MVRV Z-Score

Two lines define a Bitcoin cycle in this metric. Market cap is what every coin is worth at today’s price; realized cap is what every coin was worth the last time it moved on-chain — the network’s aggregate cost basis. The gap between them is unrealised profit (or loss) in dollar terms. The MVRV Z-Score takes that gap and divides it by the expanding-window standard deviation of market cap, turning a raw dollar figure into a count of how many standard deviations above its own history the network sits.

The chart plots that standardised gap on a linear right axis, with spot price muted on a log left axis for orientation. The pink band above +7 is Awe & Wonder’s extreme-overvaluation zone; the sage band at or below +0.1 is the deep-bottom zone. Today the reading is +0.33, the Undervalued regime, with companion raw MVRV at 1.189×.

The definition is the Awe & Wonder original:

Z = (Market Cap − Realized Cap) / Stdev(Market Cap)

The numerator is the unrealised gap. The denominator is what gives the Z-score its cross-cycle bite — and its eventual undoing. It is the expanding-window standard deviation of market cap, computed over every observation in the series up to and including today. The original Awe & Wonder series and the popular dashboard reconstruction that propagated it both use the expanding window rather than a rolling one; we follow that convention so the +7 and +0.1 reference lines stay comparable across cycles.

Why standardise at all? Raw MVRV is a unitless ratio, but its empirical thresholds drift as Bitcoin matures. The Z-score sidesteps that by translating the gap into the network’s own historical volatility units — which is why a four-figure top in late 2013 and a print near $19,800 in 2017 both registered above +7 despite an order of magnitude of price difference. The catch, examined below, is that the same expanding denominator that made the early cycles comparable now silently re-scales every modern reading downward.

Our reconstruction caveat. btc oak does not have a per-UTXO last-spent price feed. We rebuild realized cap as a two-bucket weighted sum — short-term-holder supply × STH realized price plus long-term-holder supply × LTH realized price. This approximation tracks per-UTXO ground-truth realized cap to within a few percent, which propagates to a sub-decimal-point shift in Z at typical reads. The full caveat lives on the methodology page.

The Z-score earns its keep only at the edges. The +7 band has historically fired at blow-off tops alone; the +0.1 band has bracketed every cycle bottom on the record. The middle — roughly +0.1 to +5 — covers the majority of trading days and carries weak signal on its own. Negative and near-zero readings remain the most reliable thing this metric does, and they have survived the maturation that gutted the top-side threshold. Pair any mid-zone read with cohort-spread or realized-price context (SOPR, STH/LTH realized price, Reserve Risk) rather than acting on the Z level alone.

Reading the canonical cycle anchors against the live series surfaces both halves of the decay pattern at once. Cycle peaks at +8.68, +7.98, +10.40, +6.90, +3.54, +2.97 — declining each cycle since 2017. Cycle troughs at −0.61, −0.46, −0.35 — also compressing, though far more gently. The top side has decayed by two-thirds in eight years; the bottom side has barely moved.

The cleanest view of the regime shift is the per-cycle Z-score peak, in order. Pulling the maximum reading inside each cycle’s topping window from the live series shows three hits, then three misses — and the misses run monotonically lower:

Three of six cycle peaks on record cleared +7 — all of them before 2018. Three did not. The 2021 cycle’s Z peaked at +6.90 in February 2021, within breathing distance of the ceiling but missed. November 2021 topped at +3.54, less than half the threshold. March 2024’s pre-halving high reached +2.97 — well into the mid-cycle band, despite printing a fresh all-time price high near $73,000. The trend is monotonic since 2017.

The denominator grows. The expanding stdev of market cap is computed over every observation since August 2011, and as market cap balloons, its variance grows faster than the numerator can keep pace. A market-cap-realized-cap gap of $1 trillion in 2025 maps to a smaller Z than a $200 billion gap did in 2017, because the historical stdev anchor under each is wildly different. This is the single biggest reason the +7 line has drifted out of reach.

Lost coins bias realized cap. Every Patoshi-era coin that has not moved since 2010 sits in realized cap at a near-zero per-coin valuation. Chainalysis research has estimated 2.78 to 3.79 million BTC permanently lost; Sergio Demian Lerner’s Patoshi research identifies roughly 1.1 million Satoshi-era coins untouched since 2010 (recap on Yahoo Finance). The bias is one-directional: realized cap reads low, the numerator reads high, but the effect is dwarfed by the denominator’s relentless growth.

The honest verdict: the indicator is not broken, but treating +7 as a current signal is a category error, and there are three specific failure modes worth naming with dates.

The top threshold has expired. +7 fired in April and November 2013 and in December 2017 — and never again. (The only ≥7 prints in 2011 land in August, when the expanding-stdev window held just a handful of observations, so they are an initialisation artifact at a ~$11 price rather than a tracked top; the June 2011 $30 mania predates the series entirely.) The threshold last cleared +7 on 19 December 2017, three days after the $19,665 close-basis spot top. The 2021 cycle’s Z peaked at +6.90 (22 February 2021) — the closest any cycle has come since, and it missed. Anyone waiting for +7 to confirm a top in 2021 or 2024 got no signal at all, because the calibration window — Bitcoin’s 2011–2018 history, dominated by two blow-off tops — produced a stdev anchor far smaller than today’s. By March 2024 the denominator was orders of magnitude larger than in late 2017.

The new working ceiling is closer to +5. The 2021 cycle’s Z cleared +5 (peaking at +6.90 in February 2021), but November 2021 (+3.54) and March 2024 (+2.97) fell short even of that. Those same cycles also topped raw MVRV well below its historical ~3.7 ceiling, so the shortfall is consistent across both metrics rather than a Z-score artefact. For the 2020s, reading anything above the mid-cycle band as “late and stretched” captures more of the cycle than holding out for a number the asset may never print again.

The bottom side held — that is the asymmetry to lean on. While the top threshold decayed, the +0.1 deep-bottom line fired at every cycle floor: January 2015 (−0.61), December 2018 (−0.46), the March 2020 Covid flush (−0.16), and the November 2022 post-FTX low (−0.35). Negative depth has shrunk, but the regime has triggered every time. Roughly 17% of all trading days sit at or below +0.1, which is the structural reason a steady accumulation plan quietly works — most cheap-relative-to-cost-basis days cluster there.

Our realized-cap reconstruction is cohort-based, not per-UTXO. btc oak builds the denominator from STH/LTH cohort series rather than per-UTXO last-spent prices. The approximation tracks ground truth within a few percent across the historical record, but readings sitting on a regime boundary are the most sensitive to that gap. Treat a Z near +0.1 as “within 0.05 of either side” rather than a precise cross.

If you accumulate on a schedule, the bottom side is the only part of this chart you need. Z at or below +0.1 — and especially negative — has fired at every cycle floor on the record: −0.61 in 2015, −0.46 in 2018, −0.16 at the 2020 Covid low, −0.35 at the 2022 FTX low. Treat it as a high-conviction tactical accumulation signal. The depth has shrunk cycle to cycle, but the trigger has not failed once.

If you are timing the cycle top, stop waiting for +7. The 2010s thresholds describe the 2010s; for the 2020s the +5-to-+7 Overextended band is the de-facto top range, and even that overstated the last two cycles. Use Z as the slow backdrop and let faster instruments call the turn — it moves on the order of weeks, not days. Cross-check against raw MVRV for the unstandardised comparison and Reserve Risk for the long-term-holder conviction lens before acting on any top-side read.