The Problem with Naive Allocation
The most common mistake made by investors entering crypto markets for the first time is allocating by notional value rather than by risk. An investor who decides to “put 5 percent into crypto” has made a decision that sounds conservative but may not be. If the crypto allocation has four times the volatility of the equity allocation it sits alongside, that 5 percent notional position contributes a substantially larger share of total portfolio risk — and total portfolio drawdown during adverse markets.
This is not a new problem. Commodity traders learned decades ago that allocating equal notional amounts across different commodities produces wildly unequal risk contributions, because the volatility of crude oil, natural gas, and agricultural commodities can differ by an order of magnitude. The solution — allocating in proportion to risk rather than notional value — is straightforward in principle but requires some quantitative infrastructure to implement properly.
This article walks through a practical framework for risk-adjusted position sizing in crypto, drawing on commodity trading practice and adapting it for the specific characteristics of digital asset markets.
Step 1: Estimate Realised Volatility Correctly
The foundation of any risk-based sizing approach is an accurate volatility estimate. For crypto assets, several choices are worth examining carefully.
The most common approach is to calculate close-to-close realised volatility over a trailing window — typically 30 or 90 days — and annualise it by multiplying by the square root of 365 (since crypto trades every day, including weekends). This is a reasonable starting point, but it understates true volatility in markets that gap significantly between close and open.
Because crypto trades continuously, the “overnight gap” problem is different from equities but still relevant: during low-liquidity periods (typically the late Asian session), prices can move significantly without the cushioning effect of active market making. A Parkinson or Garman-Klass estimator, which uses high-low range data rather than close-to-close returns, typically produces a 10 to 20 percent higher volatility estimate for crypto than a simple close-to-close calculation. For conservative sizing, the higher estimate is preferable.
The choice of lookback window matters significantly for assets with changing volatility regimes. Bitcoin’s realised volatility in 2020-2021 was substantially higher than in 2023-2024, and a 90-day window in a low-volatility period will understate the volatility that will be experienced during the next high-volatility regime. One practical approach is to use an exponentially weighted moving average (EWMA) that gives more weight to recent observations but does not drop the memory of volatile periods entirely, combined with a volatility floor — a minimum assumption below which you will not size as if volatility is lower, regardless of what the lookback window shows.
For a conservative practitioner, a reasonable starting assumption for Bitcoin volatility is 60 percent annualised, for Ethereum it is 75 percent, and for smaller cap assets it is 100 percent or higher. These figures incorporate a judgment that recent low-volatility periods are not representative of long-run conditions.
Step 2: Define Your Risk Budget
Risk-based position sizing requires a clear definition of how much risk you are willing to allocate to a given position or asset class. This is expressed most usefully as a percentage of total portfolio volatility, or equivalently as a maximum acceptable contribution to portfolio drawdown.
A common approach in commodity fund management is the “risk unit” framework. Each position is sized to contribute exactly one risk unit — defined as 1 percent of portfolio NAV per unit of annualised volatility — to the total portfolio risk budget. A position with 60 percent annualised volatility, sized at one risk unit, would occupy 1 percent / 60 percent = 1.67 percent of portfolio NAV. A position in gold with 15 percent annualised volatility, sized at one risk unit, would occupy 1 percent / 15 percent = 6.67 percent of portfolio NAV.
This framework makes the risk contribution explicit and comparable across assets with different volatility profiles. It also provides a natural mechanism for adjusting position sizes as volatility changes: if Bitcoin’s volatility rises from 60 to 90 percent, the position is automatically reduced (to maintain constant risk contribution) rather than left at its previous notional size.
The total number of risk units allocated to crypto as an asset class is a separate decision, driven by the overall portfolio’s objectives and constraints. A conservative institutional allocation might allow one to two risk units to crypto in total. A more aggressive allocation focused on crypto alpha might allow five to ten. What matters is that the decision is made explicitly and that it is comparable to the framework used for other asset classes.
Step 3: Account for Correlation — But with Caution
If a portfolio holds multiple crypto assets, the aggregate risk contribution depends not just on individual volatilities but on the correlations between them. This is standard portfolio theory, but it requires careful application in crypto for two reasons.
First, correlations within the crypto universe are high and somewhat unstable. Bitcoin and Ethereum have historically had a 30-day rolling correlation of 0.7 to 0.9 — high enough that they are not providing meaningful diversification from each other in most market conditions. Smaller altcoins tend to be even more highly correlated with Bitcoin, particularly during sell-offs, where correlations typically spike toward 0.9 or higher across the entire crypto market.
The practical implication is that diversification across crypto assets provides much less risk reduction than diversification across, say, different commodity sectors. A portfolio holding Bitcoin, Ethereum, Solana, and several altcoins is not five times less risky than a portfolio holding only Bitcoin. It might be 20 to 30 percent less risky on an individual-position basis, but substantially more correlated to overall crypto market risk than the asset count would suggest.
Second, and more important, correlations in the tails differ dramatically from correlations in normal conditions. The historical correlation data will show, correctly, that Bitcoin and the S&P 500 have a low average correlation. It will not adequately capture the scenario where both fall sharply together in a risk-off event. For risk management purposes, stress-testing correlations at their peak historical levels — not their averages — is more conservative and more appropriate.
One practical approach is to model crypto as if it were a single asset (using Bitcoin as the representative) for the purpose of the portfolio-level correlation calculation, even if the actual crypto exposure is spread across multiple assets. This is conservative — it ignores diversification within the crypto allocation — but given the high intra-crypto correlations, it is not far from the empirical reality during stress events.
Step 4: Set Maximum Drawdown Limits and Stick to Them
Volatility-based position sizing determines the typical day-to-day risk contribution of a position. Drawdown limits set the maximum loss a position can inflict before it is closed. Both are necessary; neither alone is sufficient.
Crypto drawdowns are not hypothetical. Bitcoin experienced a drawdown of approximately 84 percent from its November 2021 high to its November 2022 low. Ethereum experienced a comparable drawdown. Many altcoins drew down 90 to 99 percent over the same period and have not recovered their peaks. An investor who sized “conservatively” in notional terms but had no pre-defined drawdown exit was still exposed to potential losses that could have materially impaired the overall portfolio.
The drawdown limit should be set before the position is entered, not during a drawdown when emotions are running high. A reasonable approach for crypto positions is to define a maximum loss as a percentage of total portfolio NAV — for example, “no single crypto position will be permitted to lose more than 2 percent of portfolio NAV before being reduced” — and to set the initial position size so that this limit aligns with the maximum plausible drawdown of the position itself.
For a position in Bitcoin, a conservative maximum plausible drawdown from entry might be 50 percent (a figure that Bitcoin has exceeded in past cycles). If the maximum portfolio loss from Bitcoin is set at 2 percent of NAV, and the maximum position drawdown is assumed to be 50 percent, the maximum position size is 2 percent / 50 percent = 4 percent of portfolio NAV. This is consistent with — and slightly more conservative than — the volatility-based calculation above, which is appropriate.
Step 5: Rebalancing and Dynamic Adjustment
Position sizing is not a one-time calculation. As prices move and volatility changes, the risk contribution of a position drifts away from its initial sizing. A disciplined approach requires periodic rebalancing.
For daily rebalancing (common in systematic strategies), the position is adjusted each day to reflect the current volatility estimate and the current price. For less active portfolios, weekly or monthly rebalancing is more practical. The key discipline is not to allow a position to drift significantly above its risk budget simply because the asset has performed well.
This is psychologically difficult. An investor who entered Bitcoin at $30,000 and sees it trading at $80,000 is strongly tempted to maintain the full position — after all, the trade is working. But if volatility has also increased, or if the position has grown as a percentage of portfolio NAV, maintaining the position at its original notional size means accepting more risk than was originally budgeted. Trimming into strength is not capitulation; it is risk management.
A Note on Leverage
Everything above applies to unleveraged spot positions. If leverage is used — through futures, margin accounts, or options — the effective volatility of the position is multiplied by the leverage ratio, and all the size calculations change accordingly. A leveraged position that is sized to “feel” comparable to an unleveraged position can impose drastically higher risk on a portfolio than intended.
The general recommendation for traders applying commodity risk management principles to crypto is to use leverage sparingly, at least until the risk dynamics of the specific instrument being traded are well understood. The liquidation risk on leveraged crypto positions — particularly on offshore exchanges with automatic liquidation engines — can result in total loss of the position in a way that does not apply to unleveraged spot holdings. Understanding that risk, and whether it is compatible with the overall portfolio’s drawdown tolerance, is essential before deploying leverage.
Conclusion
Risk-adjusted position sizing does not guarantee profits, and it does not eliminate the possibility of significant losses in a market as volatile as crypto. What it does is ensure that the losses, if they come, are proportionate to the risk deliberately taken — not disproportionate due to naive notional allocation.
The framework described here — estimate volatility conservatively, define a risk budget explicitly, account for tail correlations, set drawdown limits in advance, and rebalance regularly — is not exotic. It is standard practice in commodity trading and has been for decades. The adaptation to crypto requires adjustments for higher volatility and more extreme tail events, but the underlying logic is the same.
Traders who apply this discipline will find that they can participate meaningfully in crypto market opportunities while managing the substantial downside risks that the asset class carries. Those who do not will find that the downside surprises, when they come, are larger than they expected.
