14 points for AZ penalty

[SouthJamaica]

---

@TheKid2 wrote:

---

@GZG wrote:

---

@SouthJamaica wrote:

---

@TheKid2 wrote:

You'd think logically that a thick, mature file would take a bigger hit then thin, young. Seems awfully penal to me for someone young taking 14 points to pay off all their cards. They like when you pay interest I guess.

---

Yeah well I find the "all zero" penalty to be oddly penal.

---

I wonder WHY users that use <10% of their credit during any given month are some minute percentage more trustworthy with credit than those who use 0% of their revolving credit in any given month. 

I'm sure they have the data that suggests that it's true, but I'd be curious to learn why it's true. 

---

Curious as well, but I'm guessing this leaks into the question of profit for lenders who use the FICO service. It's not that someone with no balances is more RISKY, it seems like that person is less ATTRACTIVE to lend since they aren't paying any interest when they don't carry any revolving balance at all. It could be that the ideal consumer isn't all about RISK, it also factors in ones willingness to pay some interest (profit for the lender).

---

Exactly. They like to say the scores are about "risk", but this aspect of the algorithms is about profitability.

[GatorGuy]

---

@TheKid2 wrote:

You'd think logically that a thick, mature file would take a bigger hit then thin, young. Seems awfully penal to me for someone young taking 14 points to pay off all their cards. They like when you pay interest I guess.

---

You can have balances report without ever paying interest. In fact probably the majority of people who have balances report never pay interest.

[Anonymalous]

---

@SouthJamaica wrote:

Just a little data point. A relative with a fairly thin, fairly young, file -- 1 store card + 3 bank cards -- got hit with "all zero" penalty; it was 14 points in FICO 8.

---

My last all zero penalty on EX FICO 8 was 14 points, as well. Four months before that, the penalty was 17 points. Young, thin file (3 bank cards + 1 AU store card, both penalties were from when I had less than a year of history).

[SouthJamaica]

---

@Anonymalous wrote:

---

@SouthJamaica wrote:

Just a little data point. A relative with a fairly thin, fairly young, file -- 1 store card + 3 bank cards -- got hit with "all zero" penalty; it was 14 points in FICO 8.

---

My last all zero penalty on EX FICO 8 was 14 points, as well. Four months before that, the penalty was 17 points. Young, thin file (3 bank cards + 1 AU store card, both penalties were from when I had less than a year of history).

---

Interesting. Penalty the same. Profile very similar.

[Zoostation1]

The more I try and understand the algorithms the more frustrated I get. I really wish that the models were more tied to whether you pay in full before your due date than what you report on your statement.  It drives me insane to know that if you take 3 people (person A, B, and C)who all spend the exact same amount each month on their cards, and all consistently pay in full before the due date can have such vastly different scores and are perceived as significantly diffferent risks. It seems like as we move forward towards newer models, the penalties for not playing the game and just paying your bill in one payment before othe due date (like my dad does) are getting worse, especially if you don't have very high limits.  If you have low limits you're effectively penalized for making larger purchases near or on your statement date.  The date in which you make a large purchase and where it falls relative to your statement closing date should never significantly matter relative to your score imo. People may disagree with me but I just don't think your statement balances should play a significant role if you're PIF and never carry an interest collecting balance month to month.  There's got to be some better way to distinguish between those who carry over a 50% balance each month and are paying interest out the wazoo and those who show a 50% balance when the statement cuts but are paid in full 5 days before the due date month after month.

In regards to the all zero penalty, I haven't had that yet cause I've had at least one card report each time.  With me having a thin file having 2 cards report seems to be a pretty big ding though esp on EQ.

[Anonymalous]

---

@Zoostation1 wrote:

The more I try and understand the algorithms the more frustrated I get. I really wish that the models were more tied to whether you pay in full before your due date than what you report on your statement.  It drives me insane to know that if you take 3 people (person A, B, and C)who all spend the exact same amount each month on their cards, and all consistently pay in full before the due date can have such vastly different scores and are perceived as significantly diffferent risks. It seems like as we move forward towards newer models, the penalties for not playing the game and just paying your bill in one payment before othe due date (like my dad does) are getting worse, especially if you don't have very high limits.  If you have low limits you're effectively penalized for making larger purchases near or on your statement date.  The date in which you make a large purchase and where it falls relative to your statement closing date should never significantly matter relative to your score imo. People may disagree with me but I just don't think your statement balances should play a significant role if you're PIF and never carry an interest collecting balance month to month.  There's got to be some better way to distinguish between those who carry over a 50% balance each month and are paying interest out the wazoo and those who show a 50% balance when the statement cuts but are paid in full 5 days before the due date month after month.

In regards to the all zero penalty, I haven't had that yet cause I've had at least one card report each time.  With me having a thin file having 2 cards report seems to be a pretty big ding though esp on EQ.

---

I think it's futile to expect fairness, because credit scoring is not about treating every person justly. It's about risk management. That's why I think it's best to see it as a game, because I don't expect the rules of a game to be an immanent truth. It's just a set of rules, and if I learn them and learn to play by them, I can get good results. Though it's also true we don't really know the rules.

I'm a thin file as well, and I've been playing with my cards. One thing I've discovered with 3 personal cards is AZEO doesn't seem to be important. I just went from 2/3 cards reporting to 1/3 cards reporting, and I gained exactly 1 point on EX FICO 8. Though my overall utilization was very low -- in both cases 1% overall, with my highest card reporting 2% or 3%.

Utilization matters more. When I had 1 card over 50%, the other 2 cards had token balances (3% or less), and overall utilization was <10%, I lost 16 points. When I had 1 card over 30%, the other 2 card had token balances, and overall utilization was <10%, I lost 13 points. When I had 1 card over 30%, the other cards had token balances, and overall utilization was <10% (by a bit more this time), I lost 9 points.

So allowing multiple cards to carry a balance doesn't seem to have much effect, but allowing individual cards to pass certain utilization thresholds does have a modest effect.

[SouthJamaica]

---

@SouthJamaica wrote:

Just a little data point. A relative with a fairly thin, fairly young, file -- 1 store card + 3 bank cards -- got hit with "all zero" penalty; it was 14 points in FICO 8.

---

Update: when a balance reported today, the 14 points came right back.

[TheKid2]

---

@Zoostation1 wrote:

The more I try and understand the algorithms the more frustrated I get. I really wish that the models were more tied to whether you pay in full before your due date than what you report on your statement.  It drives me insane to know that if you take 3 people (person A, B, and C)who all spend the exact same amount each month on their cards, and all consistently pay in full before the due date can have such vastly different scores and are perceived as significantly diffferent risks. It seems like as we move forward towards newer models, the penalties for not playing the game and just paying your bill in one payment before othe due date (like my dad does) are getting worse, especially if you don't have very high limits.  If you have low limits you're effectively penalized for making larger purchases near or on your statement date.  The date in which you make a large purchase and where it falls relative to your statement closing date should never significantly matter relative to your score imo. People may disagree with me but I just don't think your statement balances should play a significant role if you're PIF and never carry an interest collecting balance month to month.  There's got to be some better way to distinguish between those who carry over a 50% balance each month and are paying interest out the wazoo and those who show a 50% balance when the statement cuts but are paid in full 5 days before the due date month after month.

In regards to the all zero penalty, I haven't had that yet cause I've had at least one card report each time.  With me having a thin file having 2 cards report seems to be a pretty big ding though esp on EQ.

---

You have valid points and it's obvious that the playing field is tilted to the those with mature credit histories. But, I don't think we can call it unfair because that's their reference for how much debt you're carrying. Each month the creditors are expected to report a balance. You're paying your bills on time every month,  you will gain points as the accounts age, you will get CLIs to lower utilization percentage, you will get more points for that. It's definitely a long game right now.

To your point, it will be interesting to see how the new 10 models behave. Some analysis on utilization "memory" is baked in. Are your 10s your hightest scores?

P.S Making multiple payments over the course of the month will lower your interest owed. It is a money saving strategy and it helps your scores.

[ThomasJNewton]

---

@Zoostation1 wrote:

The more I try and understand the algorithms the more frustrated I get. I really wish that the models were more tied to whether you pay in full before your due date than what you report on your statement.

You understand that this is just about correlations based on millions of people who know nothing about this at all, not based on the thousands of people on here who consciously modify their behavior to improve their scores, right? Their models show that people with the behavior of having a balance on the card all the time (not even running a balance, but having one when it reports), are more of a risk. The vast majority of people know nothing about how these things work and so they use their cards without thinking about it.

It drives me insane to know that if you take 3 people (person A, B, and C)who all spend the exact same amount each month on their cards, and all consistently pay in full before the due date can have such vastly different scores and are perceived as significantly diffferent risks.

It is not about perceived risk, it is about measured risks. They look at data for groups of people who are otherwise similar and try to spot differences. This is not causative, it is correlative. The reason that it is so easy for people who understand the process to manipulate their scores with things like self loans, is that very few people know to do it. While you are right that it might seem unfair that people who make the same amount and spend the same amount might have very different scores, it is mostly because large groups of people who behave in ways similar to each of your three people are more or less likely to default within a certain number of days.

It seems like as we move forward towards newer models, the penalties for not playing the game and just paying your bill in one payment before othe due date (like my dad does) are getting worse, especially if you don't have very high limits.  If you have low limits you're effectively penalized for making larger purchases near or on your statement date.  The date in which you make a large purchase and where it falls relative to your statement closing date should never significantly matter relative to your score imo.

It matters simply because the reporting companies do not report daily balances (or even more often). If they report only once a month, then without a lot more data (e.g. Does the customer run a balance? Is this the kind of transaction they make often? What is the current interest rate on this card? etc.). Fair Isaac did not decide to punish people who made big purchases near their statement close date, they simply noticed that people who have high balances relative to their amount of credit are more likely to default. Your saying it should not matter is arguing that the data should be ignored. That either leads to more defaults leading to higher rates and fees and lower limits, or making it even harder for people with any issues to ever get credit.

People may disagree with me but I just don't think your statement balances should play a significant role if you're PIF and never carry an interest collecting balance month to month.

You say this based on what detailed data analysis you have done? How big was your dataset? What period did you study? Since this system is not about determining causes for these defaults, all they have is the data, and it shows what it shows. When I was younger, I had my first card from Gary Wheaton Bank. I had a limit of $500 or something. At the time, the only way to pay was mailing them a check, so there was no real way to pay multiple times during the month. Although I was young (under 18 when I started), I was a software developer and was making good money. I would charge my card to the limit and then when the bill came I would pay it off. I could not get a second card because that one always showed it was maxed out. Finally, I had to not use it for a month and then get a second card and eventually a third, so that I would only ever show 30% on any of the cards. My situation was not that of most of their customers, and so it was frustrating. For me to have said that they should not have been able to penalize me because of my behavior would have been misguided. As they say in law: "Extreme cases make bad laws.

There's got to be some better way to distinguish between those who carry over a 50% balance each month and are paying interest out the wazoo and those who show a 50% balance when the statement cuts but are paid in full 5 days before the due date month after month.

The card issuers would need to report more data to the credit reporting agencies and then companies like Fair Isaac would have to update their models. It might change what the models showed and then again it might not. No way to know.

[Anonymalous]

---

@ThomasJNewton wrote:

---

@Zoostation1 wrote:

The more I try and understand the algorithms the more frustrated I get. I really wish that the models were more tied to whether you pay in full before your due date than what you report on your statement.

You understand that this is just about correlations based on millions of people who know nothing about this at all, not based on the thousands of people on here who consciously modify their behavior to improve their scores, right? Their models show that people with the behavior of having a balance on the card all the time (not even running a balance, but having one when it reports), are more of a risk. The vast majority of people know nothing about how these things work and so they use their cards without thinking about it.

---

Yes, but the underlying criticism is based on a deeper mismatch, because the models are operating on limited data. The argument @Zoostation1 is making is that people who pay off their balance every month are low risk. While the data doesn't show it, it's not because the argument is wrong. It's because the data simply isn't available. The credit models are based on information reported by lenders, which only contains the balance at a particular moment in time (typically when the statement closes, but not always), and some secondary data like the highest balance over a period. It doesn't capure whether people are paying in full every month, or just the minimum balance.

So while the risk assessment is based on correlations in large sets of data, in a lot of cases it's inferring behavior based on secondary evidence that's only loosely correlated to the underlying causes. So there will be mismatches, or cases where people who consistently behave in a responsible way are punished by the system, and where people who are higher risk are able to game the system for better results.

That's why I think credit education is important, and why it's also best to treat it as a bit of a game. Because being responsible doesn't mean everyone else will automatically know I'm responsible. So it's up to me to prove it. And if I know the basic rules they use to measure responsibility, I can take the necessary steps to make sure I'm presented in a positive light. A good credit score becomes a badge or a credential, instead of 3 digit scarlet letter.

A lot of the objections seem to be based on a visceral sense of unfairness. And they're absolutely correct, it's not fair if you have the expectation that the system will correct assess everyone based on their essential characteristics. But when I see as more of a game, it doesn't have to be fair in some cosmic sense. If it's just a pseudo-arbitrary set of rules, then the rules aren't really saying anything about me as a person, and fairness is determined by whether the rules are consistently applied.