Blog
Housing market predictions on YouTube, checked
By Callsight · Published Oct 10, 2026 · Updated Oct 10, 2026
Figures as of Nov 4, 2026. They change as new windows settle.
As of Nov 4, 2026, Callsight counts 94 housing calls that three finance YouTube channels made in their own words: 64 on house prices, 22 on a housing crash and 8 on mortgage rates; Graham Stephan made 76 of them. House-price calls are judged on whether price growth sped up or slowed down, not on whether prices rose: of the 12 calls for a rise judged Incorrect at one year, 11 ended a year in which prices went up. Of the 20 crash calls that could be judged, 18 said no crash would come; at one year those stand at 12 Correct, 0 Incorrect, 2 Partly correct and 4 Pending.
What did finance YouTubers predict about the housing market?
The 94 counted calls come from three channels and were made between June 2017 and August 2026. Calls by Joseph Carlson on these subjects are listed, but none is counted. Graham Stephan made 76 of the 94: 54 on house prices, 17 on a housing crash and 5 on mortgage rates. The article on what Graham Stephan said about housing goes through his calls subject by subject.
| Channel | House prices | Housing crash | Mortgage rates | All |
|---|---|---|---|---|
| Graham Stephan | 54 | 17 | 5 | 76 |
| Coin Bureau | 4 | 3 | 3 | 10 |
| Andrei Jikh | 6 | 2 | 0 | 8 |
On house prices, 19 were recorded as a rise, 18 as a fall and 27 with no direction that could be judged; on a crash, 18 of the 20 that could be judged said none would come; on mortgage rates, 5 called for higher rates and 1 for lower.
Another 139 calls on these subjects are listed and never counted: 81 forecasts read out from someone else, each shown with its source, and 58 in the channels’ own words that the record holds back, 54 of them because their direction could not be read with confidence or because a review questioned how they were recorded. Each call’s page gives the reason. Two more counted calls are on the median house price, a separate series, and are not in these figures.
Did YouTubers predict a housing market crash?
Callsight makes no forecast of a crash. Of the 22 counted calls the tracked channels made on a US housing crash, 20 could be judged: 18 said no crash would come and 2 said one would. At one year, the calls that said none would come stand at 12 Correct, 0 Incorrect, 2 Partly correct and 4 Pending; the two that said one would come are both Incorrect. Neither named a time; a call with no deadline is headlined at one year. Graham Stephan made 16 of the 18 that said none would come. Two more were about housing in another country and are Not verifiable.
A crash here is a fall of at least 10% in the Case-Shiller index from its highest figure since the call, inside the window; a fall of at least 5% is Partly correct. A call that no crash will come is Correct while the fall stays under 5%. The index is not seasonally adjusted, and prices usually dip in winter, so part of a fall from a summer high can be the season. The method stores no base rate for a crash call, so this section gives counts only.
One call that said none would come, from a channel other than Graham Stephan’s, chosen by the rule at the end of this article:
“But I wouldn't expect a crash like some people predict because there's a lot of people waiting on the sidelines to buy a a”Reading when said
312.95
Oct 9, 2023
Reading at close
324.22
Oct 8, 2024
Change
+11.27
The house price index (CSUSHPINSA, first releases) stood at 313.0 in October 2023, the figure at the start of the window. Between then and October 2024, its deepest fall from a running peak was −0.8%, in January 2024. A fall of at least 10.0% counts as a housing crash, and at least 5.0% as partial. The call was that no housing crash would come.
30 days
Correct
90 days
Correct
180 days
Correct
1 year
Correct
No deadline was stated; the headline is the one-year window, where it is Correct. Its window at 3 years is not yet decided, and its window at 5 years is still open.
Every call on the subject is on the housing crash page. Callsight’s article on who predicted the recession sets these calls beside the recession and stock-market crash calls.
What did finance YouTubers predict for mortgage rates?
Eight counted calls are about mortgage rates: 5 from Graham Stephan and 3 from Coin Bureau; 5 called for higher rates and 1 for lower. They are judged on Freddie Mac’s weekly average 30-year fixed rate (MORTGAGE30US), as first published: the last weekly figure on or before the day of the call against the last one inside the window. A change smaller than 0.5 points is Within band, and a call that names a level is Correct if any weekly figure inside the window reaches it. At one year they stand at 2 Correct, 1 Incorrect, 1 Within band, 2 Pending and 2 Not verifiable. Three decided calls are too few to give as a rate. Every call is on the mortgage rates page.
One of them, chosen by the rule at the end of this article:
“it's no surprise that rates will probably go a lot higher as the Federal Reserve continually raises rates for the foreseeable future”Reading when said
4.16%
Mar 23, 2022
Reading at close
6.42%
Mar 23, 2023
Change
+2.26 pp
Mortgage rates stood at 4.16% at the start of the window and 6.42% at Mar 23, 2023, a change of +2.26 pp (MORTGAGE30US, first releases). The call was that it would rise. Moves smaller than 0.5 pp are treated as no change.
30 days
Correct
90 days
Correct
180 days
Correct
1 year
Correct
No deadline was stated; the headline is the one-year window, where it is Correct. At 3 years it is Correct; its window at 5 years is still open.
Why can a call that prices will rise be Incorrect when prices rose?
House prices are checked on the S&P Cotality Case-Shiller U.S. National Home Price Index (CSUSHPINSA on FRED), as first published, and the rule reads that index’s year-on-year change: how much higher or lower prices were than twelve months earlier. The rate for the month the call was made is compared with the rate for the last month inside the window. A call that prices will rise is Correct if the second rate is at least 2.0 points above the first, and Incorrect if it is at least 2.0 points below; a call that prices will fall is judged the other way round. A smaller change is Within band, which is neither Correct nor Incorrect. No verdict is given until the last month’s figure is published.
So the rule asks whether price growth sped up or slowed down. Suppose prices were rising 10% a year when a call said they would rise, and 4% a year twelve months later. Prices went up, and the call is Incorrect, because the pace fell by 6 points. A call made the same day that prices would fall is Correct, although prices never fell. The rule in full is on the methodology page.
Did the house price predictions come true?
On house prices the record’s verdicts answer whether the pace of price growth moved the way a call said, not whether prices went up or down, and the two answers often differ. Of the 17 calls recorded as a rise that named no figure and are settled at one year, the index ended the window higher than in the month of the call in 16. The record has 12 of them Incorrect, 11 of those in a year when prices rose. Of the 15 such calls recorded as a fall, the index ended lower in 3. The record has 11 of them Correct, 8 of those in a year when prices rose. These are counts, not rates: the record keeps a base rate for the pace and none for the level.
Some calls said both at once: that prices would keep rising, but more slowly. In the rule’s terms that is a fall in the growth rate. A call like that can be recorded as a rise instead, and then it is judged as a call that growth would speed up and is Incorrect when growth slows, as its words said it would. Each card shows the words beside the verdict; if you think a call was misread, the corrections page says how to report it.
The table shows which way each channel’s counted calls were recorded. It gives no verdicts: each call’s verdict is on its card, beside the words and the two rates it was read from, and every call is on the house prices page.
| Channel | Recorded as a rise | Recorded as a fall | No direction judged | All |
|---|---|---|---|---|
| Graham Stephan | 18 | 10 | 26 | 54 |
| Andrei Jikh | 1 | 5 | 0 | 6 |
| Coin Bureau | 0 | 3 | 1 | 4 |
One call whose words are about the pace itself, chosen by the rule at the end of this article:
“going forward i don't think that kind of growth is going to be sustainable and i think the market has already started to soften”Reading when said
20.03%
May 23, 2022
Reading at close
−0.27%
May 23, 2023
Change
−20.30 pp
House prices stood at 20.03% at the start of the window and −0.27% at May 23, 2023, a change of −20.30 pp (CSUSHPINSA, first releases). The call was that it would fall. Moves smaller than 2.0 pp are treated as no change.
30 days
Within band
90 days
Correct
180 days
Correct
1 year
Correct
No deadline was stated; the headline is the one-year window, where it is Correct. At 3 years it is Correct; its window at 5 years is still open. Over its first year the index itself also fell.
Why can’t many house price predictions be checked?
At one year, 26 of the 64 counted calls on house prices are Not verifiable, each with the reason on its card. Of these, 18 are about a local market, such as one city or one state, and a national index says little about one market. Four were recorded with a figure the year-on-year rate had already passed when the words were said, so the window could not test them. Four name no direction at all. The method prefers a missed verdict to a wrong one, and none of them is in the settled counts above.
How were these calls checked?
Each call is quoted word for word from the video and linked to the second it was said. It is checked at six windows, from 30 days to 5 years, and the one-year window is the headline. Economic figures are read as first published, so a later revision does not change a verdict. A forecast read out from someone else is listed with its source and never counted. The rules are the same for every channel, and the methodology page gives them in full.
How these examples were chosen
The cards were chosen on Oct 10, 2026 by a fixed rule, from counted calls with a Correct or Incorrect verdict at one year. On each subject the rule takes the call from the most widely watched video, passing over a call whose reading is being checked again, whose words give only a possibility or name no time, report someone else’s view or the general mood, or describe a crash abroad or worldwide. On house prices it also passes over a call whose words are not about the pace of price growth, which is what the rule measures, or whose recorded direction is not the one the words give. On a housing crash it takes a channel other than Graham Stephan, whose calls the recession article already shows. All three cards are Correct at one year, because on each subject the first call to pass the rule was Correct; each topic page gives every verdict: housing crash, mortgage rates and house prices.
Figures are computed by code from the same record as the rest of the site. This article was drafted with the help of AI tools and reviewed by a person before publication. It is not investment advice.