Every week we ask the same Marion question twice, seven days apart, and report what changed. This morning the query returns 191 active listings; a week ago it returned 187. And the most interesting number in it is one that is not a price at all.
One record went from $20,000 to $209,000, and a date field is how we caught it
| What the record carries | October 28 | November 4 |
|---|---|---|
| Asking price | $20,000 | $209,000 |
| Property-type code | 8 (land) | 6 (house) |
| Bedrooms | — | 3 |
| Floor area | — | 1,216 sq ft |
| Year built | — | 2026 |
| Lot size | 4,791 sq ft | 4,791 sq ft |
| Badge | — | New Construction |
| Days on market | 44 | 7 |
One record, the same record identifier in both pulls, at 907 W White St. Every line is a field the record carries; none is our interpretation. The lot size is identical, which is what makes it the same piece of ground.
Read the last line first. The days-on-market count went 44 to 7 — 37 days backwards across a 7-day gap. A listing cannot get younger. That is the tell, and it is the only field on the record that could not possibly be a market event.
Everything else follows from it. A week ago the record carried the land code and no bedroom count, no floor area and no year built. This morning it carries the house code, three bedrooms, 1,216 square feet, a year built of 2026 and a new-construction badge. The lot size is identical in both weeks, which is how we know it is the same piece of ground rather than two different listings that happen to share a street number.
A separate source confirms it by a completely different mechanism. Our four-county listing feed, pulled for November 2’s Market Monday over an overlapping week, carried this address as Land at $20,000 on October 26 and as Residential at $209,000 on November 2 — under a different listing identifier. So one source reused the record and the other issued a new one, and both describe the same change. We do not combine any count from the two sources; the second one is here only to show that the event is real and not an artefact of one feed.
That also sharpens something November 2’s post said. It described this address as a property that changed listing record without going anywhere. That was true and incomplete — it did not go anywhere, but it stopped being a lot and started being a house, and the post had no way to see that from one feed.
What we are not going to do is tell you what happened on the ground. Whether something was built, when, by whom, or whether the earlier listing was simply replaced is not in any field we hold. Every figure in the table above is a record field. And this record is excluded from every price-move figure in this post, because a $189,000 move on a record that changed what it is describing is not a price move.
And a second record aged 15 days in 7
10570 Steele Acres Ln held its asking price at $86,000 and kept its badges, but its property-type code moved from 6 to 1 and its days-on-market count went from 45 to 60 — that is 15 days across a 7-day window, 8 more than it allows. Another record event, a quieter one, caught by the same field.
Worth being precise about this one: yesterday’s Lake of Egypt post reported the same row, jumping the same way across its own seven days. The two windows overlap by six days, and the counts line up day for day, so this is one event seen through two windows, not two anomalies. Counting the same thing twice is the easiest way to manufacture a pattern out of nothing.
Across the 179 records on this query both weeks, the days-on-market count advanced by exactly seven days on 163 of them. Two records did not, and those are the two records above. Another 14 carry no days-on-market figure at all in one or both weeks.
Which is why the house median gets its sensitivity printed
| Houses on this query | October 28 | November 4 | Move |
|---|---|---|---|
| How many | 111 | 115 | +4 |
| Median asking price | $244,900 | $239,900 | −$5,000 |
| Average asking price | $308,246 | $301,669 | −$6,577 |
| Median asking price, without the reclassified record | $244,900 | $242,450 | −$2,450 |
| Median days on market | 54 | 51 | −3 |
The days-on-market medians are over the houses that carry the figure — 104 of 111, then 109 of 115. A missing figure is not a zero and not an average; it is a smaller denominator, and it is printed.
The median asking price on the houses fell $5,000 and the average fell $6,577. Both fell, which is the only reason we are willing to call it a fall. But about half of the median move is that one reclassified record joining the house set at $209,000, below the median and pulling it down. Take that record out and the median fall is $2,450.
We print both figures rather than the bigger one. A statistic that halves when a single record is removed is a statistic you should see the sensitivity of, and a reader who is told only the first number has been given a worse answer than one who is told both.
What is left after that is turnover, the same thing a flat count hides anywhere.
| Houses | Median asking | |
|---|---|---|
| Came on the query | 11 | $209,900 |
| Came off the query | 7 | $244,900 |
| Difference | +4 | −$35,000 |
Coming off this query is not a sale. A record can go under contract, expire, be withdrawn or be cancelled, and this query asks only for active listings, so we cannot tell those apart and do not try.
Eleven houses came on the query and seven came off, and the ones arriving ask a median $35,000 less than the ones leaving. That is turnover rather than repricing, and none of it is a seller changing a number.
9 of the 10 price moves were reductions
Of the 179 records on this query both weeks, 10 changed price. Set the reclassified record aside and all 9 of the rest are reductions, at a median of $9,000 and running from $2,500 to $30,900. Not one asking price on this query went up for a reason that was about price.
The land side did nothing at all: 69 land records a week ago and 67 now, at a median of $85,000 both weeks — the same figure, not a rounded one.
And the arrivals on this query really are arrivals
Twelve records joined the query this week. Eleven of the twelve carry a week or less on market; the last one carries 38 days. So when something appears on this query it has very nearly always just come on the market.
That is worth saying because it is not true of every way of watching a market. A query scoped by postal code asks the market a question; a query scoped by words in a listing’s write-up asks the words a question, and things join and leave it for reasons that have nothing to do with houses moving. Which instrument you are reading changes what an arrival means, and that is a point about method rather than about this week.
The closed-sale series has not moved, and we checked rather than assumed
September 2026 is still the newest published closed month for this postal code. We also compared all 60 monthly observations against the copy we pulled on October 28 — every median sale price, every sold count, every days-on-market figure. Not one value changed. That check is only possible because we kept the earlier pull, and it is the difference between reporting that a series has not moved and assuming it.
No closed-sale figure appears anywhere in this post, because none of them is new.
What this post is not saying
Nothing here says anything sold. All 191 of these listings are active and the eight records that left are not described as sold, pending, contingent, withdrawn or expired, because we cannot see which. Nothing here is a forecast, and no cause is given for any price move. Nothing here subtracts an asking median from a closed median — different sets of houses. No property-type code beyond the two we tested is given a name: nine records sit in five other codes and are described by their own fields. One record carries no address at all — the payload holds a placeholder where the street line should be, the same record in both weeks — so it cannot join an address-level comparison, and it is counted everywhere else. There is no price per square foot and no property-tax figure anywhere on this page, and no interest rate or payment figure either.
And one honesty note about the universe itself. This query is scoped by postal code, and seven records in it carry a postal field that is either a ZIP+4 variant or a neighbouring code. They are in the comparison because they were in last week’s pull too, and dropping them this week would have made a difference of question look like a difference of market. So this is what the query returns, not exactly one postal code.
If you are pricing a house in Marion
The useful thing from this week is not the median. It is that two records on this query changed in ways that no price figure would show you, and that a figure built from the whole list halves when one of them is removed. The number that helps you is narrower: what is actually listed at your size and in your band this week, with the record events taken out. Somebody has to build that and keep it current. The number is 618.925.8654, or ask any of our brokers.
💬 Greg Holthaus on reading a listing record
“A record went from $20,000 to $209,000 here in seven days, and nobody raised a price. The thing that gave it away was a date running backwards, which is the sort of detail that never makes it into a headline about a market. What I take from it is narrow and it is the same thing I would tell a seller: a number built off a whole list is only as good as the records in the list, and this week two of them were describing something different than they were describing a week ago. Ask for the short list of what is really comparable to your house. That one is worth building properly.”
— Greg Holthaus, Owner & Designated Managing Broker, Southern Illinois Realty Experts
About the numbers: Active-listing figures come from two pulls of one query of local MLS listing data, October 28 and November 4, seven days apart: 187 active records then 191, every record active in both pulls. The query form was re-established by membership before anything was compared, because the earlier module recorded a description of its query rather than the query itself: the form used here is the one that returns for-sale-by-owner and new-construction badged records, which the narrower form omits, and both are present in both pulls. The postal-code filter mattered more than the address of the query did — an exact-match filter returns seven fewer records than the comparable form and would have excluded rows the earlier pull contains, so no post-filter is applied and the universe is described as what the query returns. Property-type codes are tested against the rows, never inherited: every record on the house code carries both a bedroom count and a floor area (111 of 111, then 115 of 115), while records on the land code almost never do and mostly carry a lot size. No other code is given a type name. Week-over-week figures are computed over the 179 records present in both pulls, not over either pull as a whole. Arrivals and departures are matched on street line plus town and reconciled against each record’s own identifier, which agreed both ways this week at 12 = (12 − 0) + 0, and the same on departures; no address was rewritten in place, nothing was relisted under a new identifier, and no address carries two records. The reclassified record is excluded from every price-move figure and its effect on the house median is printed separately. The second source is used only to corroborate that one event and no count from it is combined with these — the two sources are known not to overlap fully. Closed-sale figures were re-pulled this morning with the three metrics matched by label and never by position; all 60 monthly observations were compared against the October 28 copy and none changed, September 2026 is still the newest published month, and no closed-sale figure is quoted anywhere in this post. A missing field is treated as a smaller denominator and never as a zero, and the denominators are printed. No listing is described as sold; a record leaving this query can have gone under contract, expired, been withdrawn or been cancelled, and we cannot tell those apart. No price per square foot is computed anywhere and no property-tax figure appears. Asking prices are not sale prices, and nothing here forecasts whether prices will rise or fall. Listings change daily; confirm anything you intend to rely on. Source note: figures on this page include local MLS data as published by Redfin (redfin.com).
