as easy to do with XIRR.

Additionally, I would think that the use of IRR as I've seen it described here at .org would penalize portfolios where people let cash sit idly in their accounts. I don't really understand the rationale there - we're trying to report the returns you can earn by investing your money in a portfolio of loans, not the value of letting it sit on the platform.

And not including it would give the ROI a useless bump that doesn't include idle time of money that sits waiting for:
* the bidding process to end.
* wasted time for loans that fail to fund
* wasted time for loans that fail to originate
* time to takes for money to transfer in
* time it takes for money to transfer out
* time wasted waiting for a loan worth bidding on comes up
* Funds sitting in the account that are too small to bid out.

Now we come to the area of personal responsibility.  People who don't bother to bid all that often, only felt like bidding once a week, or twice a month since funds didn't come in that often, or those that grew disenchanted with the performance & didn't feel like bidding as often.  This still affects their performance on Prosper.  Excluding this by no means magically makes their portfolio magically enough to beat out a 7% bond fund they may be eying as an alternative.  Excluding that is like excluding loans to HRs or loans that have current delinquencies.  Clearly these are areas where the lender has been a complete idiot & it's their fault for making such stupid decisions.  But removing them in no way magically improves their real annual ROI.

Now this isn't all bad for Prosper.  Whole calculations can be made based on smart bidding strategies such as Experiean AAs only with no current delinquencies & no more then 2 inquiries.   But Prosper needs to calculate an annual ROI that we can trust is accurate.  It's much easier for Prosper to be able to do so since they have access to all of the information.

All that said, I'm neither the person responsible for determining which calculation to use, nor am I the person responsible for its calculation. And to Ira's point, I may be wrong on the details or just not communicating them very well. But, given where things stand, I'm not sure what more I can be doing here.

It might be of value to get the description of what is included & not included from the calculation from the person who implements it him or herself.  The other end is to provide feedback to the person who determined which calculation to use.  Of course doing so now would be very premature since I'm not sure we accurately understand what is included in the calculation, nor have we spent enough time evaluating the calculation to see if it fails in certain situations.

[Urbi_et_Orbi]

And not including it would give the ROI a useless bump that doesn't include idle time of money that sits waiting for:
* the bidding process to end.
* wasted time for loans that fail to fund
* wasted time for loans that fail to originate
* time to takes for money to transfer in
* time it takes for money to transfer out
* time wasted waiting for a loan worth bidding on comes up
* Funds sitting in the account that are too small to bid out.

True.  Also, don't forget the time the monthly incoming payments sit in transfer before the funds are even available as idle, biddable money in your account.

[JammingJAY]

You guys better chill out or prosper shira will ban you.