Algorithms

[Anonymous]

Just wondering. 

Do banks set their algorithms for each individual card holder? In this case, lets discuss risk tolerance. When does the computer decide to balance chase, close your account or CLD?

At what utilization across all cards does the computer say "red flag red flag"?

Is it inquiries?

Is it new accounts?

Why can some get AA and others don't?

We know new baddies can cause AA. For discussion purposes, let's talk about clean files with decent scores of 700 plus.

Any thoughts, rhymes or reasons, rythems & tempos?

Keep Barclays out of it. We all know if the wind blows in the wrong direction your account could get whacked.

[Anonymous]

---

@Anonymous wrote:Do banks set their algorithms for each individual card holder?

---

I am sure that cannot be true.  If it were then a bank's algorithm would have a million different if-then statements: if it is consumer Bob Smith do this, if it is Bob Jones do that, if it is Jane Doe do something else.  No reason to have an algorithm if it is designed for each of its cardholders. 

I do think that different CC issuers are likely to have different sets of criteria: Chase might have one set, Citi another, Amex a third, and so on.

You are right that some people seem to be able to have all cards for several months at a very high utilization with no problem, and with others the CC issuer gets freaked out and begins to balance chase.

One very specific sub-species of risk is consumer bust-out: which is where...

an individual applies for a credit card(s), establishes a normal usage pattern and solid repayment history, then racks up numerous charges and maxes out the cards with no intention of paying the bill. Bust-out consists of an initial phase where the individual works to develop the card issuer’s trust and a strong credit profile with the goal of opening numerous accounts and receiving credit line increases so that more funds are available for the second phase of the fraud, where the individual makes transactions that he or she doesn’t plan to repay.

https://www.investopedia.com/terms/b/bustout.asp-0

There have been discussion threads about bust-out here on the Forums.  This reddit thread gives a good overview of many of the things a CC issuer's bust-out detector might look for:

https://www.reddit.com/r/churning/comments/7jla34/card_shutdowns_and_bustout_score_risk_factors/

There are other species of fraud and then again still other kinds of risk where the CC holder is not commiting intentional fraud but is simply being foolish with his money and getting over his head.  All can have similar consequences (the CC issuer shutting down the card) but there may be different algorithms for each.

[simplynoir]

In the end to me it's these main points: balances carried on how many cards, how long said balances have been carried, any evidence of MS, and Bust Out possibility. The first two are tied to each other and any prolonged period of balances probably >50% where minimum payments were made probably raises red flags. MS is understood to be a no-no so that one speaks for itself especially if it's large amounts of money. Bust Out is a little tricky only because sometimes people that are truly innocent get caught in the crossfire but seems like from reading shutdown threads this plays a lot into accounts opened in a short period of time especially with one creditor like Chase.

Another consideration is some banks employ their own risk assessment score to go along with FICO scoring models such as NFCU, AMEX, and Chase. Best guess is those with a higher score both with SP reports and their own risk model probably receive a higher threshold for how long they'll hold back on taking any action from CLD/balance chasing to account closure. We've seen some users get balanced chased on one card but look at another card with similar stats and nothing happens. So I think profile strength if BO isn't a concern probably helps to back them off. At least that's the way I see it.

[RobertEG]

I would suppose that those are all proprietary trade secrets that a business does not publicly share either with consumers or their competitors.

You rarely see publications that disclose any details on such algorithms and their decision points/factors, and businesses rarely use patents to protect such algorithms, as obtaining a patent requires full disclosure of how to make and use the algorithm.

They usually choose the trade secret route, leaving others to attempt to reverse engineer a method only by seeing outcomes.

As a result, one would need a large number of specific anecdotal scenarios with very specific controls to determine the innards of the algorithms.

[Anonymous]

---

@Anonymous wrote:

---

@Anonymous wrote:Do banks set their algorithms for each individual card holder?

---

I am sure that cannot be true.  If it were then a bank's algorithm would have a million different if-then statements: if it is consumer Bob Smith do this, if it is Bob Jones do that, if it is Jane Doe do something else.  No reason to have an algorithm if it is designed for each of its cardholders. 

I do think that different CC issuers are likely to have different sets of criteria: Chase might have one set, Citi another, Amex a third, and so on.

You are right that some people seem to be able to have all cards for several months at a very high utilization with no problem, and with others the CC issuer gets freaked out and begins to balance chase.

One very specific sub-species of risk is consumer bust-out: which is where...

an individual applies for a credit card(s), establishes a normal usage pattern and solid repayment history, then racks up numerous charges and maxes out the cards with no intention of paying the bill. Bust-out consists of an initial phase where the individual works to develop the card issuer’s trust and a strong credit profile with the goal of opening numerous accounts and receiving credit line increases so that more funds are available for the second phase of the fraud, where the individual makes transactions that he or she doesn’t plan to repay.

https://www.investopedia.com/terms/b/bustout.asp-0

There have been discussion threads about bust-out here on the Forums.  This reddit thread gives a good overview of many of the things a CC issuer's bust-out detector might look for:

https://www.reddit.com/r/churning/comments/7jla34/card_shutdowns_and_bustout_score_risk_factors/

There are other species of fraud and then again still other kinds of risk where the CC holder is not commiting intentional fraud but is simply being foolish with his money and getting over his head.  All can have similar consequences (the CC issuer shutting down the card) but there may be different algorithms for each.

---

I get the bust out thing. I believe Experian was the first to come up with some type of algorithms to use for that risk assessment. I could or very well may be wrong on that. Please correct me if I am.

I have read where numerous posters on here have kept cards maxed for years and never heard a peep from any of their creditors. That's what made me get opinions on this subject.

[Anonymous]

I know, you're right.  Some people have 91% + on all cards for many months in a row with no AA.  Others trigger balance chasing.

[Anonymous]

---

@Anonymous wrote:

I know, you're right.  Some people have 91% + on all cards for many months in a row with no AA.  Others trigger balance chasing.

---

Many months and years!

[Kree]

---

@Anonymous wrote:

---

@Anonymous wrote:

I know, you're right.  Some people have 91% + on all cards for many months in a row with no AA.  Others trigger balance chasing.

---

Many months and years!

---

I once spent years over 100% with Discover. When I paid it down, they gave me an auto CLI.