If Bitcoin (BTC) Supply-in-Profit RSI Crosses the Threshold, Glassnode Caps Loss at 20%
Glassnode's backtest of an RSI rule on Bitcoin supply-in-profit data capped maximum drawdown at 20%, against 53% for continuous holding in the same window.
AI SummaryAI
- Glassnode released an RSI-based Bitcoin strategy backtest on its Alpha Lab platform on October 27, 2026.
- The backtest capped maximum drawdown at 20%, against 53% for continuous holding.
- The strategy posted a 63% win rate and a 1.53 Sharpe ratio across 79 trades.
- Annualized return reached 52% for the strategy against 38% for buy-and-hold.
Glassnode's 20% Drawdown Ceiling
For the rule to work the way Glassnode describes it, one condition has to hold first: the 30-day RSI of Bitcoin's share of supply in profit must cross a threshold the firm has chosen not to publish. Only above that line does the model open a long position; below it, the strategy stays in cash. The backtest behind these numbers appeared on October 27, 2026, when the on-chain analytics firm released results from its Alpha Lab platform for
Bitcoin (BTC). The signal does not read the Bitcoin (BTC) price directly. It applies the RSI to a blockchain gauge that estimates how much of the circulating supply is held by entities that bought below the current market level, a group that runs from small wallets to each crypto whale. Because that measure tracks unrealized profit across holders, the model reacts to shifts in investor profitability rather than to chart patterns. A high share of supply in profit alone says nothing certain about direction: markets have climbed while most holders were in profit and fallen from the same condition. Glassnode ran the comparison over the period after July 2023, a window that also contains the most recent Bitcoin halving. Across that span, the strategy's maximum drawdown, the deepest peak-to-trough decline recorded during the test, came to 20%. Investors who held continuously, the pure HODL approach, absorbed a maximum decline of 53% over the same window. The two figures rest on the same market history and the same assumptions, which is what makes the comparison meaningful. They do not say the strategy makes 20% losses impossible in live trading. They say that, in this backtest, timing entries off holder profitability cut the worst sustained loss to well under half of what an investor who never sold endured.
Update (28 Oct 05:18 UTC): Into this debate over where
Bitcoin (BTC) heads next, Coinbase CEO Brian Armstrong added a supply-side argument on Monday, October 27, when he compared the 21 million coin cap to gold and diamonds, two assets that laboratories are learning to manufacture.
A separate read on the same tape arrived from Galaxy Research, whose October 22 report to clients, published October 26, argued that Bitcoin's bear market ended in June. Head of firmwide research Alex Thorn wrote that Bitcoin sits in an accumulation phase of the kind that preceded past advances and sees conditions in place for a return to $125,000. The evidence included five consecutive weekly closes above the 50-week moving average after 44 weeks below it, and a slowdown in old coins moving: monthly flows of Bitcoin dormant at least five years fell from 67,000 BTC in 2024 to 34,000 BTC in 2026.
A third voice entered the tape later on Thursday, when digital asset analyst Killa (KillaXBT) posted a roadmap on X for Bitcoin (BTC) drawn from past early-bull-market patterns. From the current level of $80,800, he projected $100,000 in 160 days, and a later reading of $126,000 in 280 days — by 15:50 UTC this stood against the $125,000 target cited earlier — and $160,000 within 500 days, gains of roughly 24%, 56% and 98%. His chart placed "
Bitcoin (BTC) NOW" beside "BTC BEFORE," arguing the market is repeating a sequence of sharp drop, sideways consolidation and base formation seen at prior bull-market starts. He also framed the current pullback as a correction inside an uptrend, noting the previous bear market cut Bitcoin about 78% while this one ended near 54%, which he said could shrink bull-market drawdowns to 10–15% from the 20–22% of past cycles.
Into this debate over where Bitcoin heads next, Coinbase CEO Brian Armstrong added a supply-side argument on Monday, October 27, when he compared the 21 million coin cap to gold and diamonds, two assets that laboratories are learning to manufacture. Citing Allied Market Research figures, he noted the lab-grown diamond market was worth $24 billion in 2022 and is projected to reach $59.2 billion by 2032 at 9.6% annual growth, while San Francisco startup Marathon Fusion reported last week that it converted liquid mercury into stable gold atoms. Armstrong said nothing of the kind can be invented for Bitcoin, whose 21 million limit is enforced by every independent full node and could only be raised through a network-wide hard fork. ProCap Financial CEO Anthony Pompliano backed the view.
79 Trades, Four Days Each
The performance table fills in the rest of the picture. The strategy completed 79 trades across the test, held the average position for four days, and was active for roughly 44% of the period, while continuous holding sits in the market 100% of the time by definition. Its successful-trade ratio came to 63%. Its Sharpe ratio, a measure that sets a strategy's returns against the risk taken to earn them, reached 1.53 against 0.92 for holding through the full window. Annualized returns favored the model as well: 52% against 38%. Spending less than half the time in the market is the mechanism behind the narrower drawdown: the model simply was not positioned during the stretches that produced holding's deepest losses, and the four-day average hold means it captured short bursts rather than riding the full trend.
| Metric | RSI strategy | Continuous holding |
| Annualized return | 52% | 38% |
| Maximum drawdown | -20% | -53% |
| Sharpe ratio | 1.53 | 0.92 |
| Successful-trade ratio | 63% | Not stated |
| Trades | 79 | Not stated |
| Average position | 4 days | Not stated |
| Time in market | 44% | 100% |
The 20% and 53% readings describe peak-to-trough declines, not total losses, and the returns shown are model outputs from historical testing rather than realized investor gains. One gap remains: the published material does not state the numerical RSI threshold, so the full set of trading conditions cannot be reconstructed from the disclosure alone. Fees, slippage and the properties of the underlying data set can all change how a rule like this performs once it leaves the backtest, and parameters that look strong in one stretch of history can lose that edge in the next.
Risk Control, Not a Buy Signal
The result matters more as a study in risk control than as a trading plan. What carried the performance was the time out of the market, not the entries alone: a model positioned 44% of the time escaped the stretches that drove holding's 53% peak-to-trough loss. That framing fits the tape our recent coverage tracked, from Bitcoin's slide to a five-week low to the net exchange outflow of $1.01 billion recorded on October 27, and readers can weigh the model's assumptions against the levels in our Bitcoin technical analysis and the wider tape on our Bitcoin tag. The condition furthest from being met is verification: the RSI trigger is unpublished, no outside desk can reproduce the entries, and until Glassnode prints the threshold or the model holds up across a different cycle, the 20% ceiling stays exactly what it is, a backtest result.
AI-generated, AI-reviewed, under COINOTAG editorial oversight.

