topic Re: Anyone else think the algorithm is wrong??? in Understanding FICO® Scoring

Anyone else think the algorithm is wrong???

AlgorithmIsWrong — Fri, 22 Mar 2024 14:33:50 GMT

Like seriously, you get one late payment and you get hit with a 20+ point decrease. While making consistent, on-time monthly payments and your score increases slower than molasses . A car note gets paid off, there goes your score down the drain. There's something definitely wrong with the whole model and I wonder if anyone within the three credit bureaus and myFICO know how to reimagine the system.

Re: Anyone else think the algorithm is wrong???

FlaDude — Fri, 22 Mar 2024 15:37:32 GMT

FWIW, FICO is definately being tweaked. While FICO 8 scores (and even older models) are still commonly used, FICO 9 exists and 10 is there too.

Since FICO is meant to be a predictor of creditworthiness, I think it is expected that not making a payment on time should have a negative score impact. As far as the credit mix issue (paying off your car loan), that may be more debatable, but if they have statistics that show that people with no active installment loans are more likely to default, then that should be factored into the scoring.

If you haven't ready any of this thread, you might want to look at some of it:

We-re-Tom-Quinn-amp-Tommy-Lee-FICO-Score-Experts-Ask-us-anything

Re: Anyone else think the algorithm is wrong???

AlgorithmIsWrong — Fri, 22 Mar 2024 16:07:17 GMT

Prediction is pretty much an opinion in my eyes and peoples credit worthiness should not be judged based on an opinion. If one late payment hurts you by 20+ points when all your prior payments were on time, one would take into account all the positive history and slightly ding the person for a late payment. Key word there is "slightly." Now, if the late payments continue, then sure, I could understand the huge decline. This is why the system is flawed.

Perhaps, delayed reporting from agencies to the bureaus is warranted. Where a 3 to 6 month period is scored rather than individual months.

Re: Anyone else think the algorithm is wrong???

CYBERSAM — Fri, 22 Mar 2024 16:16:20 GMT

To pay your bills on time is not a favor to credit word, its an obligation. They don't see that behavior as amazing, only not as negative. However when you miss payments then that is a big red flag.

Having an Excellent credit score, dosen't mean you are special. It's mean you are doing what you suppose to do.

You are living in their word, not other way around.

Re: Anyone else think the algorithm is wrong???

AlgorithmIsWrong — Fri, 22 Mar 2024 16:44:53 GMT

@CYBERSAM , for being a Senior Contributor, one would think you'd bring in more insight and factual information, but that response was just not good. Full of opinion just like the bureaus themselves.

Step it up my guy. I know you can be better.

Re: Anyone else think the algorithm is wrong???

coldfusion — Fri, 22 Mar 2024 17:16:29 GMT

@CYBERSAM does just fine.

Re: Anyone else think the algorithm is wrong???

Slabenstein — Fri, 22 Mar 2024 17:43:34 GMT

Wrong with respect to what? A score model that didn't penalize properly for delinquency wouldn't be useful to lenders for underwriting, and so wouldn't be used in any case.

Re: Anyone else think the algorithm is wrong???

CorpCrMgr1 — Fri, 22 Mar 2024 17:43:50 GMT

It was once explained to me in the Forums it's like your significant other cheating on you one time. Ten years of flawless behavior and then the cheating. How much could you trust this person?

I had stated prior it's like with a baseball hitter not getting a hit 50% of the time would still be considered amazing player.

Re: Anyone else think the algorithm is wrong???

Brian_Earl_Spilner — Fri, 22 Mar 2024 17:46:28 GMT

@AlgorithmIsWrong wrote:
Prediction is pretty much an opinion in my eyes and peoples credit worthiness should not be judged based on an opinion. If one late payment hurts you by 20+ points when all your prior payments were on time, one would take into account all the positive history and slightly ding the person for a late payment. Key word there is "slightly." Now, if the late payments continue, then sure, I could understand the huge decline. This is why the system is flawed.

Perhaps, delayed reporting from agencies to the bureaus is warranted. Where a 3 to 6 month period is scored rather than individual months.

The score is a quick glance at likelihood to perform and is based on datapoints. As mentioned before, if they find a correlation between datapoints that is consistent across enough people, it can be assumed that it is a fairly accurate predictor of what can happen and factored into the algorithm.

Now, is it fair? I don't think so. I just finished a 7 year rebuild meaning I went 7 years since my last derogatory, but was penalized until the end. I also don't like the fact that paying my chargeoffs and collections still hurt me more than a bankruptcy. I fulfilled my obligations, people who filed bankruptcy did not. I didn't hit the 700s until about 5 years in, while some filing bankruptcy were able to do it in as early as 2 years. I was relegated to toy limits while people literally a week out of bankruptcy were able to get multi-thousand dollar limits from a lender they just burned. So I get what you're saying in regards to scoring us on the overall reporting versus a single incident. But that's not how it works. The score is based on a snapshot of your profile at that moment in time. The new trending scores will hopefully help in a situation like mine, showing I was paying everything down and off and scoring me appropriately for it.

Having said all that, the score is just a number and doesn't dictate likelihood of approval. Lenders will always use their own criteria for approvals. I've always been an outlier and able to get approved for things my scores say I shouldn't. It also goes the other way. Cap1 is not friendly to people with 800+ scores. I use it as just a guideline and to quickly verify nothing is wrong. If I see my score drop 30 points, I know there's something that needs my attention. Otherwise, I don't bother checking it before an app as the lender will either approve me, or they won't.

Re: Anyone else think the algorithm is wrong???

Kforce — Fri, 22 Mar 2024 18:07:08 GMT

@AlgorithmIsWrong wrote:
Like seriously, you get one late payment and you get hit with a 20+ point decrease. While making consistent, on-time monthly payments and your score increases slower than molasses . A car note gets paid off, there goes your score down the drain. There's something definitely wrong with the whole model and I wonder if anyone within the three credit bureaus and myFICO know how to reimagine the system.

I think they are doing a good job.

Not perfect but not really a broken system

Not everyone will agree with every point loss or gain, but the overall

things that are looked at and considered make sense.

I wish not having a house or car loan in the last 15 years was not depressing

my score, however I understand that I have no recent history to look at.

Same with all zero, seems petty to punish for paying everything off,

but if you look at it as they have no history of you borrowing anything !

The few points lost are made up by years of payment history

and low utilization. All zero and no auto:mortgage loans and you

have 840's not 850. Even -30 for a 30 day and you are still top tier.

Re: Anyone else think the algorithm is wrong???

Kforce — Fri, 22 Mar 2024 18:20:28 GMT

@Brian_Earl_Spilner wrote:
Now, is it fair? I don't think so. I just finished a 7 year rebuild meaning I went 7 years since my last derogatory, but was penalized until the end. I also don't like the fact that paying my chargeoffs and collections still hurt me more than a bankruptcy. I fulfilled my obligations, people who filed bankruptcy did not.

I agree this is not fair

Having said all that, the score is just a number and doesn't dictate likelihood of approval. Lenders will always use their own criteria for approvals. I've always been an outlier and able to get approved for things my scores say I shouldn't. It also goes the other way. Cap1 is not friendly to people with 800+ scores. I use it as just a guideline and to quickly verify nothing is wrong. If I see my score drop 30 points, I know there's something that needs my attention. Otherwise, I don't bother checking it before an app as the lender will either approve me, or they won't.

I believe many worry way to much about score.
Agree 100%

Re: Anyone else think the algorithm is wrong???

Thomas_Thumb — Fri, 22 Mar 2024 22:49:33 GMT

@AlgorithmIsWrong wrote:
Prediction is pretty much an opinion in my eyes and peoples credit worthiness should not be judged based on an opinion. If one late payment hurts you by 20+ points when all your prior payments were on time, one would take into account all the positive history and slightly ding the person for a late payment. Key word there is "slightly." Now, if the late payments continue, then sure, I could understand the huge decline. This is why the system is flawed.

Perhaps, delayed reporting from agencies to the bureaus is warranted. Where a 3 to 6 month period is scored rather than individual months.

Twenty points is a trivial decline. A huge decline is 120 points or more -IMO. A single 30+ day late is bad and indicates increased risk. An extended late 60+ or multiple lates represent greater risk and a 90+ day late is a red flag.

Recent Fico algorithms have already been tweaked to mute the impact of derogatories. Thus, the inflation of scores in recent years.

Data shows strong correlation between risk of future derogatories and presence of lates. The impact of lates on score does decline over time and the reduced risk assessment is backed by data.

Bottom line - the scoring algorithms are data driven and are not based on opinion. Issues such as paid vs unpaid collections as well as medical debt have been addressed in recent models. Paid = lower risk to lenders.

Re: Anyone else think the algorithm is wrong???

Thomas_Thumb — Fri, 22 Mar 2024 22:43:59 GMT

@Brian_Earl_Spilner wrote:
Lenders will always use their own criteria for approvals. I've always been an outlier and able to get approved for things my scores say I shouldn't. It also goes the other way. Cap1 is not friendly to people with 800+ scores.

Doom!?

Will my Discover card CL be slashed when Cap One takes over? Will my card be cancelled?

Re: Anyone else think the algorithm is wrong???

Horseshoez — Sat, 23 Mar 2024 02:40:10 GMT

@Thomas_Thumb wrote:
Bottom line - the scoring algorithms are data driven and are not based on opinion. Issues such as paid vs unpaid collections as well as medical debt have been addressed in recent models. Paid = lower risk to lenders.

You beat me to it; when it comes to assembling the tables for credit scoring based upon millions and billions of bits of actuarial data, there is literally zero "opinion" involved. @AlgorithmIsWrong, you are welcome to your opinion, however, you missed the mark by a wide margin.

Re: Anyone else think the algorithm is wrong???

ThomasJNewton — Sat, 23 Mar 2024 16:41:53 GMT

@AlgorithmIsWrong wrote:
Prediction is pretty much an opinion in my eyes and peoples credit worthiness should not be judged based on an opinion.

Sorry, it is not an opinion, but the result of statistical analysis of millions of records. FICO finds groups of profiles that are similar and have defaulted, then they go back to the period they defaulted and find profiles that are similar but did not default and look for differences. Those differences are the basis of their models.

If one late payment hurts you by 20+ points when all your prior payments were on time, one would take into account all the positive history and slightly ding the person for a late payment.

You act as if this is a value judgement. It is not. It is simply an observation that people whose profile looks like the one in question have a certain percentage greater chance of defaulting after a 30-day late. If it really is a simple problem (your bank merged and so your account number changed causing your autopay to fail), your credit card company will often remove it as a courtesy.

Key word there is "slightly." Now, if the late payments continue, then sure, I could understand the huge decline. This is why the system is flawed.

You feel it is not a big deal, and so you should not take a big hit. Unfortunately, statistics show something else. The system's problem is that it knows nothing about why something happened, only that it did. There are a million reasons that a payment was missed, some of which having nothing to do with financial stress. Unfortunately, there is no way for a fully automated system to take those into account, as it cannot know enough about you to know what the issue was. The entity that can take those things into account is your credit issuer. That is why they can do courtesy deletes.

Perhaps, delayed reporting from agencies to the bureaus is warranted. Where a 3 to 6 month period is scored rather than individual months.

First, many issuers already delay thirty days (if you miss a payment date, they do not report it until it is 30 days late). Second, payment history is taken into account in their models. If one has 10 cards with perfect 10 year payment histories, and one that has a 30 day late on it, one's score will be higher than someone with many lates. I am not sure how you think they are scoring individual months. They score one's aggregate behavior over time for things like payments. What they do not do is score trended data for balances and utilization.

Re: Anyone else think the algorithm is wrong???

Horseshoez — Sat, 23 Mar 2024 17:46:55 GMT

I get the feeling @AlgorithmIsWrong was just here for a fly-by rant and we won't be hearing from this individual again.

Re: Anyone else think the algorithm is wrong???

tricie17 — Sat, 23 Mar 2024 19:31:57 GMT

I for one believes that the system is flawed. The guidance given to the Bureaus I doubt is read correctly. I mean after all if I get a credit card, my longevity goes down but if I don't get a credit card, I do not have enough credit-who made this mess up. The Bureaus are so busy creating their own portfolio that your credit gets information that do not even belong to you or it is wrong altogether. FICO only gets to accept the flawed information from each of the three Bureaus which does not make things better. CFPB is so overloaded with crap messes that is placed incorrectly on reports and the investigation starts with (sent back to) the company reporting the wrong information in the first place. I could go on and on. It is all a catch 22 program. Oh and I forgot to mention that there are so many secondary companies with access to your credit that it becomes a markery.

Re: Anyone else think the algorithm is wrong???

Horseshoez — Sun, 24 Mar 2024 02:05:46 GMT

@tricie17 wrote:
I for one believes that the system is flawed. The guidance given to the Bureaus I doubt is read correctly. I mean after all if I get a credit card, my longevity goes down but if I don't get a credit card, I do not have enough credit-who made this mess up. The Bureaus are so busy creating their own portfolio that your credit gets information that do not even belong to you or it is wrong altogether. FICO only gets to accept the flawed information from each of the three Bureaus which does not make things better. CFPB is so overloaded with crap messes that is placed incorrectly on reports and the investigation starts with (sent back to) the company reporting the wrong information in the first place. I could go on and on. It is all a catch 22 program. Oh and I forgot to mention that there are so many secondary companies with access to your credit that it becomes a markery.

I have a tin-foil hat I'd like to sell you.

Re: Anyone else think the algorithm is wrong???

FicoMike0 — Sun, 24 Mar 2024 20:26:47 GMT

My sympathy to the op and his pain. That said, since the whole purpose of fico score is to predict being late, is anyone really surprised that it responds to (drumroll) being late?

Re: Anyone else think the algorithm is wrong???

CorpCrMgr1 — Mon, 25 Mar 2024 19:23:53 GMT

@tricie17 well said!