Calculate your percentile rank vs other LC accounts

[Fred]

I found what was causing this.  Our tool uses a static image for the background chart instead of dynamically creating the chart using all the data points from the JSON file.  Our percentile tool use the data from the JSON so it is accurate, but the chart we have can be out of date.

Your tool is faster compared to LC's, which I think is due to the static chart image. 

However, I would still like to see the chart image to reflect the latest data from LC.  Therefore, perhaps generating the chart image once daily would be enough.  Or weekly?  Not sure how often LC updates the JSON data.

[rawraw]

Your Percentile:
vs All Accounts: 87
vs Similar Age Accounts: 94

Pretty cool.

[SeanMCA]

I've made a change to allow older ages but give a caveat that "vs similar age" is not accurate because we have no data on that.  We are just comparing with with other 30 month old accounts.

Wait, does similar age here refer to the length of time accounts with lending club have been open? I thought it referred to the weighted average age of the portfolio.

[turing]

Your tool is faster compared to LC's, which I think is due to the static chart image. 

However, I would still like to see the chart image to reflect the latest data from LC.  Therefore, perhaps generating the chart image once daily would be enough.  Or weekly?  Not sure how often LC updates the JSON data.

Glad you noticed the speed.  We cache the entire JSON file for 24 hours to speed up processing so we don't even need to wait for LC to give us the data.

If we do generate the chart dynamically we would want to keep the speed so we would probably do it daily too.  Having 24,000 points in html/css might slow things down even if the JSON doesn't need to be processed.  I will have to play with it to find the best combination of speed and accuracy.

Feedback is good so far so it is probably worth the time to improve it.

[thezfunk]

I have way more losses (-16.91% on 1126 notes) than gains on Folio, so this feel good chart is fun for me even though I know it isn't for real.

Your Percentile:
vs All Accounts: 99
vs Similar Age Accounts: 99

Age:  16.6 mos.
# of notes (not counting the 1126 solds):  6315 (4679 issued and current)

I am also in the 99 percentile category and like you have a loss running on Folio where i dump the notes that go bad or are predicted to go bad of -8.5 or so.  It's getting better as I am getting more proactive about putting a note on Folio instead of waiting for it to get really bad.

[Lovinglifestyle]

I am also in the 99 percentile category and like you have a loss running on Folio where i dump the notes that go bad or are predicted to go bad of -8.5 or so.  It's getting better as I am getting more proactive about putting a note on Folio instead of waiting for it to get really bad.

PeerCube has my Folio account at -8.76 (half as bad as LC's NAR).  I like knowing the numbers--it helps to keep me motivated toward early  management.  I don't use an automatic method--yet.  Thanks for sharing!

[rawraw]

My Folio returns are 8% NAR based on FICO drops.

[turing]

Maybe LC will give a handy JSON comparison tool for Folio and overall performance.  When they do we will be sure to add a percentile tool for that too.

[lascott]

Re: Our model has a percentile score of 98 vs all accounts. And 92 vs accounts with the same age.

You may want to keep those those 2 number for each x.y

Weighted Average Age of Portfolio:   x.y mos

That way you can graph the change over the weeks.

[turing]

Re: Our model has a percentile score of 98 vs all accounts. And 92 vs accounts with the same age.

You may want to keep those those 2 number for each x.y

Weighted Average Age of Portfolio:   x.y mos

That way you can graph the change over the weeks.

That is a very good suggestion.  I will do that for sure.  Spreadsheet created now.

[turing]

I've implemented some of your suggestions.

1) Percentile vs Similar Age and Similar WAIR


2) Chart dynamically generated (no longer static image)

3) See similar accounts (age & wair) on the chart

[lascott]

Appears to be working. Thanks.