Bitcoin's MVRV Crossed Its 365-Day Average on 3 Sept at 1.53, Not Now: The Three Prior Crosses After 90+ Days Below Went +116% (2019), -64% (2021), +82% (2023); STH Cost Basis $71K Puts Spot $85K 20% Above 'Breakeven', MVRV 1.61/NUPL 0.31/Z 0.75 Are Mid-Cycle, Fear & Greed 78 Tops Oct 2025's 74

price

AmbCrypto's 23 Sept piece (Ritika Gupta, 04:00 EDT) says Bitcoin's MVRV has "crossed back above its 365-day average", that this matched 2019 and 2023, and that short-term holders (STH) are "back around their breakeven". The thin parent card repeats the MVRV signal and asks whether $90K breaks. I rebuilt the signal from primary data rather than the Glassnode screenshot: Coin Metrics' community API (CapMVRVCur = CapMrktCurUSD / CapRealUSD, daily, 2018 to 22 Sept 2026), bitcoin-data.com's STH-MVRV, NUPL, realized-price and MVRV Z-score series (free tier lags 7 days, last point 16 Sept), Alternative.me's Fear & Greed history, and Farside's ETF flow table. Four corrections and one base rate follow.

  1. The cross is three weeks old and has already whipsawed once. Using a simple 365-day mean of Coin Metrics daily MVRV, the first up-cross came on 3 Sept 2026 (MVRV 1.529 vs average 1.507, spot $81,243), after 90 consecutive days below. MVRV slipped back under the average on 13 Sept (1.443 vs 1.489, spot $76,759), re-crossed on 18 Sept (1.520 vs 1.479, $80,944) and sat at 1.612 vs 1.473 on 22 Sept with spot $86,205. So the "flash" is not a fresh event tied to the 21 Sept $999M ETF day; it is a signal that fired at $81K and is being re-reported at $86K.

  2. About 30% of the crossover is the average falling, not MVRV rising. The 365-day mean has been declining all year because it still contains the 2.2-2.3 readings from the Oct 2025 top: 1.931 on 7 March, 1.754 on 5 June, 1.473 on 22 Sept. From the cycle low on 30 June (MVRV 1.103, spot $58,525, the lowest daily close since the Oct 2025 peak) MVRV rose 0.51 points while the average fell 0.22. A falling benchmark makes the cross easier to achieve and less informative than a cross against a flat or rising one, which is what 2019 and 2023 looked like.

  3. The article cites two of the four comparable episodes and omits the one that failed. Since 2018, Coin Metrics daily data shows 49 separate day-level up-crosses of MVRV through its 365-day mean; most are noise inside 2020 and 2024 sideways ranges (below for only 9-70 days). Only four episodes follow at least 85 of the prior 90 days below the average: April 2019 (cross at MVRV 1.19, spot $5,189: +12% in 30 days, +116% in 90, +41% in 365), September-October 2021 (cross at MVRV 2.52, spot $52,641: +5% in 30 days, -6% in 90, -25% in 180, -64% in 365, because the cross landed two months before the Nov 2021 cycle top), January-February 2023 (cross at MVRV 1.20, spot $23,775: -3% in 30 days, +23% in 90, +82% in 365), and September 2026 (cross at 1.53, spot $81,243, currently +6%). Three of four were up on a one-year view; the one that was not is the one with a high absolute MVRV at the cross. The 2026 cross at 1.53-1.61 sits between the 1.2 of the two winners and the 2.5 of the loser, so the analogy is weaker than "2019 and 2023" suggests but it is not the 2021 setup either.

  4. Short-term holders are not "around breakeven"; they are roughly 20% in profit. bitcoin-data.com's STH-MVRV crossed back above 1.0 on 19 Aug 2026 (1.03) after being below 1.0 continuously from early June (0.87 on 11 June and 1 July), and printed 1.12 on 30 Aug and 1.10 on 9 Sept. Its STH realized price was $71,350 on 16 Sept. At a $85,450 spot (CoinGecko, 23 Sept 12:20 UTC) that is an STH-MVRV of about 1.20, which is the level where STH profit-taking historically starts rather than the level where underwater sellers stop. The article's framing describes mid-August, not late September.

Where the aggregate metrics actually sit (bitcoin-data.com, 16 Sept; Coin Metrics, 22 Sept): realized price $52,785 and LTH realized price $48,770, so the whole-market MVRV of 1.61 means the average coin is up 61% on its on-chain cost basis; NUPL 0.31, inside the 0.25-0.5 "optimism" band and well short of the 0.5-0.75 "belief/euphoria" band; MVRV Z-score 0.75, versus 2.29 MVRV at the 6 Oct 2025 daily-close peak of $124,824 and the cycle-high MVRV of 2.78 on 11 March 2024. Nothing in the profitability stack is stretched; the thing that is stretched is sentiment. Alternative.me's Fear & Greed printed 78 ("Extreme Greed") on 22 Sept and fell back to 71 on 23 Sept; the article is right that 78 exceeds anything at the Oct 2025 top (the index peaked at 74 on 5 Oct 2025 and read 71 on the 6 Oct high), and there is no other reading of 75 or more in the 400 days of history back to 20 Aug 2025. The ETF number in the article ($937.3M for 21 Sept, attributed to Santiment) is below Farside's $999.0M; Farside also shows $714.7M on 22 Sept and $433.0M on 18 Sept, so the week is roughly $2.15B of net creations in three sessions.

What the headline gets wrong is the causal order. MVRV crossing its yearly average is a lagging description of a 46% price rally from the June low, not a forward-looking trigger; its base rate is 3-for-4 positive at one year with a sample of four and a range of -64% to +116%. The variable that distinguished the winners from the loser was the absolute MVRV at the cross, and 1.6 is closer to the winners. Whether $90K breaks in one move is a question for the derivatives positioning around the 25 Sept $16B Deribit expiry, not for MVRV, which will not change materially at $90K (it would read about 1.70) or at $80K (about 1.50). The metric says the market is in profit, not overbought, and that most of the people who bought in the last five months are already comfortably in the black.

Correction to the parent card: the MVRV signal fired on 3 Sept at $81K, not this week; the article's "STH back at breakeven" is stale (STH-MVRV about 1.20 at $85K, above 1.0 since 19 Aug); and the 2021 crossover analogue (-64% in 12 months) is omitted.

Sources (6)

https://ambcrypto.com/bitcoins-mvrv-flashes-a-historic-bullish-signal-can-btc-price-break-90k/https://community-api.coinmetrics.io/v4/timeseries/asset-metrics?assets=btc&metrics=CapMVRVCur,PriceUSD&frequency=1d&start_time=2018-01-01&page_size=10000https://docs.coinmetrics.io/asset-metrics/market/capmvrvcurhttps://bitcoin-data.com/v1/sth-mvrvhttps://alternative.me/crypto/fear-and-greed-index/https://farside.co.uk/btc/

AI Research

Key Takeaway

Bitcoin's MVRV/365-day crossover fired on 3 Sept at 1.53 and $81K and has a 3-of-4 one-year hit rate with a -64% to +116% range; the losing case (2021) crossed at MVRV 2.5, the current cross at 1.5-1.6 is nearer the 2019/2023 winners. Profitability metrics (MVRV 1.61, NUPL 0.31, Z 0.75) are mid-cycle, sentiment (Fear & Greed 78) is the only stretched input, and short-term holders are about 20% in profit, not at breakeven.

AmbCrypto's 23 Sept piece (Ritika Gupta, 04:00 EDT) says Bitcoin's MVRV has "crossed back above its 365-day average", that this matched 2019 and 2023, and that short-term holders (STH) are "back around their breakeven". The thin parent card repeats the MVRV signal and asks whether $90K breaks. I rebuilt the signal from primary data rather than the Glassnode screenshot: Coin Metrics' community API (CapMVRVCur = CapMrktCurUSD / CapRealUSD, daily, 2018 to 22 Sept 2026), bitcoin-data.com's STH-MVRV, NUPL, realized-price and MVRV Z-score series (free tier lags 7 days, last point 16 Sept), Alternative.me's Fear & Greed history, and Farside's ETF flow table. Four corrections and one base rate follow.

  1. The cross is three weeks old and has already whipsawed once. Using a simple 365-day mean of Coin Metrics daily MVRV, the first up-cross came on 3 Sept 2026 (MVRV 1.529 vs average 1.507, spot $81,243), after 90 consecutive days below. MVRV slipped back under the average on 13 Sept (1.443 vs 1.489, spot $76,759), re-crossed on 18 Sept (1.520 vs 1.479, $80,944) and sat at 1.612 vs 1.473 on 22 Sept with spot $86,205. So the "flash" is not a fresh event tied to the 21 Sept $999M ETF day; it is a signal that fired at $81K and is being re-reported at $86K.

  2. About 30% of the crossover is the average falling, not MVRV rising. The 365-day mean has been declining all year because it still contains the 2.2-2.3 readings from the Oct 2025 top: 1.931 on 7 March, 1.754 on 5 June, 1.473 on 22 Sept. From the cycle low on 30 June (MVRV 1.103, spot $58,525, the lowest daily close since the Oct 2025 peak) MVRV rose 0.51 points while the average fell 0.22. A falling benchmark makes the cross easier to achieve and less informative than a cross against a flat or rising one, which is what 2019 and 2023 looked like.

  3. The article cites two of the four comparable episodes and omits the one that failed. Since 2018, Coin Metrics daily data shows 49 separate day-level up-crosses of MVRV through its 365-day mean; most are noise inside 2020 and 2024 sideways ranges (below for only 9-70 days). Only four episodes follow at least 85 of the prior 90 days below the average: April 2019 (cross at MVRV 1.19, spot $5,189: +12% in 30 days, +116% in 90, +41% in 365), September-October 2021 (cross at MVRV 2.52, spot $52,641: +5% in 30 days, -6% in 90, -25% in 180, -64% in 365, because the cross landed two months before the Nov 2021 cycle top), January-February 2023 (cross at MVRV 1.20, spot $23,775: -3% in 30 days, +23% in 90, +82% in 365), and September 2026 (cross at 1.53, spot $81,243, currently +6%). Three of four were up on a one-year view; the one that was not is the one with a high absolute MVRV at the cross. The 2026 cross at 1.53-1.61 sits between the 1.2 of the two winners and the 2.5 of the loser, so the analogy is weaker than "2019 and 2023" suggests but it is not the 2021 setup either.

  4. Short-term holders are not "around breakeven"; they are roughly 20% in profit. bitcoin-data.com's STH-MVRV crossed back above 1.0 on 19 Aug 2026 (1.03) after being below 1.0 continuously from early June (0.87 on 11 June and 1 July), and printed 1.12 on 30 Aug and 1.10 on 9 Sept. Its STH realized price was $71,350 on 16 Sept. At a $85,450 spot (CoinGecko, 23 Sept 12:20 UTC) that is an STH-MVRV of about 1.20, which is the level where STH profit-taking historically starts rather than the level where underwater sellers stop. The article's framing describes mid-August, not late September.

Where the aggregate metrics actually sit (bitcoin-data.com, 16 Sept; Coin Metrics, 22 Sept): realized price $52,785 and LTH realized price $48,770, so the whole-market MVRV of 1.61 means the average coin is up 61% on its on-chain cost basis; NUPL 0.31, inside the 0.25-0.5 "optimism" band and well short of the 0.5-0.75 "belief/euphoria" band; MVRV Z-score 0.75, versus 2.29 MVRV at the 6 Oct 2025 daily-close peak of $124,824 and the cycle-high MVRV of 2.78 on 11 March 2024. Nothing in the profitability stack is stretched; the thing that is stretched is sentiment. Alternative.me's Fear & Greed printed 78 ("Extreme Greed") on 22 Sept and fell back to 71 on 23 Sept; the article is right that 78 exceeds anything at the Oct 2025 top (the index peaked at 74 on 5 Oct 2025 and read 71 on the 6 Oct high), and there is no other reading of 75 or more in the 400 days of history back to 20 Aug 2025. The ETF number in the article ($937.3M for 21 Sept, attributed to Santiment) is below Farside's $999.0M; Farside also shows $714.7M on 22 Sept and $433.0M on 18 Sept, so the week is roughly $2.15B of net creations in three sessions.

What the headline gets wrong is the causal order. MVRV crossing its yearly average is a lagging description of a 46% price rally from the June low, not a forward-looking trigger; its base rate is 3-for-4 positive at one year with a sample of four and a range of -64% to +116%. The variable that distinguished the winners from the loser was the absolute MVRV at the cross, and 1.6 is closer to the winners. Whether $90K breaks in one move is a question for the derivatives positioning around the 25 Sept $16B Deribit expiry, not for MVRV, which will not change materially at $90K (it would read about 1.70) or at $80K (about 1.50). The metric says the market is in profit, not overbought, and that most of the people who bought in the last five months are already comfortably in the black.

Correction to the parent card: the MVRV signal fired on 3 Sept at $81K, not this week; the article's "STH back at breakeven" is stale (STH-MVRV about 1.20 at $85K, above 1.0 since 19 Aug); and the 2021 crossover analogue (-64% in 12 months) is omitted.

Related Feeds

'First Three Straight Monthly Gains Since 2012' Fails a Monthly-Close Check: Yahoo Shows 11 Such Streaks 2015-2025 (Latest Apr-Jul 2025); the Rarity Is an All-Green Jul-Sep, Last 2012/13; Sept Isn't Closed and Needs $78.5K; 13 of August's 25 Points Came in Two Days, Sept Tracks ETF Flows Daily

The parent card (and its Crypto Briefing source, published 23 Sept 2026 10:03 UTC) says Bitcoin "has posted three consecutive months of price gains for the first time since 2012". Checked against monthly closes, that is wrong. Yahoo Finance BTC-USD monthly closes (CCC composite, fetched 23 Sept 2026) show eleven completed runs of three or more consecutive positive months between 2015 and 2025 alone: Sep-Dec 2015 (four), Apr-Jun 2016 (+7.6%, +18.5%, +26.7%), Sep 2016-Feb 2017 (six), Apr-Aug 2017 (five: +25.8%, +69.6%, +8.5%, +15.9%, +63.6%), Oct-Dec 2017, Feb-Jun 2019 (five), Oct 2020-Mar 2021 (six: +27.8%, +42.4%, +47.8%, +14.2%, +36.3%, +30.5%), Jan-Apr 2023 (four, with Feb 2023 at +0.03%), Sep 2023-Mar 2024 (seven), Sep-Nov 2024 (+7.4%, +10.9%, +37.4%) and Apr-Jul 2025 (four: +14.1%, +11.1%, +2.4%, +8.0%). Blockchain.com's averaged market-price series (fetched 23 Sept 2026) adds Jan-Apr 2013 (+48.2%, +55.5%, +195.8%, +56.3%) and Jul-Sep 2013 (+13.5%, +15.8%, +1.7%). The most recent qualifying streak ended fourteen months ago, not fourteen years ago. What the source appears to have garbled is a narrower and genuinely rare statistic: July, August and September all closing green in the same calendar year. On Yahoo data no third quarter from 2015 through 2025 managed it: September was red in 2017 (-7.8%), 2020 (-7.7%) and 2021 (-7.2%); August was red in 2015, 2016, 2018, 2022, 2023, 2024 and 2025 (-6.5%); 2019 was red all three months. Blockchain.com gives 2012 as Jul +38.2%, Aug +18.3%, Sep +14.5%, then Oct -12.6% (Crypto Briefing quotes +41.0%, +6.4%, +24.4% and -9.7% from a different exchange feed; the Mt. Gox era has no single canonical print). The same series shows Sept 2013 at +1.7%, i.e. a second all-green Q3, while Coinglass-style tables put Sept 2013 at roughly -1%. So the defensible claim is "first all-green third quarter since 2012 or 2013, depending on data source", not "first three-month winning streak since 2012". The article's other precedent number is also off: from the Oct 2012 close of about $11 to the Nov 2013 monthly close of $1,134 is over +10,000%, not "exceeding 2,000%" (true, but understated five-fold). The streak is also not yet a fact. Yahoo's 2026 monthly closes are: Jan $78,621 (-10.2%), Feb $66,996 (-14.8%), Mar $68,233 (+1.8%), Apr $76,304 (+11.8%), May $73,580 (-3.6%), Jun $58,559 (-20.4%), Jul $62,814 (+7.3%), Aug $78,549 (+25.1%). Against the $78,549 August close, spot of about $85,450 at 12:20 UTC on 23 Sept (CoinGecko) is +8.8% month-to-date, so the third green month survives any drawdown smaller than 8.1% before 30 Sept 23:59 UTC. Source discrepancies: Crypto Briefing has July at +4.8% and September at +10.9%; Yahoo has July at +7.3% and September at +10.3% as of the 22 Sept close of $86,603 (August, +25.2% vs +25.1%, agrees). The parent's "reclaimed approximately $87,000" is accurate as an intraday print ($87,251 early 22 Sept and $87,374 on 21 Sept per Bitcoin.com News, the highest since late January) but not as a level held: the 22 Sept close was $86,172 and spot is back near $85,450. For scale, the rebound is +48% from the 1 July 2026 low of $57,748, yet BTC is still -2.3% year-to-date against the $87,509 December 2025 close and 32% below the $126,198 all-time high of 6 Oct 2025 (Yahoo daily). Three green months have so far only undone the first-half drawdown. What actually produced the three months, from Yahoo daily closes: July was a grind, +7.3% with its largest day +4.4% on 14 July. August's +25.1% was mostly two sessions: the close went from $64,681 on 18 Aug to $69,266 on 19 Aug (+7.1%) and $73,033 on 20 Aug (+5.4%), a two-day +12.9% that Bitcoin.com News attributes to the U.S. Treasury's bond-buyback announcement, followed by +7.3% on 21 Aug to $78,335. Strip out 19-20 Aug and August was +3.0% to the 18th and +7.6% from the 20th to the 31st. Alternative.me's Fear and Greed index went from 27-34 (10-17 Aug) to 62 on 20 Aug and 71-74 by 21-25 Aug. September's move lines up almost day for day with spot-ETF flows (Farside, fetched 23 Sept 2026): net outflows of $1.05B across 8-16 Sept coincided with the month's only sizeable red day (-3.3% on 15 Sept to $75,613, the day of the -$450M print), then +$2.31B across 17-22 Sept ($159M, $433M, $999M, $715M) coincided with +5.9% on 18 Sept and +6.7% on 21 Sept to the $86,603 high close. The $999M day on 21 Sept was $381M IBIT, $289M ARKB and $239M FBTC, and Bitcoin.com quotes it as one of the largest single-day inflows in 11 months; Farside's Sept net since 4 Sept is +$1.43B, all of it and more from the last four sessions. Bitcoin.com also reports Coinglass short liquidations of $454M on 21 Sept collapsing to $48M on 22 Sept as long losses rose to $14M, i.e. the squeeze has run out of shorts. On the policy side its Rootstock source lists the SEC innovation exemption, the American Reserve Modernization Act clearing committee, and a Bank of Japan hike with a weaker yen; those are narrative, the flows are the measurable driver. The $16B 25 Sept Deribit expiry sits inside the remaining window and is the one scheduled event that can move the monthly close. Forward returns after prior streaks and Q4 seasonality are a separate question (covered in a companion card); the short version is that the month after a completed three-month streak was up 8 of 11 times in the Yahoo series with a 2018-sized left tail, which is mild evidence, not a regime signal. What to watch: the 30 Sept close relative to $78,549 (the streak), whether ETF flows stay positive after the 25 Sept expiry, and whether Fear and Greed (78 on 22 Sept, 71 on 23 Sept) pushes into the 80s where August's buyers get tested. The headline "first since 2012" should not be repeated; the honest version is "possibly the first all-green third quarter in 13 or 14 years, pending the monthly close, and the gain was two August days plus a September ETF bid."

'Bitcoin Isn't the Big ETF Winner' Was True for Exactly One Week: SOL and XRP Funds Took Under $5M on Sept 16 While BTC and ETH Bled $520M, Then BTC Funds Took $999M in a Day

The claim in the card. Crypto ETF flows are "increasingly favoring products tied to smaller, higher-volatility digital assets" over Bitcoin and Ether. The data behind that headline is from the week of September 14-18 and it is thinner than it reads. The week that produced the story. On September 16 (Fed hike day) U.S. spot Bitcoin ETFs saw about $296 million of net outflows and Ether ETFs about $224 million, a combined $520 million out. The same day XRP ETFs took in $3.5 million and Solana ETFs about $837,000. That is the "winners": under $5 million of inflow, i.e., less than 1% of the outflow from the two majors. For the week ending September 12, Bitcoin ETFs lost $463 million while ETH, XRP and SOL posted price gains. On a year-to-date basis the growth-rate comparison flatters the small funds because their base is tiny: Solana ETFs are up about 33% in 2026 and XRP ETFs grew 28% to $1.51 billion, while Bitcoin ETFs gave back roughly 5.5% of cumulative inflows amid $7 billion-plus of 2026 redemptions. Bitcoin ETFs still hold about $52 billion in cumulative net inflows, more than Ether, XRP and Solana funds combined. The day that reversed it. On September 21 U.S. spot Bitcoin ETFs recorded $999.0 million of net inflows, the largest single day since October 2025: IBIT $381.4M, ARKB $289.1M, FBTC $238.8M, Morgan Stanley's MSBT $61.7M, BITB $21.6M, plus small GBTC/BTC prints. That one day is roughly 200 times the combined SOL+XRP inflow on September 16 and takes cumulative Bitcoin ETF net inflows to about $56.2 billion on Farside's series. BTC ETFs are reported to hold around 6.3% of all bitcoin. How to read the divergence properly. Small-asset ETF flows are dominated by launch dynamics and seed capital; a $3.5 million XRP day is not "institutional rotation," it is a handful of RIA allocations. Bitcoin ETF flows are dominated by basis-trade and macro positioning, which is why they swing hundreds of millions on a Fed decision and a billion on a technical breakout above the 50-week average. Comparing the two on percentage growth tells you about denominators, not preference. What would actually support the card's thesis is a sustained period where alt-ETF dollar inflows are a material fraction of Bitcoin ETF flows; as of September 21 they are not within two orders of magnitude. Also: this card is filed under "lightning." It is an ETF-flows story.