Major Banks/Credit Card Companies & Algorithims

**ms277017** · Aug 20, 11, 2:39 pm

I am new around here and I already feel at home. I became interested in credit reporting about a year and a half ago.

So I was sitting around board thinking of credit reporting etc. and I began to think: How do the card companies figure out what kind of consumer I am?

So immediately I though of a monkey chained to a terminal banging on a keyboard haha j/k

Anyways, I thought it is an algorithm a massive computer brain that does trillions of calculations for risk on every account in house. Ok so what are the parameters it considers? So I started brain storming thinking of a few: how many payments per cycle, payment amounts per cycle, activity, beginning and ending balance, no activity, over limit etc.

I am obsessive compulsive to a degree and I cannot stand my bank account to have change in it ie: $1200.06 etc. so I started send all of my uneven change to citibank and capital one. Yea I know I am nuts right? Well, as it has turned out for whatever reason I started getting limit increase on capital one and now on citibank. So anything said here is speculation at best alright haha

Anyways, I am leaning towards the result that the algorithms love multiple payments and not really taking the amounts into consideration. I mean the algorithm does take amounts into consideration I am sure, but it seems to really like getting my random change. Literally capital on blows me up with promotions since I started this a year ago. We all know cap one has ugly APR, but every 9 months I get a promo to lower to 11.9% since I started this change sending thing.

Kind of hard for a human to wrap their brain around the idea that a massive computer determines how much the card company 'likes' you haha

Better check to see if your bank charges for sending a ton of few cent payments every month before you run off and try this LOL

Chase does not chage me I send about 30 payments a month on my account, well I have never been charged you would probably want to investigate that beforehand.

Anyone else have any fun speculation on what these algorithms might be looking for in a responsible borrower. Just for fun of course the program that does this stuff cannot be contained in a mere human brain, but I find it interesting to see what other credit reporting junkies might have to say about what they think these algorithms are looking for?

**jeffysdad** · Aug 21, 11, 5:28 am

That's an interesting question and I can't even pretend to know the answer. However, it doesn't seem likely to me that multiple small payments would move the needle much if at all. Maybe they just think you're a trustworthy customer for other reasons?

Your hypothesis assumes that the algorithm is "looking" for reasons to raise your limit. It probably is in fact doing this. But it also must be looking for reasons to lower your limit. Most of these would be obvious: late/missed payment, over limit/high balance, etc. I remember reading a while back that some issuers were monitoring customer behavior via charges; i.e., they'd flag you if they saw that you were charging at a liquor store frequently, for instance. I wonder if that's really true and if so, how often it happens.

**ms277017** · Aug 21, 11, 5:38 am

Exactly Jeffy, the point of me starting this thread is not to say: Hey I know this..

I started it because I find it just mind boggling that they have a super massive computer brain that is just attached to your account lmao

So yea, little payments maybe maybe not eh, but there are multitudes of things and I agree with you that they are looking for bad more so than good haha

I really do put effort into and care for all of my credit accounts no misses, no lates, under limit and at least one transaction a month etc. etc. so you are probably right when you say I am just being rewarded for all of the good actions not just one here or there.

This algorithm is smart, smart, smart though Jeffy and boy you could fool your grandma before you could fool it hahaha

And the liquer store idea... very intresting, not only how much you spend etc., but perhaps where do you spend and is that advantagious to your risk with a given company?

See with these algorithms it is not just what you do do, equally it is what you might do next. predictabillity is corralated with lower risk, that is almost a given if the company feels like they know you and what you might do in the future they have to relax a little on you unless you give them a reason to wake up and start looking at your account again for some reason.

I think the thing the computer must just eat up and love the most is timely payments.... On time never late, maybe not the full balance, but higher than minimum etc. How many times you pay is probably like a 3/4th on the list, but then you have the side of use and how do you use etc. it is a big undertaking to try and think about how it might work haha

eh, but I find it fun to think about stuff like this from time to time. There are obvious things if you do with the compnay they will gain favor with you and equally there are obvious things if you do you can lose favor. Biggest favor loser is a missed payment 30...60..90...etc. So in my opinion missing a payment is probably the worst, spending predictably and paying more-than-min ON TIME is probably the best what do you guys and gals think?

**Iroquois** · Aug 23, 11, 12:40 pm

First off, the bank model cares little about you as an individual. Explicit factors and linked factors of other people's performances that have smilar characteristics are used to predict how you are likely to perform.

If 100 other people have made payments everyday and then go bankrupt then that is what the payment behavior will suggest.

How much you owe, what you buy, how you pay, where you live, how much you make, how long your history is, what your credit mix is, amongst other factors including that can be purchased on the outside (type of car registered to you, have you registered to vote, do you have green eyes reported on your drivers license; those types of things) can factor into the various different scores that will be calculated, such as acceptance score, collection score, bankruptcy score, profitability score etc. If there is statistical significance to you including zip plus 4 rather than just zip in your warrant registrations, it can be found, included and evaluated against others who have done the same and how they have performed over time. Each financial institution will have its own scores that it figures out from the raw data with few relying solely or at all on things like FICO.

What what I've read one of the recent challanges of the past few years is how to predict who was going to just walk away from their mortgage. People with perfectly good credit would walk away and "strategically" default, so willingness to pay behavior and trait research has been undoubtablly done and then matched against who has walked away, where did they do it, etc. In this case someone may be a wonderful client but if others like that individual have walked they could well get a lot of scruitiny.

**Mogul of Pineapples** · Aug 25, 11, 1:10 am

Welcome to the forum, ms277017. Analytical minds are great, think you will fit in just fine here with us!

Anyways, I am leaning towards the result that the algorithms love multiple payments and not really taking the amounts into consideration.

Strangely enough, a couple years back I experienced the opposite. On my AmEx accounts I would regularly make multiple payments per month. The reason was being obsessed with not forgetting, and hence, whenever I logged into my account I would pay the full balance whatever it may be.

Then sometime later (many months) after regularly doing that, I had requested a credit limit increase and was denied. The reason AmEx gave me was something to the effect that my monthly payments were too low (paraphrasing since I do not recall the precise language used). I found that hilarious since the cumulative amount of my payments always equaled or exceeded my monthly charges and there was no balance on that account. The algorithm must have only been looking at the first payment made in a cycle and ignoring the rest.

Speaking of micro-payments, there was a bank that used to offer some checking accounts a few pennies for each online bill pay made. There are members here who have claimed to made hundreds or thousands of payments per month, adding those pennies up to some real money. However that bill pay perk wasn't widely offered and has since been discontinued.

I remember reading a while back that some issuers were monitoring customer behavior via charges; i.e., they'd flag you if they saw that you were charging at a liquor store frequently, for instance. I wonder if that's really true and if so, how often it happens.

American Express took a lot of heat for that and has claimed to have ended it. I'm not sure if other issuers are doing it, because I'm not sure if AmEx would have discontinued the practice if it weren't for the public pressure to do so. Thus far no other banks have been exposed of doing it that I am aware of, but that's not to say some aren't doing it today.

I think the thing the computer must just eat up and love the most is timely payments.... On time never late, maybe not the full balance, but higher than minimum etc. How many times you pay is probably like a 3/4th on the list, but then you have the side of use and how do you use etc.

Agreed but at the same time, some of the cards aimed at balance carriers, the banks may actually like those who pay late on occasion because it's an opportunity for a late fee. The couple people I know that have great creditworthiness but are careless and pay late sometimes, the both of them seem to be bombarded with credit card deals for balance transfers and promotional 0% on purchases. Perhaps the occasional late payer that the algorithm determines to be a low risk would be desirable given the potential for greater revenue from the account, right?

do you have green eyes reported on your drivers license; those types of things) can factor into the various different scores that will be calculated, such as acceptance score, collection score, bankruptcy score, profitability score etc

Well I'm not so sure about the eye color part, being that some ethnicities only have one eye color, that would certainly be racial profiling and a big class action lawsuit.

Thread: Major Banks/Credit Card Companies & Algorithims

LinkBack

Thread Tools

Display

Major Banks/Credit Card Companies & Algorithims

Your behavior is matched against others who look like you. You'll be viewed based on