Well, I dont see how you can call it accurate, when it seems to have gotten about half of the members that have tried it within its 50 point range. 50 points is a lot, lol. If it tells me I'm between a 700 and 750, theres a big difference between those two scores.
[Edit] FWIW, it estimated me at 632 to 682. So it did hit the range. It didnt ask for hardly any information, so its basically taking an educated shot in the dark, hence the 50 point range. It seems to be no more or less accurate than all the other estimators.
Starting Score: EQ: 665 - TU: 687
I used this estimator from the original post and got a range of 781 to 831. I also used the estimator on this site, producing a range of 750 to 800. Actual FICO scores as of this morning are EQ - 804 and TU - 821. Thus, the first was more accurate, but it's not hard to hit a 50 point range!
Starting Score: TU: 821/EQ: 800
It's my experience that estimators included in purchased FICO products are reasonably accurate. I find the various ranges in given situations just right when taken into consideration factors such as reverse engineering and the seller's obligation to the customer. The customer of course should not take an estimator for more than what it is. If an estimate is based on one primary variable changing along with secondary variables such as aging an rebucketing, it's in the customer's own interest to understand this very frame that comes with it and not be carried away in wishful dreams as tempting as that may be. 
Another example of a FICO product-related estimate is how long a myFICO Score Watch subscriber should have to wait until the arrival of the score update following a key change to your credit reports. When DW recently earned a hard inquiry on EQ from applying for a credit card, myFICO SW sent her a credit alert within 24 hours. I find that reasonable. I had hoped I would have received an alert myself within such a window stemming from a new TL's first statement cutoff, but that wasn't to be. 
@wednesdayaddamsd wrote:I don't know why people get so worked up over estimators.
Neither do I. 
But when your landline phone connection has been out of service for days, and you wake up four a'clock on a Sunday morning and your Internet connection has gone down the tubes as well, what's a man to do? DW likely wouldn't be in the mood to serve you coffee and eggs in bed!
So I spent the quiet hours by daybreak writing my own estimator. I came up with something that'll suffice for now. Includes handling of AAoA (individual and overall), util (individual and overall), inquiries. Doesn't do derogs (don't have any), mix of credit, rebucketing, logarithmic algorithms and that sort of stuff.
It is of course based on the experience of DW and me. So how's that for verification that it works? 