Experian DPs on a thin file

[Anonymalous]

I don't know if these DPs will help, but I have an extremely thin and clean file so it should be easy to isolate factors.

Experian (got the CCT trial on 10/29, free prior):

• 10/3: Unscorable, AAoA 5 mos, AoOA 5 mos
• 10/28, 29, 31: FICO 8 763, AAoA 11 yrs 9 mos, AoOA 23 yrs 1 mo
• 11/1: FICO 8 767, AAoA 11 yrs 10 mos, AoOA 23 yrs 2 mo

I got my first credit card in mid-May. I was added as an AU to a much older account, which first reported on 10/15. While Experian doesn't list the AAoA for each of the cards, it's clear that the age used for my first card in October was 5 mos (average (23 yrs 1 mo, 5 mos) = 11 yrs 9 mos), and both cards aged 1 month on 11/1. Three conclusions:

1. The addition of the AU card is what made me scorable, not the age of my first card (which wasn't FICO's minimum of 6 months when I first got a score).
2. Cards age on the 1st of the month.
3. The +4 increase must be due to age, though I didn't pass any year threshold.

I compared every page of the 10/31 and 11/1 reports, and the only difference is age. None of the other information changed. Neither my personal card nor the AU card updated, and I have nothing except some personal information anywhere else on the report, not even a hard pull (for some reason, Capital One didn't pull Experian). But I still ended up with radically different What's helping/What's hurting:

10/31 (for FICO 8):

What's hurting
- No revolving activity

11/1 (for FICO 8):

What's helping
- No missed payments
- Low revolving credit usage
- Not seeking credit
- Recent credit card usage
What's hurting
- Few accounts paid on time

How can 2 cards aging 1 month, with no change to the underlying data, somehow result in a completely different set of recommendations? It doesn't seem to make much sense, at a human level. Which suggests the recommendations use machine learning. Except....

Experian/CCT also provides FICO 2, two auto, and three bankcard scores. They're also bizarre. Between 10/31 and 11/1 they increased 0 to 13 points, except for one that dropped 19 points (Auto 8). Again, why? This is hard to blame on machine learning, because even FICO 8 dates back to 2009.

[NYC_Fella]

Thanks, interesting DPs.

[Anonymous]

You get bumps for accounts aging before 12 months. I recently got a significant bump from 3 cards aging 3 months (I rejoined the credit world after taking a break for a few years) . I checked my reports and nothing else had changed.

---

@Anonymalous wrote:

I don't know if these DPs will help, but I have an extremely thin and clean file so it should be easy to isolate factors.

Experian (got the CCT trial on 10/29, free prior):

• 10/3: Unscorable, AAoA 5 mos, AoOA 5 mos
• 10/28, 29, 31: FICO 8 763, AAoA 11 yrs 9 mos, AoOA 23 yrs 1 mo
• 11/1: FICO 8 767, AAoA 11 yrs 10 mos, AoOA 23 yrs 2 mo

I got my first credit card in mid-May. I was added as an AU to a much older account, which first reported on 10/15. While Experian doesn't list the AAoA for each of the cards, it's clear that the age used for my first card in October was 5 mos (average (23 yrs 1 mo, 5 mos) = 11 yrs 9 mos), and both cards aged 1 month on 11/1. Three conclusions:

1. The addition of the AU card is what made me scorable, not the age of my first card (which wasn't FICO's minimum of 6 months when I first got a score).
2. Cards age on the 1st of the month.
3. The +4 increase must be due to age, though I didn't pass any year threshold.

I compared every page of the 10/31 and 11/1 reports, and the only difference is age. None of the other information changed. Neither my personal card nor the AU card updated, and I have nothing except some personal information anywhere else on the report, not even a hard pull (for some reason, Capital One didn't pull Experian). But I still ended up with radically different What's helping/What's hurting:

10/31 (for FICO 8):

What's hurting
- No revolving activity

11/1 (for FICO 8):

What's helping
- No missed payments
- Low revolving credit usage
- Not seeking credit
- Recent credit card usage
What's hurting
- Few accounts paid on time

How can 2 cards aging 1 month, with no change to the underlying data, somehow result in a completely different set of recommendations? It doesn't seem to make much sense, at a human level. Which suggests the recommendations use machine learning. Except....

Experian/CCT also provides FICO 2, two auto, and three bankcard scores. They're also bizarre. Between 10/31 and 11/1 they increased 0 to 13 points, except for one that dropped 19 points (Auto 8). Again, why? This is hard to blame on machine learning, because even FICO 8 dates back to 2009.

---

This added like 25 points between the 3 (and pushed me into the 740s). 

[Anonymalous]

Interesting, @Anonymous. My FAKO (VS 3.0) scores were between 674 and 681 for the 2nd to 5th months of my first card, only increasing +1 to +3 points per month, before jumping almost 80 points with the addition of the AU card. If what you're saying is true, that's a major difference from the FICO model, where you saw approximately a +8 increase per month. Even if we assume the bonus applies to each your 3 cards separately, that's still higher than what I was seeing in my VSs.

But if that applies now, after the addition of the AU card, that means it's based on the age of the newest card hitting 6 months, not on the average age or the oldest, because neither of those passed a year threshold. And the distribution of the increases is still very strange -- +4, +6, +12, -19, +11, +0, and +13 scattered among the various FICO scores provided by Experian isn't a consistent pattern. It seems random.

[Anonymous]

Just to clarify -- I did not see that jump PER month. It jumped all at once after the accounts aged 3 months. Must be a threshold in FICO.