1 card from 89% to 48% --> +26 points!

[Anonymous]

---

@Anonymous wrote:

So Cassie, is your theory that SJ's 89.xx% is being rounded up [to 90%] in terms of what the Fico algorithm is considering, but at the same time it's being truncated to 89% on the Experian front end and thus because SJ is seeing his score move he is then believing that 89% is in fact the meaningful number being considered, not 90%?  If so, you're theory is that Experian fluff exists on their front end just like many other CMS.

Your thoughts on that, SJ?

I don't know that I really believe that the years of rounding testing done on here has all been false.  I mean there have been tons, literally hundreds of tests right around the known threshold points (mostly the main one for aggregate) where people have gone to the hundreths percentage wise going across and back and all cases that I've heard of near as I can remember have verified upward rounding on percentages that contain decimals past the whole number.  It just seems impossible to me that all of those tests were somehow false positives.  Am I missing something here?

---

It's very possible that many of those tests were inaccurate due to using exact balances vs. the truncated balances. All balances should be truncated before calculating utilization. I didn't know this before the AMA with people from FICO. I also didn't know that <8.99 was wrong either. < 9.5 is correct when talking about avoiding 10% utilization.

Was that 89.573 calculated using the truncated balance on that card? Depending on the limit, incorrectly including up to an additional 99¢ could have put that number at 89.5+ for the manual calculation, but not at Experian because they would never see a fractional balance amount. So naturally Experian would show 89%.

[SouthJamaica]

---

@Anonymous wrote:

---

....

Are you absolutely sure it's 89.573?

....

I'm never absolutely sure of anything having to do with FICO

The credit limit was 15000.00

The rounded balance was 13436.00

The actual balance was 13436.20

[SouthJamaica]

---

@Anonymous wrote:

So Cassie, is your theory that SJ's 89.xx% is being rounded up [to 90%] in terms of what the Fico algorithm is considering, but at the same time it's being truncated to 89% on the Experian front end and thus because SJ is seeing his score move he is then believing that 89% is in fact the meaningful number being considered, not 90%?  If so, you're theory is that Experian fluff exists on their front end just like many other CMS.

Your thoughts on that, SJ?

I don't know that I really believe that the years of rounding testing done on here has all been false.  I mean there have been tons, literally hundreds of tests right around the known threshold points (mostly the main one for aggregate) where people have gone to the hundreths percentage wise going across and back and all cases that I've heard of near as I can remember have verified upward rounding on percentages that contain decimals past the whole number.  It just seems impossible to me that all of those tests were somehow false positives.  Am I missing something here?

---

I think I've been following experian.com on a daily basis for at least 2 years now, and I'm pretty convinced that the rounded percentages are the ones that move my EX FICO scores.

[Anonymous]

But how can you be sure that whatever is being rounded/truncated with EX is what is being considered by the Fico algorithm?  I don't know that there's any real way to verify that with 100% certainty.

EX may show you X, but the algorithm sees at as X+1.  If X+1 is the Fico threshold, wouldn't X then move your scores and thus present you with a false positive as to the threshold point?

[SouthJamaica]

EX is also supplying me with the EX FICO scores

[Anonymous]

---

@Anonymous wrote:

But how can you be sure that whatever is being rounded/truncated with EX is what is being considered by the Fico algorithm?  I don't know that there's any real way to verify that with 100% certainty.

EX may show you X, but the algorithm sees at as X+1.  If X+1 is the Fico threshold, wouldn't X then move your scores and thus present you with a false positive as to the threshold point?

---

@Anonymous I agree with your analysis of how Experian's FE could be one point off, that's the reason all the tests were thought to be correct over the years, because they thought the threshold was at 9 instead of 10.

If you're under 9 it's always going around to 9 or less, so it's always going to be below 10, so you're always going to get the award. 

and yeah I think you can absolutely tell, because if you truncate your balances and do your utilization yourself, you can then watch it as it corresponds to your score changes to determine where the threshold is. 

it's just like if you install a second speedometer in your car. If the original one is still there and off by 5 miles an hour, they're both going to move at the same rate, but the original one is just going to read 5 miles an hour less. So of course the point changes are going to move with the utilization on Experian, but that doesn't mean its showing the right number.

yes Experian FE calculations could qualify for your fluff status BBS, I don't think FICO had any input in development. 

[dragontears]

Wanted to add a data point around this,

This month I had a card go 0% => 75% utilization (total utilization was under 10%), EQ8 dropped 22 points when the card reported 0 again EQ went back up 22 points.