Miner economics · Bitcoin
Cost of Production, explained
A modelled floor under Bitcoin's price, built from network difficulty and daily issuance — the same economics that force miners to sell. How the floor is calculated, why price rarely trades below it, and how far above it we are today.
- Formula
- α·D^0.45 / I
- Smoothing
- 90-day average
- Cost zone
- ×1.20
- Heatmeter weight
- 8 / 100
difficulty^0.45 over daily issuance
applied to both difficulty and issuance
band drawn above the smoothed floor
Bitcoin Heat Meter, one of 14 inputs
What is Cost of Production?
A modelled floor under Bitcoin's price, built from two things every miner on the network actually faces: how much computational work (difficulty) is required to find a block, and how much Bitcoin (issuance — the block subsidy plus transaction fees) they're paid for finding one. Combine those and you get an approximation of the marginal cost of producing a single coin.
It's built as a recreation of the published "BTC Cost Of Production" model, using CoinMetrics' community data for issuance and fees, and blockchain.info for network difficulty. Both sources are public and free for anyone to check.
How is the cost floor calculated?
The raw daily floor is α × difficulty^0.45 / daily issuance, where the 0.45 exponent on difficulty is carried over unchanged from the original model, and alpha is a scaling constant (more on that below). A single day's raw number is volatile — it swings with mining luck and how much issuance a given day happens to produce — so both difficulty and issuance are first smoothed with a 90-day simple moving average, and that smoothed pair produces the stable blue line you see on the chart.
A second line, the cost zone, is drawn at 1.20 times the smoothed floor — the upper edge of the band where price has typically rested when supported by, but not yet far removed from, production economics.
Why calibrate alpha instead of using the original constant?
The original model uses alpha = 1/1800. That constant assumes whatever difficulty scale its own source used. Our difficulty numbers come from blockchain.info's absolute network difficulty, which sits on a different numeric scale — plugging 1/1800 directly into that input produces a floor of roughly $3, off by a factor of about 21,853×.
Because difficulty and issuance are shaped the same way regardless of provider — only the absolute scale of difficulty differs — the fix is a single constant-factor correction: solving the model against known reference points gives alpha ≈ 12.14 (= 1/1800 × 21,853). With that value the model lines up with its reference points and, historically, the smoothed floor sits below Bitcoin's price the large majority of the time, as the underlying economics would predict.
Has Cost of Production ever been wrong?
Bitcoin has rarely closed below its modelled production cost, and the times it has were brief. But "a physical floor" is a strong claim for any model to make, and this one has real limits.
Alpha is a number chosen to make the model fit
The original published model uses alpha = 1/1800. Our difficulty figures come from a different source (blockchain.info) on a different scale, so alpha is recalibrated to 12.14 to line the model up with a known reference point. It is a number picked to make the model fit, not one worked out from mining hardware costs or electricity prices.
It's a network average, not any one miner's economics
Real mining costs vary enormously by hardware efficiency, electricity price, and hosting deals. The model collapses all of that into one constant applied to network-wide difficulty and issuance — it describes the network as a whole, not your specific mining rig.
The 0–100 score ranks each day against the whole history
The score is where today's price-to-floor ratio ranks against every day on record — including, for any past date you look at, days that came after it. So a reading from years ago can shift slightly as new extremes are added. The same is true of MVRV, AHR999, RHODL and Puell on this site.
On the Ethereum Heat Meter, this input is Bitcoin's floor, not an ETH model
Ethereum is no longer mined — the Merge moved it to proof of stake, where the network is secured by coins staked rather than by mining — so it has no mining difficulty and there is no Ethereum version of this calculation. The Ethereum Heat Meter’s Cost of Production input is the Bitcoin figure exactly as it appears here, just weighted 5/100 instead of 8/100 — a Bitcoin stand-in, not Ethereum’s own production cost.
How do we turn Cost of Production into a risk score?
The score isn't built around the 1.20 cost-zone multiple directly. Instead, for every date with both a smoothed floor and a price, we take price ÷ smoothed cost floor and then rank today's ratio as a percentile against every ratio in the whole history — the share of all days with an equal or lower ratio, on a 0–100 scale.
A ratio near its all-time low — price sitting close to the floor — scores near 0. A ratio at its all-time high — price at its most extended above production cost — scores near 100. Because the ranking uses the whole history, a reading from the past is ranked against days that hadn't happened yet at the time, not only the days before it. Several other on-chain inputs on this site are scored the same way.
Miners can't sell at a loss forever. The cost floor is what happens when the cheapest sellers get squeezed out.
How do you read the Cost of Production chart?
Watch the distance to the blue floor, not just the price level. A price sitting inside or just above the cost zone has historically meant miner economics are providing a natural backstop — miners forced to sell near break-even create supply pressure that dries up as price approaches the floor, since they can't profitably keep selling below it.
A close below the floor is the rare, extreme case. It has historically been brief and has coincided with generational entry points rather than a stable state — the model treats a sustained stay below production cost as economically unsustainable for the mining network as a whole.
Where it fits
Cost of Production is a miner-economics indicator — it answers whether Bitcoin is trading close to what it actually costs the network to produce a coin, not whether sentiment is stretched or a moving average has crossed. That makes it a genuinely different lens from valuation ratios like MVRV or technical signals like Pi Cycle.
It's one of fourteen inputs feeding the Bitcoin Heat Meter, currently at 8 out of 100 — tied with the Puell Multiple for fifth, behind AHR999 (13), Log Regression (11), RHODL (11) and Risk Wave (10) — meaning it carries meaningfully more say in the score than most of its thirteen peers. We adjust those weights as the indicators are re-tested. It also appears on the Ethereum Heat Meter, but there it's a stand-in — the same Bitcoin figure reused at a lower weight (5/100), since Ethereum has no mining difficulty of its own to model.
Common questions
What does "cost of production" mean for Bitcoin?
It's a modelled estimate of what it costs miners, in aggregate, to produce one Bitcoin — derived from network difficulty (how much computational work is required) and daily issuance (the block subsidy plus transaction fees miners are paid). It approximates the marginal cost of mining, not a specific miner's electricity bill.
How is the cost floor calculated?
The raw daily floor is alpha × difficulty^0.45 ÷ daily issuance in BTC, using a calibrated alpha of 12.14 and an exponent of 0.45 carried over from the original model. Because a single day’s raw figure swings with mining luck and with how much issuance that day happens to produce, both difficulty and issuance are first smoothed with a 90-day simple moving average, and that smoothed pair produces the stable blue floor line you see on the chart.
What is the "cost zone" band?
A second line drawn at 1.20 times the smoothed cost floor — an upper edge marking the region where price typically sits when it's supported by, but not yet far removed from, production economics. Price sitting inside that band is read as fair value; price well above it is read as increasingly stretched.
Why calibrate alpha to 12.14 instead of the original 1/1800?
The original model's alpha assumes whatever source supplied its difficulty numbers. Ours comes from blockchain.info's absolute network difficulty, which sits on a different numeric scale — using 1/1800 directly against that input produces a floor of roughly $3, off by a factor of about 21,853×. Difficulty and issuance behave the same way whoever reports them, so only the absolute scale differs. That means a single correction — 1/1800 × 21,853 ≈ 12.14 — realigns the model to known reference points without changing its shape.
How does Blockchain Decoded turn this into a risk score?
We take the ratio of live price to the smoothed cost floor for every date where we have both, then rank today's ratio against every ratio in the whole history — the share of all days with an equal or lower ratio, on a 0–100 scale. A ratio sitting near its all-time low (price close to the floor) scores near 0; a ratio at its all-time high (price most stretched above the floor) scores near 100.
Does this apply to Ethereum too?
No independent Ethereum version exists. The Ethereum Heat Meter includes a Cost of Production input, but it reuses the Bitcoin figure rather than working anything out from Ethereum’s own data — Ethereum is no longer mined, so it has no mining difficulty to feed the model. It carries a lower weight there (5/100 vs 8/100 on Bitcoin), as a Bitcoin stand-in.
Related guides
Others in on-chain — What the blockchain itself says about holders, miners, and what people paid for their coins.
- RHODL RatioThe balance between recently moved coins and long-dormant ones, which peaks near cycle tops.
- STH Realized PriceThe average price at which short-term holders bought — support in a bull market, resistance in a bear.
- Supply in Profit / Loss (NUPL)How much of all the Bitcoin out there is sitting on a paper profit, mapped to phases from panic to euphoria.
- Stock-to-FlowThe scarcity model that values Bitcoin from its issuance rate, shown with its deviation from actual price.
- Volume MomentumTrading volume against its own yearly baseline — speculative heat rather than price.
Every indicator we track — the full glossary.
See where price sits above the floor right now
Everything above is the metric on its own terms. The live chart plots the modelled miner cost floor from difficulty and issuance, and how far price is trading above it, with our 0–100 risk score on top. That score is also one of the fourteen inputs behind the Bitcoin Heat Meter.