ResearchPro's Actual Performance

[ResearchPro]

Since some people would like to track my portfolio, here is the information to help them. I plan to update it monthly.

Monthly averaging method:
Accounting is done by the following scheme:

Principal:  beginning + additions - redemptions - charge-offs = ending
$PnL = interest - fees - charge-offs
Monthly return = $PnL / beginning
I calculate ROI as:
Average Monthly ROI = SUM($PnL)/SUM(beginning), where summation is performed by all complete months.
Annualized Average ROI = (1 + Average Monthly ROI)^12-1

Note: I charge off 1 month lates at 85%, 2 month lates and more at 100%.

Month			Principal			Interest	Fees	$PnL	Return	ROI-to-date
	Beginning	Additions	Redemptions	Charge-offs	Ending					(Annualized)
Aug-08	$0.00	$1800.00	$0.00	$0.00	$1800.00	-	-	-	-	-
Sep-08	$1800.00	$1150.00	$34.95	$0.00	$2915.05	$39.28	$1.28	$38.00	2.11%	28.49%
Oct-08	$2915.05	$400.00	$51.62	$0.00	$3263.43	$74.92	$2.35	$72.57	2.49%	32.07%
Nov-08	$3263.43	$0.00	$98.87	$0.00	$3164.56	$69.07	$2.31	$66.76	2.05%	30.19%
Dec-08	$3164.56	$0.00	$83.34	$41.10	$3040.13	$87.35	$2.84	$43.41	1.37%	26.54%
Jan-09	$3040.13	$0.00	$73.65	$47.36	$2919.12	$78.80	$2.51	$28.93	0.95%	23.29%
Feb-09	$2919.12	$0.00	$138.65	$127.71	$2652.76	$63.67	$2.01	$(66.04)	-2.26%	13.67%
Mar-09	$2652.76	$0.00	$73.13	$92.97	$2486.66	$68.70	$2.60	$(26.87)	-1.01%	9.95%
Apr-09	$2486.66	$0.00	$80.52	$60.55	$2345.59	$66.57	$2.61	$3.41	0.14%	8.99%
May-09	$2345.59	$0.00	$50.89	$57.76	$2236.94	$62.67	$2.76	$2.15	0.09%	8.22%
Jun-09	$2236.94	$0.00	$98.37	$90.73	$2047.84	$55.39	$1.77	$(37.11)	-1.66%	5.75%
Jul-09	$2047.84	$0.00	$138.89	$80.70	$1828.24	$55.38	$2.77	$(28.09)	-1.37%	4.11%
Aug-09	$1828.24	$0.00	$51.35	$1.22	$1775.67	$49.91	$4.74	$43.95	2.40%	5.66%
Sep-09	$1775.67	$0.00	$85.19	$44.27	$1646.21	$46.53	$1.72	$0.54	0.03%	5.36%
Oct-09	$1646.21	$0.00	$58.29	$66.56	$1521.36	$45.98	$1.81	$(22.39)	-1.36%	4.27%
Nov-09	$1521.36	$0.00	$40.32	$44.94	$1436.09	$33.86	$1.60	$(12.68)	-0.83%	3.65%
Dec-09	$1436.09	$0.00	$57.13	$68.29	$1310.67	$44.00	$1.77	$(26.06)	-1.81%	2.64%
Jan-10	$1310.67	$0.00	$50.51	$(19.71)	$1279.87	$35.34	$2.28	$52.76	4.03%	4.25%
Feb-10	$1279.87	$0.00	$40.71	$59.77	$1179.39	$24.69	$0.81	$(35.88)	-2.80%	2.98%
Mar-10	$1179.39	$0.00	$57.99	$(2.11)	$1123.52	$37.47	$3.11	$36.47	3.09%	4.00%
Apr-10	$1123.52	$0.00	$110.54	$(4.55)	$1017.53	$32.23	$3.06	$33.72	3.00%	4.90%
May-10	$1017.53	$0.00	$56.43	$96.72	$864.38	$16.42	$0.54	$(80.83)	-7.94%	2.45%
Jun-10	$864.38	$0.00	$79.59	$(16.36)	$801.14	$39.14	$2.00	$53.50	6.19%	3.90%

Total Beginning: $43,855.01
Total $PnL: $140.20
Average Monthly ROI: 0.32%
Average ROI Annualized: 3.90%

Alternative method, using Excel XIRR() function:
Money in = additions, assumed mid-month date;
Money out = principal redemption + interest - fees, assumed mid-month date.
Last assumed money out = latest month end principal balance, at this month's end.

$(1,800.00)	21-Aug-2008
$(1,150.00)	21-Sep-2008
$(400.00)	15-Oct-2008
$72.95	15-Sep-2008
$124.19	15-Oct-2008
$165.63	15-Nov-2008
$167.85	15-Dec-2008
$149.94	15-Jan-2009
$200.31	14-Feb-2009
$139.23	15-Mar-2009
$144.48	15-Apr-2009
$110.80	15-May-2009
$151.99	15-Jun-2009
$191.50	15-Jul-2009
$96.52	15-Aug-2009
$130.00	15-Sep-2009
$102.46	15-Oct-2009
$72.58	15-Nov-2009
$99.36	15-Dec-2009
$83.57	15-Jan-2010
$64.59	15-Feb-2010
$92.35	15-Mar-2010
$139.71	15-Apr-2010
$72.32	15-May-2010
$116.74	15-Jun-2010
$801.14	1-Jul-2010

XIRR = 4.03%

[go4reward]

Your return is great! Good for you! 

I am barely kept with making even.

[ResearchPro]

Your return is great! Good for you! 
I am barely kept with making even.

Thanks! I'm sure it will worsen with time though, 26% is just too good to be true. But I hope to end up with 10% or little more.

[go4reward]

Yes, I would be happy if I got 10% return overall.

[Investar]

Percentages are such fun. Using percentages, you can get 50% of the statistics to lie 100% of the time and 100% of the statistics to lie 50% of time. Where have I heard that? Anyway, I've been doing my best to understand the calcs you are doing and their subtleties. Other than math as a hobby (or obsession, your choice), I've also been trying to understand why you are doing this. Surprise! The ranking report I assembled here provides that answer. You are looking for the real answer.

Myself, I pretty much called LendingStats number close enough for me. I always had a hunch Eric's calc was too optimistic and the snapshot below seems to bear that out. But my goodness, look at the spread between LS and Eric's. And based on your work, the number I always ignored (Experian) looks like it's right in there. Only question now, which Experian number?

ResearchPro's ROI
based on the most recent snapshot of from all sources

23.92% - LendingStats estimating method
25.24% - ResearchPro alternate Excel method
25.93% - Experian ROI reported by LendingStats
26.54% - ResearchPro preferred averaging method
26.93% - Experian ROI reported by Erics
28.32% - Eric CC calculation method
- - - - - - -
26.15% - ResearchPro's actual ROI using Investar secret method

Okay, so my secret is I averaged all six of the stats above. Can't go wrong averaging the averages, can ya? Well Mr. Pro, I'm having fun watching you have fun and I may yet learn something. I think I already learned why you chose "ResearchPro" as your user name. A bit of housekeeping: you show "Total Beginning: $11,143.04" -- shouldn't that be 1k rather than 11k?

[go4reward]

My current ROI is about 0%. It stinks.

[ResearchPro]

Life itself is supposed to be fun... Or suffering... - depends on your personal beliefs.
Also, I guessed that you averaged it before I red your explanation
I think that the Experian ROI is close by accident, it won't be close in time (I don't believe I can make 20%).

I think I already learned why you chose "ResearchPro" as your user name.

I'm intrigued: I don't know the right answer on this one myself

A bit of housekeeping: you show "Total Beginning: $11,143.04" -- shouldn't that be 1k rather than 11k?

No, 11K is correct. The idea is to line-up all 1-month periods like I pass through them simultaneously. So the beginning would be combined beginnings of all of them and PnL would be a combined PnL of all of them.
ROI shouldn't be affected whether all 1-month periods are consequent of all at the same time.

[ResearchPro]

My current ROI is about 0%. It stinks.

LS shows you at 3.5%, Eric's shows 11.9%. Both are far from 0.

[go4reward]

LS shows you at 3.5%, Eric's shows 11.9%. Both are far from 0.

Yes and no. As you have already aware of the reporting mechanism by other sites, all (<15 days) loans are not reported yet. If they count those in, I am sure it will be close to it or even less.

My performance maybe worse in the SCP group. I was daunt aggressive at the time. I have improved my loan stat in the later months think to helpful members forum as well as being more conservative.

Ken

[Investar]

A bit of housekeeping: you show "Total Beginning: $11,143.04" -- shouldn't that be 1k rather than 11k?

No, 11K is correct. The idea is to line-up all 1-month periods like I pass through them simultaneously. So the beginning would be combined beginnings of all of them and PnL would be a combined PnL of all of them.
ROI shouldn't be affected whether all 1-month periods are consequent of all at the same time.

I'm not quite with you (haven't actually done the math exercise longhand) but it all seems a bit convoluted to me. I'd kinda like to post your numbers to a temporary asset in my database and see what my math comes up with. My numbers are only accurate for full quarters of activity so I'd need your Jan 08 data to complete Qtr 4.

The math package I built into my custom portfolio database addresses ROI skew on assets with a compounding or fluctuating cost basis across the 12 months (such as CDs and the Prosper asset). It adds up the income and divides by a 'phantom' cost basis that starts with dollars invested at the beginning of the year but revalues dollars added/removed during the year to account for the period of time they were "working" or "not working" in the asset.

Mechanically, it's a little coarse. It assumes all new dollars in the first quarter appeared on a singular day and worked .8333 of the year, dollars appearing the second quarter worked .5833 of the year, and so on. It gets Cds exactly right and gawd darn close on most everything else. It has big trouble with Money Market slush funds due to the high volume of dollar movement that occurred other than on day .8333 or day .5833 during the year.

[ResearchPro]

Yes and no. As you have already aware of the reporting mechanism by other sites, all (<15 days) loans are not reported yet. If they count those in, I am sure it will be close to it or even less.

Most of my <15 days lates returned to current, so I don't think you can treat those as losses yet. Also, LS is predicting the future somewhat, so if your lates will develop kind of at the same speed as before, you are still at 3-4%.

[ResearchPro]

The math package I built into my custom portfolio database addresses ROI skew on assets with a compounding or fluctuating cost basis across the 12 months

Wow...

It adds up the income and divides by a 'phantom' cost basis that starts with dollars invested at the beginning of the year but revalues dollars added/removed during the year to account for the period of time they were "working" or "not working" in the asset.

Sounds somewhat like XIRR in Excel.
Anyway, we'll see after Jan is complete.

[Investar]

Okay, I plugged ResearchPro's numbers into my data base.
It says his annualized APY was 22.48%.

I know something the computer doesn't know -- he was only at this half a year. My math adjusts valuation of dollars at work to compensate for the length of time they were working during the year, so I did a basis adjustment to compensate and came up with 26.22%. This is all more helpful to me than it is to you. Since I trust RP's math to be pretty concise, it confirms my formulas are good 'nuff for me. That said and assuming a static portfolio, it's unlikely the 26.22% number is real -- too high. As for XIRR, I don't think it has a leg up on Research's method. Both his methods are more accurate month-on-month than mine. Still, my work suggests 12 months of actual data will show something closer to 22%.

above edited for clarity and to correct results (after PM to ResearchPro, I fixed a data entry error I had) 

An accounting question cropped up -- Reporting income: From the box on top of page two of Prosper's monthly statement, "Loan Summary" I post the line "Payments: Interest" (plus fees paid if any) as my income. The service fee I paid Prosper is already removed from that number? This is what they actually paid me?

[sosocratic]

Yes and no. As you have already aware of the reporting mechanism by other sites, all (<15 days) loans are not reported yet. If they count those in, I am sure it will be close to it or even less.

Most of my <15 days lates returned to current, so I don't think you can treat those as losses yet. Also, LS is predicting the future somewhat, so if your lates will develop kind of at the same speed as before, you are still at 3-4%.

I agree - most of my <15 days late return to good. One of mine went 1 month and then returned to good (out of 3-4 looking bad) so we'll see how that hold up. But I also would not do anything with the 15 days late until they go one month late. Then it is probably reasonable to take half of it off or so and then at 2 months consider it probably gone.

[ResearchPro]

An accounting question cropped up -- Reporting income: From the box on top of page two of Prosper's monthly statement, "Loan Summary" I post the line "Payments: Interest" (plus fees paid if any) as my income. The service fee I paid Prosper is already removed from that number? This is what they actually paid me?

Yes, I believe so.