How many comps an ARV needs
People ask how many comps an ARV needs as if the answer were a moral minimum. The report’s count math is smaller and stricter than that. The count portion of the confidence score is the number of comps times 6, capped at 30. One comp is 6 points. Two is 12. Three is 18. Four is 24. Five is 30. Six is still 30. A seventh does not add count points either. Five sales fill the bucket. More sales can still change the value. They do not make the count portion larger. This page is that table, plus one invented list that shows the median move after the count has something to say. The recency guide is about how old those sales are. Age and count are different scores.
What the cap does and does not do
The cap is only on the count portion. Similarity, the share of sales inside six months, and how tightly the prices cluster are separate buckets, up to 25, 25, and 20 points. A sixth sale that is recent and close to the others can still raise those buckets. A sixth sale that is a wild price can lower the consistency bucket even while the count stays at 30. Adding comps to “get to five” by including a house you would not buy is a way to fill a 30 point bucket and damage the value the bucket is scoring. Membership first, then the count.
A single comp is a special case of the median: the median is that price. The count portion is 6, not 30, and not zero. The formula does not refuse to print an ARV. It prints one and a score that has already given away 24 of the 30 count points. If you are staring at one perfect sale and calling the ARV done, you are using the price and ignoring the 6. Write both down. The throw-out guide is what to do when the next sale you find is the one that does not belong. Leaving it out on purpose is a count of the sales that remain, not a failure to reach a magic number.
Adding sales changes which prices are in the middle
Start with two invented sales, $180,000 and $220,000. An even count averages the two central prices. There are only two, so the plain median is $200,000. Add $190,000. The sorted list is $180,000, $190,000, $220,000. The middle is $190,000. One additional sale moved the hand median by $10,000, and the count portion went from 12 points to 18. Add a fourth sale at $300,000, a number chosen to be obviously unlike. The sorted list is $180,000, $190,000, $220,000, $300,000. The two central prices are $190,000 and $220,000. Their average is $205,000. The outrageous sale did not become the median, but it pushed $220,000 into the central pair, so the median rose from $190,000 to $205,000. Count points went from 18 to 24. The points went up. The value moved toward a sale you do not believe.
That is the trap in “I need more comps.” The fourth price increased the count score and worsened the median, because an even count lets a high sale enter the central pair without becoming the middle by itself. A fifth sale at $192,000 would sort to $180,000, $190,000, $192,000, $220,000, $300,000, and the middle would be $192,000. The count portion would finally read 30, and the median would have climbed back down only because you added a sale near the others. The $300,000 sale is still in the list, still hurting consistency, and still not the median. Five was not a blessing. It was the cap. The MAO calculator will take whichever of these you decide is the ARV. It will not know that $300,000 is still hiding in the set.
A 70 percent offer on the median you kept
Suppose you throw out the $300,000 sale and keep $180,000, $190,000, and $192,000. The plain median is $190,000. Count points on three comps are 18, not 30. An invented rehab of $32,000 produces a 70 percent offer of $133,000 minus $32,000, which is $101,000. You can see that on the ARV and 70% rule calculator by typing the three prices you kept and leaving the $300,000 sale out of the form. If you type it anyway, four prices change the median, as the walk above already showed. The calculator does not cap confidence. It only medians what you give it. The report is the place the count cap lives.
Do not hunt a sixth comp to satisfy the score. The score stopped caring at five. Hunt a sale that matches, and stop when the next candidate is a different house. These prices are invented so the central-pair arithmetic is visible. They are not a neighborhood.
Questions
Do six comps score more count points than five?
No. The count portion is the number of comps times 6, capped at 30. Five comps are 30. Six comps are still 30. A sixth sale can still change the median. It just does not add count points.
Is a one-comp median useless?
It is the price you typed, and the count portion is 6 points out of 30. The other confidence buckets can still add points. A single sale is a thin set even when those other buckets look fine. Do not treat 6 count points as a full score.
run a full analysis free when the median you can do by hand is not enough. A full analysis weights the comps, adjusts beds and baths, and prices the repair tiers. The sample figures there are invented, the same way every dollar on this page is invented.