From AZEO to 100% cards reporting-no change

[Anonymous]

---

@Anonymous wrote:
Cardnut The way the algorithm works, it only looks at your current everything. It does not consider the past that’s called trended data and it’s only used in VS4.

Read about trended data and you’ll find evidence to support your view on algorithms that use trended data.

---

Yeah, it can be a snapshot, but there might be a weighted attribute on that card at the time of snapshot. 

Any scoring model that doesn't weight use over time would have no credibility with me and would not be a serious model in my opinion. 

Otherwise, if I'm a lender unknowingly looking at someone's credit report with a high score thinking, wow, this applicant has a tremendous 10 year payment history on 10 credit cards when that applicant SD'd all 10 cards for 10 years and never used them. 

If that were the case, open up as many cards as you can, SD them until the issuer closes them, open more, and SD your way to great credit. I hope that's not possible. 

[Anonymous]

If you’ll go read what I told you you’ll see the truth for yourself. There is no weighted attribute overtime that is considered by FICO algorithms, that’s the whole selling point of the VS4 trended data.

Whether fortunately or unfortunately, it doesn’t matter whether the algorithm holds weight with you or me or suits any of our opinions. The only person‘s opinions they cared about where the creditors. And apparently the creditors find some reason to believe in them. But that’s why that’s not the only thing taken into consideration in a UW decision.

And yes many people do just that.

[Revelate]

There are many things which I think are bat**bleep** stupid in some of the FICO algorithms, but what I or others outside of the FICO data analytics groups think is in a word irrelevant.

Trended data is not used, at all, in any released mainstream FICO algorithm or industry option to my knowledge. Lenders have gone back and forth on reporting trended data but I think most do but without total compliance on that it loses resolution fast.

VS 4 was developed after FICO 9 and with their market share they could afford to go out on a limb. I do think trended data will be used in FICO 10 or whatever it is called but it hasn’t to date and FICO models are the overwhelmingly pulled scores.

Some internal scores like Chase might use it, impossible to say but for everything that truly counts: utilization has no memory.

10 years from now maybe that changes.

[Anonymous]

This is a table of 28 FICO scores with score factors on each one that shows the effect of AZEO across all 3 CRA's on a clean scorecard with only 2 cards.

All 10 EX scores aren't affected that much, with the best results coming from some of the FICO 9 models and EX 2 +6pts. EX 8 didn't move at all.

But at TU/EQ it's a much different story: EQ 8 +19 and TU 8 +14. As much as +27pts on the FICO 9 models, which could make a huge difference for someone who wants to lease a car.

I see a lot of posts about AZEO that use the daily Experian data and wonder what all the fuss is about. EX 8 doesn't seem to care about number of accounts with balances on any scorecard.

It's always better to have more data, so thanks for posting!

[Anonymous]

---

@Anonymous wrote:
If you’ll go read what I told you you’ll see the truth for yourself. There is no weighted attribute overtime that is considered by FICO algorithms, that’s the whole selling point of the VS4 trended data.

Whethrr fortunately or unfortunately, it doesn’t matter whether the algorithm holds weight with you or me or suits any of our opinions. The only person‘s opinions they cared about where the creditors. And apparently the creditors find some reason to believe in them. But that’s why that’s not the only thing taken into consideration in a UW decision.

And yes many people do just that.

---

@Birdman7, it's not that I don't believe you, what I don't believe is there is no scoring model as of yet, 2020, that doesn't take that into account yet. Maybe there are no models yet that do. Maybe some are in the works, or maybe there are a couple we don't know about. In 2020, we don't have decent scoring models yet? That's too bad if that's the case. 

If the scoring models don't take income into effect, then they're not good risk assessors at all anyway. I don't know of any that do. If income is included, AZEO means nothing for any future models (it means nothing in the current newer models). And trending data could be used more efficiently. 

For example, for me, my average monthly balance on 9 cards is about $400 per card. When all these report, the monthly payment amount is going to be around $40 each (average). That shows up as $360 monthly total due each month. Is that a lot? On a 40K income, yes. For a 400k income no. The scoring model would need to reflect that.. The risk is different for someone earning 40K and someone earning 400K obviously. 

So back to OP, like I said, not sure if it matters to report balances every month or not. Maybe it's in a scoring model or not. Can't prove it. If someone would say I've been reporting balances on all my cards the last 6 months and my scores didn't change, great. I do it and will continue to do so, and it might be factored in future scoring models. As time goes by, almost everything  becomes more sophisticated, and trending data might be trending lol. AZEO is too much work in my opinion. And either way it doesn't matter with my scores. If I need to borrow in the future, +/- 20 points wouldn't affect my interest rate. 

[Ficoproblems247]

I agree 100% with your speculation on trended data. What the computers are able to extrapolate now from the patterns they see are more sophisticated than we realize probably. They see how much we let report on each card relative to our limit each month, and are able to determine more and more advanced usage patterns the longer our accounts have been open. Another thing I don’t understand is that some say that our usage has “no memory.” Yes, of course if you max out your cards, let them report for months and take a hit, then pay them back down your score will rebound as the score itself has no memory. The bureaus however keep a record of your usage relative to your limits each month. So if you apply to another lender, they can see your current snapshot looks good, but can still tell you had a shady history.

[Anonymous]

There is no maybe. FICO scores do not use trended data, do the research. Yes in 2020 they’re is a model that uses trended data. I’ve already told you it’s vantage score 4.0.

No if somebody did that for six months it wouldn’t prove anything. Because aging factors would come into play.

[Anonymous]

---

@Anonymous wrote:
There is no maybe. FICO scores do not use trended data, do the research. Yes in 2020 they’re is a model that uses trended data. I’ve already told you it’s vantage score 4.0.

No if somebody did that for six months it wouldn’t prove anything. Because aging factors would come into play.

---

Yeah, probably. I wouldn't take anything as 100% accurate though, because the scoring models are proprietary and they would not want all details revealed. But who knows, maybe they don't care and the entire algorithm for each scoring model is available. That might be a first, since the developers of the models have competitors. 

I don't take anything from anyone, even experts in the field, as 100% gospel. I learned that lesson long ago what happens when you do.

[Anonymous]

You’re right it is proprietary and they’re not gonna tell you the algorithm. However they’re not gonna tell a bald face lie and state that they do not use trended data.

Furthermore how would the CRA‘s market vantage score 4.0 as the only algorithm with Trended data if that were true?

Believe what you want. Everyone’s entitled to their own opinion and perspective.

[Anonymous]

Like I said, you're probably right. 

If we really do only have one model using trended data, how many more are in the works? How many other lesser known models use it? Internal models. Or another variation not referred to as trended data? Call it something else so that the general population doesn't know about it. Who knows. 

If VS4 is marketing itself as the only one with trended data, be sure others will follow suit, or some variation of it. Income might be the next enhancement, which will produce much better predictive results in my opinion.