Scoring a perfect 300

[Anonymous]

Although it is rare to score a perfect 850, there are probably thousands of people who have done it,

The flip side (perfect badness = 300) is something I am guessing, however, that no one has ever attained.  Does anyone know of a documented case of this?

In fact, I wouldn't be surprised if nobody has come even close.  Scoring in the top 50 (the "800 club") is not unusual at all.  The chart below suggests that 18% is the approximate number of consumers year by year who attain an 800.  But I don't know of any data that show the % for the bottom 50.  Even the chart below lumps all the bad scouts into a big 200-point range (300-499).

http://www.fico.com/en/blogs/risk-compliance/fico-score-distribution-remains-mixed/

Anyone have a guess as to what the lowest anyone has ever scored?  My thought is that it's probably in the upper 300s, but if someone produced hard evidence that the worst scorer of all time was (say) 413, it wouldn't surprise me.

PS.  Note: I am confining this to FICO algorithms where the lowest theoretical score is 300.  Various specialty metrics can go lower. (Auto, Bankcard enhanced, etc. I think go down to 250.)

[Revelate]

Couldn't hit the minimum number on any previous generation score similarly to how you couldn't hit an 850 either.

FICO 8 not sure.  Guess could trawl the RYC forum and see but don't know that anyone's ever really tracked that.

[Thomas_Thumb]

---

@Anonymous wrote:

Although it is rare to score a perfect 850, there are probably thousands of people who have done it,

The flip side (perfect badness = 300) is something I am guessing, however, that no one has ever attained.  Does anyone know of a documented case of this?

In fact, I wouldn't be surprised if nobody has come even close.  Scoring in the top 50 (the "800 club") is not unusual at all.  The chart below suggests that 18% is the approximate number of consumers year by year who attain an 800.  But I don't know of any data that show the % for the bottom 50.  Even the chart below lumps all the bad scouts into a big 200-point range (300-499).

http://www.fico.com/en/blogs/risk-compliance/fico-score-distribution-remains-mixed/

Anyone have a guess as to what the lowest anyone has ever scored?  My thought is that it's probably in the upper 300s, but if someone produced hard evidence that the worst scorer of all time was (say) 413, it wouldn't surprise me.

PS.  Note: I am confining this to FICO algorithms where the lowest theoretical score is 300.  Various specialty metrics can go lower. (Auto, Bankcard enhanced, etc. I think go down to 250.)

---

© 2012 Experian Information Solutions. All rights reserved.

© 2012 FICO

Minimum Experian/FICO® Risk Model v8   316

MaximumExperian/FICO® Risk Model v8   850

File date: April 2012

EXF5G-C

[Anonymous]

Thanks, TT.  Where is that from?  Is there a link for it?

I am gussing this means that, during 2012, someone scored a 312 using the EX data?

[Thomas_Thumb]

Yes,The information is from a 2012 Experian document. The minimum reported Fico 8 score may be from 2012 (or earlier).

Unfortunately, I was unable to copy the link over to this post. However, you can find the link and other interesting ones in the original post under "the many flavors of FICO" topic on this web site. 

[Anonymous]

Hey TT!  Thanks. 

Jello77's post "The Many Flavors of FICO" is invaluable and it it was one of the first things I ever read on the MF forum.  (As you can see it is a summary of a detailed discussion begun in 2012.  I even contributed to it, far down the thread and a while back.)

Jello77's idea of "real world ranges" refers to the minimum and maximum values that could really ever be achieved by that model with that data.  That's different from the minimum and maximum that have been achieved by some definite person.  "Real" as Jello77 is using it is being contrasted with the official but inaccurate (i.e. unreal) range given by FICO's public relations people for many years (300-850).  (If you read his preface and also slowly read through the thread, this becomes clear.)

Thus, for example, J77 gives 334-818 as the real world range of FICO 04 as implemented by Equifax.  That doesn't mean that people could have scored higher than 818, but for some enigmatic reason the highest some guy ever scored was 818.   It means that, as Equifax implemented it in reality, the maximum score that is theoretically possible with that model is 818. 

Likewise, the Experian FICO 8 "real world range" that Jello77 gives is 316 - 850.  That doesn't mean that 316 is the lowest score ever recorded by some real profile.  It means that the way that the model is implemented, 316 is the lowest possible value.

The Experian link supports this, because there is a gigantic jump from the lowest score theoretically possible (316) to the next number (465).  Contrast that with the top of the scale, where we see much more of a finely graded continuum as we move down.  We know that real people have really scored 850, and 849, and 848, etc.  In the case of the bottom up, we do not see a similar gradation, but rather a gigantic leap from the lowest the model could score to the just under 1%.

The Experian chart does answer one important question, which is that there are definitely people scoring less than 465.  (At the time that chart was created, about 1% of consumers were scoring less than 465.)  What it doesn't tells us is the lowest score ever recorded, which might be (say) 380, or even higher, like 430.

[Revelate]

---

@Anonymous wrote:

Hey TT!  Thanks. 

Jello77's post "The Many Flavors of FICO" is invaluable and it it was one of the first things I ever read on the MF forum.  (As you can see it is a summary of a detailed discussion begun in 2012.  I even contributed to it, far down the thread and a while back.)

Jello77's idea of "real world ranges" refers to the minimum and maximum values that could really ever be achieved by that model with that data.  That's different from the minimum and maximum that have been achieved by some definite person.  "Real" as Jello77 is using it is being contrasted with the official but inaccurate (i.e. unreal) range given by FICO's public relations people for many years (300-850).  (If you read his preface and also slowly read through the thread, this becomes clear.)

Thus, for example, J77 gives 334-818 as the real world range of FICO 04 as implemented by Equifax.  That doesn't mean that people could have scored higher than 818, but for some enigmatic reason the highest some guy ever scored was 818.   It means that, as Equifax implemented it in reality, the maximum score that is theoretically possible with that model is 818. 

Likewise, the Experian FICO 8 "real world range" that Jello77 gives is 316 - 850.  That doesn't mean that 316 is the lowest score ever recorded by some real profile.  It means that the way that the model is implemented, 316 is the lowest possible value.

The Experian link supports this, because there is a gigantic jump from the lowest score theoretically possible (316) to the next number (465).  Contrast that with the top of the scale, where we see much more of a finely graded continuum as we move down.  We know that real people have really scored 850, and 849, and 848, etc.  In the case of the bottom up, we do not see a similar gradation, but rather a gigantic leap from the lowest the model could score to the just under 1%.

The Experian chart does answer one important question, which is that there are definitely people scoring less than 465.  (At the time that chart was created, about 1% of consumers were scoring less than 465.)  What it doesn't tells us is the lowest score ever recorded, which might be (say) 380, or even higher, like 430.

---

That's not correct

The maximum attainable score on EQ Beacon 5.0 was 818.  It simply was not possible to get higher than that under the model.  818 was "perfect credit" with EQ Beacon 5.0 aka EQ 04, same as 850 is today on EQ Beacon 9.0 aka EQ 8.  The same holds true for the rest of them.

[Anonymous]

---

@Revelate wrote:

---

@Anonymous wrote:

Hey TT!  Thanks. 

Jello77's post "The Many Flavors of FICO" is invaluable and it it was one of the first things I ever read on the MF forum.  (As you can see it is a summary of a detailed discussion begun in 2012.  I even contributed to it, far down the thread and a while back.)

Jello77's idea of "real world ranges" refers to the minimum and maximum values that could really ever be achieved by that model with that data.  That's different from the minimum and maximum that have been achieved by some definite person.  "Real" as Jello77 is using it is being contrasted with the official but inaccurate (i.e. unreal) range given by FICO's public relations people for many years (300-850).  (If you read his preface and also slowly read through the thread, this becomes clear.)

Thus, for example, J77 gives 334-818 as the real world range of FICO 04 as implemented by Equifax.  That doesn't mean that people could have scored higher than 818, but for some enigmatic reason the highest some guy ever scored was 818.   It means that, as Equifax implemented it in reality, the maximum score that is theoretically possible with that model is 818. 

Likewise, the Experian FICO 8 "real world range" that Jello77 gives is 316 - 850.  That doesn't mean that 316 is the lowest score ever recorded by some real profile.  It means that the way that the model is implemented, 316 is the lowest possible value.

The Experian link supports this, because there is a gigantic jump from the lowest score theoretically possible (316) to the next number (465).  Contrast that with the top of the scale, where we see much more of a finely graded continuum as we move down.  We know that real people have really scored 850, and 849, and 848, etc.  In the case of the bottom up, we do not see a similar gradation, but rather a gigantic leap from the lowest the model could score to the just under 1%.

The Experian chart does answer one important question, which is that there are definitely people scoring less than 465.  (At the time that chart was created, about 1% of consumers were scoring less than 465.)  What it doesn't tells us is the lowest score ever recorded, which might be (say) 380, or even higher, like 430.

---

That's not correct

The maximum attainable score on EQ Beacon 5.0 was 818.  It simply was not possible to get higher than that under the model.  818 was "perfect credit" with EQ Beacon 5.0 aka EQ 04, same as 850 is today on EQ Beacon 9.0 aka EQ 8.  The same holds true for the rest of them.

---

Hi Revelate.  You're misunderstanding me.  I agree that the maximum attainable score was 818 and that it was simply not possible to get higher under that model.  That's why I say that the upper bound of 818 doesn't mean simply that this is the maximum that anybody has scored so far,.  Rather, I say, it means that the maximum score that is theoretically possible with that model is 818.

My post is explaining to TT that the ranges he is finding in Jello77's post are descriptions about what is the maximum theoretically obtainable and the minimum theoretically attainable (per model and CRA).  We can't deduce from those the answer to my original question, which is: what is the lowest that some real person with a real credit profile every scored? 

[Revelate]

---

@Anonymous wrote:

---

@Revelate wrote:

---

@Anonymous wrote:

Hey TT!  Thanks. 

Jello77's post "The Many Flavors of FICO" is invaluable and it it was one of the first things I ever read on the MF forum.  (As you can see it is a summary of a detailed discussion begun in 2012.  I even contributed to it, far down the thread and a while back.)

Jello77's idea of "real world ranges" refers to the minimum and maximum values that could really ever be achieved by that model with that data.  That's different from the minimum and maximum that have been achieved by some definite person.  "Real" as Jello77 is using it is being contrasted with the official but inaccurate (i.e. unreal) range given by FICO's public relations people for many years (300-850).  (If you read his preface and also slowly read through the thread, this becomes clear.)

Thus, for example, J77 gives 334-818 as the real world range of FICO 04 as implemented by Equifax.  That doesn't mean that people could have scored higher than 818, but for some enigmatic reason the highest some guy ever scored was 818.   It means that, as Equifax implemented it in reality, the maximum score that is theoretically possible with that model is 818. 

Likewise, the Experian FICO 8 "real world range" that Jello77 gives is 316 - 850.  That doesn't mean that 316 is the lowest score ever recorded by some real profile.  It means that the way that the model is implemented, 316 is the lowest possible value.

The Experian link supports this, because there is a gigantic jump from the lowest score theoretically possible (316) to the next number (465).  Contrast that with the top of the scale, where we see much more of a finely graded continuum as we move down.  We know that real people have really scored 850, and 849, and 848, etc.  In the case of the bottom up, we do not see a similar gradation, but rather a gigantic leap from the lowest the model could score to the just under 1%.

The Experian chart does answer one important question, which is that there are definitely people scoring less than 465.  (At the time that chart was created, about 1% of consumers were scoring less than 465.)  What it doesn't tells us is the lowest score ever recorded, which might be (say) 380, or even higher, like 430.

---

That's not correct

The maximum attainable score on EQ Beacon 5.0 was 818.  It simply was not possible to get higher than that under the model.  818 was "perfect credit" with EQ Beacon 5.0 aka EQ 04, same as 850 is today on EQ Beacon 9.0 aka EQ 8.  The same holds true for the rest of them.

---

Hi Revelate.  You're misunderstanding me.  I agree that the maximum attainable score was 818 and that it was simply not possible to get higher under that model.  That's why I say that the upper bound of 818 doesn't mean simply that this is the maximum that anybody has scored so far,.  Rather, I say, it means that the maximum score that is theoretically possible with that model is 818.

My post is explaining to TT that the ranges he is finding in Jello77's post are descriptions about what is the maximum theoretically obtainable and the minimum theoretically attainable (per model and CRA).  We can't deduce from those the answer to my original question, which is: what is the lowest that some real person with a real credit profile every scored? 

---

Hah apparently I've lost control of my command of English.  Mea culpa, you're right, I misread it.

That's what I'd thought you'd asked in the OP then got confused, oh well, been one of those days months.

[Anonymous]

OK, that strikethru really made me smile.  I certainly know how days months can be.