With FICO 8, if all bureaus had the exact same information, would the scores be the same?

[Anonymous]

But this was not the case with old flavors, correct? Still trying to figure it all out.

[CH-7-Mission-Accomplished]

---

@Anonymous wrote:

But this was not the case with old flavors, correct? Still trying to figure it all out.

---

No.  

To make things even more confusing, each CRA has 'tweaked" the algorithm so there are really three times the number of FICO 08 scores, plus three times the FICO 08 derivations like auto enhanced, etc.

Although the score is useless because it is not used, at least Vantage score is supposed to be the same algorithm across the three since it was developed by all three CRAs.

[Anonymous]

CH-7 is right.  The actual FICO 8 Classic algorithm differs slightly between the three CRAs.  Not just the data upon which the algorithm runs but the actual algorithm itself.

That said, it would be interesting if somehow you could take 1000 clean profiles at random, eliminate all profiles that do not have exactly the same accounts and the same number of inquiries, and then see how much different the three FICO 8 scores were.  I am guessing that you would find that on average the difference was extremely small.

I remember being surprised this summer when I pulled my scores here to see that my FICO 8 Classic scores were almost identical.  Two were exactly the same and one was exactly one point off.  I had expected a bigger difference.  All my other FICO scores (especially the earlier models) had a much bigger difference between the three CRAs.

[RobertEG]

While it was several years ago, Fair Isaac stated in one of their old webinars that with identical data in their credit files, the vairaiton in the algorithms they customize for each CRA could result in a score difference between CRAs of up to a max of 30 points.

[Thomas_Thumb]

---

@CH-7-Mission-Accomplished wrote:

---

@Anonymous wrote:

But this was not the case with old flavors, correct? Still trying to figure it all out.

---

No.  

To make things even more confusing, each CRA has 'tweaked" the algorithm so there are really three times the number of FICO 08 scores, plus three times the FICO 08 derivations like auto enhanced, etc.

Although the score is useless because it is not used, at least Vantage score is supposed to be the same algorithm across the three since it was developed by all three CRAs.

---

Correct - VantageScore 3.0 is touted as being applied uniformly across all three CRAs. However, the score is not useless.

Based on my own results, VS3 scoring parity certainly appears to be true - unfortunately, it is unlikely that I will ever have identical data in reports due to a different total account count (11 vs 10) in Oct (10 vs 9) in Nov or an inquiry (1 vs 0) depending on CRA but not another. The EQ inquiry - is between 1 and 2 years old (VS goes to 24 months for INQs)

Even with these minor differences, spread between CRA scores is 2 or 3 points month to month. I really can't speak to Fico 8 due to the buffer - but Fico 8 enhanced (both bankcard and Auto) indicates score model tweaking - with EQ showing greater divergence.

[elim]

I can't figure it out. I have the same inq's on EQ and EX (1 on TU usually)

[Thomas_Thumb]

Nice graphs! Where do those come from?

Do you have any difference in # accounts among the 3 CRAs and or AAoA? A 20 point score spread on VS3 seems rather large "all things being equal". A 20 point score spread on Fico 8 would not be surprising - but your 40 point spread is.

Also can you provide an exact INQ count by CRA for your last set of reported scores? Hopefully no INQs are associated with Auto loans as they may (or may not) be counted as one in scoring

Anything lurking in the Badlands, such as a 30 day late with a couple years age on it?

[Anonymous]

---

@Thomas_Thumb wrote:

Nice graphs! Where do those come from?

Do you have any difference in # accounts among the 3 CRAs and or AAoA? A 20 point score spread on VS3 seems rather large "all things being equal". A 20 point score spread on Fico 8 would not be surprising - but your 40 point spread is.

Also can you provide an exact INQ count by CRA for your last set of reported scores? Hopefully no INQs are associated with Auto loans as they may (or may not) be counted as one in scoring

Anything lurking in the Badlands, such as a 30 day late with a couple years age on it?

---

Great questions by TT!  Note, however, that even if the number of accounts was the same you could still have a huge difference in credit profile.  For example, EQ and TU could both show exactly five credit cards: 

          EQ = cards A, B, C, D, and X

          TU = cards A, B, C, D, and Y

but cards X and Y coulld tilt the score very differently (X is maxxed out and Y has a zero balance, and both have very high credit limits).

So the point is that it's important that the accounts themselves be the same, not only should the summary figure be the same.

Even when the accounts themselves are the same it's possible to have wildly different credit scores at an instant in time.  Here's an example.  Suppose somebody has exactly two credit cards, A and B, both of which tend to report on the 7th.  The two cards were both nearly maxxed out, but then they were abruptly paid down (card A to $0 and card B to 1% utilization).  You check your FICO scores on the 8th.  EQ has not yet registered the new balances, TU has registered the new balance of A but not B, and EX has registered both new balances.  EQ will show an overall util of 98%, TU of 50%, and EX of 1% -- all with exactly the same accounts and possibly the same inquiries.

So in the hypothesis of identical data, it's important that every piece of data be the same.  Identical accounts, identical balances on corresponding accounts, identical derogs in the same places, identical age of oldest, identical AAoA, etc.

I personally have had times in my credit history where my accounts were superficially identical and it took a while of carefully cross checking each item to find the place where they were different (and significantly so). 

Interesting discussion thus far.  Thanks!

[elim]

The charts are in Google Sheets.  I track all my stuff on them and have many.  I think I can even embed them as interactive here (trying now)

https://docs.google.com/spreadsheets/d/12SC6wA1ixAu-sO-TZ5hylXfoCq2wsMhinzw22sCUtLw/pubchart?oid=682...

edit: The charts won't embed like they will on a regular site but they link solely to the chart and not the entire sheet (and are interactive with real time changes).  Still very efficient

I have a major derog on all 3 being 2010 tax lien.  All accounts are equal on reports except for an old paid auto loan that is solely on EX.  One old Home Depot CC from 06' ($5600) is being reported as "open" on TU which is the cause of my nice TU scores. It will fall off next year so i'll see what happens there.

EQ seems to plummet when i let balances report.

[Anonymous]

Great thread! Same questions run through my mind ... I have the same information between all three bureaus on the FICO 08 yet the scores deviate by 8 points. Yes, for me the Vantage Score (which is really useless) is the same for all three bureaus. Good info glad I am not alone wondering about this.