As of Nov 4, 2026, Callsight has judged 51 of Andrei Jikh's calls Correct or Incorrect at one year: 30 correct and 21 incorrect. On the 49 of them with a comparable base rate (how often the called move happened anyway), 57.1% were correct, against a base rate of 41.7%. Every call links to the second it was said.
Hit rate, 1 year
58.8%(51 calls)
30 correct · 21 incorrect (out of 51 checked calls) After 3 years: 58.8% (34 calls) · on the 33 of 34 calls that have a base rate: 57.6% correct, base rate 48.7%
Above the base rate
+15.4pts
On the 49 of 51 calls with a comparable base rate: 57.1% correct, against a base rate of 41.7%.
Return vs benchmark
−6.4pts
Average over 15 price calls with a benchmark. 95% range −31.4 pts to +18.6 pts.
Predictions tracked
232
1 still open · 55 relayed from someone else · 74 own calls not scored
AI Ask an AI about Andrei Jikh’s videos
An AI simulation that answers from Andrei Jikh’s own videos and shows the exact words it relied on.
Every direction call with both results, both directions, so neither is picked.
Still correct after 1 year
8 of 13
Fewer than 20 calls have both results, so a count is shown and no percentage. Over the same windows, 67.6% of 30-day moves in the same instruments were still in the same direction after 1 year (measured on 13 of 13 calls). Correct after 30 days, still open at 1 year: 1
30 days → 1 year
30 days / 1 year
Correct 1 year
Incorrect 1 year
Correct 30 days
Correct 1 year8
Incorrect 1 year5
Incorrect 30 days
Correct 1 year4
Incorrect 1 year5
Incorrect after 30 days, correct after 1 year: 4 of 9 · Methodology
Calls per month
The last 24 months
UpDownOther
Recent predictions
What was said on the left, what happened on the right.
PredictionSPYwill fallChecked at fixed windows, from 30 days to 5 years
Recorded as markets, a word for a group: this call is judged on SPY, the stand-in the method assigns to it.
What happenedafter 90 days
Correct so far · 90 days
Entry price (first close after the video)
$768.56
Aug 6, 2026
Price at close
$754.35
Nov 3, 2026
Total return, dividends included
−1.6%
SPY closed at $768.56 on Aug 6, 2026. On Nov 3, 2026 it closed at $754.35, a total return of −1.6% (dividends included). The call was that it would fall.
“So basically what we know is that every single time the yen got stronger really fast, it meant that markets somewhere in the world were starting to break.”
PredictionGLDwill riseChecked at fixed windows, from 30 days to 5 years
What happenedafter 90 days
Correct so far · 90 days
Entry price (first close after the video)
$368.38
Jun 30, 2026
Price at close
$377.91
Sep 28, 2026
Total return, dividends included
+2.6%
S&P 500 +2.8%
GLD closed at $368.38 on Jun 30, 2026. On Sep 28, 2026 it closed at $377.91, a total return of +2.6% (dividends included). SPY's total return over the same window was +2.8%. The call was that it would rise.
“It looks like gold is going through a very important level on the technical charts and we could see gold go down to the $3,000 level potentially in the next year or two or maybe three before we see a shift from the stock market and into the commodities via this super cycle theory.”
“And that pattern is what I think will most likely happen with AI. I think it's the initial investors, the people buying at 1.75 trillion dollar valuations, the passive funds that are forced in by these rule changes, they're”
Watch at 31:34Your 401K Is Their Exit Strategy (SpaceX, Anthropic, OpenAI)
PredictionBubble burst— not judged: the rule reads only the US stock market in generalChecked at fixed windows, from 30 days to 5 years
What happenedafter 1 year
Not verifiable
Not verifiable: subject not covered. The call names a subject other than the US stock market as a whole; this rule reads SPY only.
“And higher yields are the one thing that could pop everything we've talked about, like all the paper profits and the circular spending and the $150 billion in corporate bonds that all these companies issued to fund their AI spending, right?”
Watch at 29:08Your 401K Is Their Exit Strategy (SpaceX, Anthropic, OpenAI)
PredictionBubble burst— not judged: the rule reads only the US stock market in generalChecked at fixed windows, from 30 days to 5 years
What happenedafter 1 year
Not verifiable
Not verifiable: subject not covered. The call names a subject other than the US stock market as a whole; this rule reads SPY only.
Housing market calls by finance YouTubers, checked against Case-Shiller and Freddie Mac. House prices are judged on the growth rate, not the price level.
Ethereum calls and price targets from Coin Bureau and other YouTube channels, quoted to the second and checked against real prices from 30 days to 5 years.
Andrei Jikh's own Bitcoin, S&P 500 and Fed calls, each linked to the second it was said and checked from 30 days to 5 years. Sample sizes shown.
Questions people ask
How accurate are Andrei Jikh's predictions?
As of Nov 4, 2026, Callsight has judged 51 of Andrei Jikh's calls Correct or Incorrect at one year: 30 correct and 21 incorrect. On the 49 of them with a comparable base rate (how often the called move happened anyway), 57.1% were correct, against a base rate of 41.7%. Every call links to the second it was said. After 3 years: 58.8% (34 calls) · on the 33 of 34 calls that have a base rate: 57.6% correct, base rate 48.7%
How are Andrei Jikh's predictions checked?
Each one is quoted word for word from a public video and linked to the second it was said. A price call is measured from the first closing price after the video went public, and every call is checked at 30 days, 90 days, 180 days, 1 year, 3 years, 5 years, by fixed rules that are the same for every channel.
Do Andrei Jikh's price calls beat their benchmark?
Return against the benchmark at one year: −6.4 points, the average over 15 price calls with a benchmark (95% range −31.4 to +18.6 points). How the return is measured is on the methodology page.
Is this investment advice?
No. This is a record of what people said in public and what happened next. Nothing here tells you to buy or sell anything.
Hit rates count each distinct call once, however often it was repeated, and leave out calls that could not be checked. Methodology