pith. sign in

arxiv: 2605.20528 · v1 · pith:R6JGQMD3new · submitted 2026-05-19 · 💻 cs.CE

Modern Portfolio Theory in the Crypto-Wilderness

Ivan Vynyavskyy , Stefan Kitzler , Bernhard Haslhofer , Aviv Yaish This is my paper

Pith reviewed 2026-05-21 06:00 UTC · model grok-4.3

classification 💻 cs.CE
keywords cryptocurrencymodern portfolio theoryefficient frontierblockchain dataportfolio optimizationmarket timingEthereum accountsrealized returns
0
0 comments X

The pith

Market entry timing predicts cryptoasset returns far better than any portfolio allocation strategy.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Using complete on-chain records, the study reconstructs portfolios across more than 116 million Ethereum accounts from 2015 through 2025. It finds that the month an account first enters the market accounts for 70 to 79 percent of the variance in realized returns, while the choice of how to divide holdings among assets adds almost nothing. Simple market-capitalization weighting produces higher median returns than any mean-variance optimization technique tested. The data also show extreme concentration: more than four-fifths of accounts hold only a single asset.

Core claim

Reconstruction of every Ethereum account's holdings from public token-transfer data reveals that market entry month alone explains 70-79 percent of realized-return variance, dwarfing the explanatory power of allocation choice; passive market-capitalization weighting outperforms every tested mean-variance optimization rule, and observed portfolios sit closest to the efficient frontier simply because they contain so few assets rather than because they have been deliberately optimized.

What carries the argument

On-chain portfolio reconstruction from token-transfer history, which permits measurement of each account's distance to the mean-variance efficient frontier defined by the assets it actually holds.

Load-bearing premise

That every account's true risk exposures and holdings are fully captured by on-chain transfers without material off-chain activity, multi-account splitting, or unrecorded liquidity constraints.

What would settle it

A direct test showing that entry-month explanatory power drops below 50 percent once off-chain holdings or multi-account strategies are incorporated into the same accounts.

read the original abstract

Modern Portfolio Theory (MPT) prescribes how to maximise the return of an asset portfolio for a given level of risk. The optimal trade-off between return and variance defines the efficient frontier. Whether actual cryptoasset portfolios approximate this prescription and whether proximity to the frontier translates into realised performance remain difficult to test at large scale in traditional markets due to their opaque nature and the inaccessibility of data. As we show, public blockchains make these questions measurable: every token transfer is recorded, thus enabling complete portfolio reconstruction for every account at any point in time. We leverage this transparency to reconstruct cryptoasset portfolios for over 116M Ethereum accounts across the full chain history (2015-2025), measure their distance to the constrained efficient frontier, and quantify how deviations translate into realised performance. Here we show that market entry timing, not allocation choice, is the dominant predictor of realised cryptoasset returns. On-chain wealth is highly concentrated and portfolios are pervasively under-diversified, with single-asset holdings accounting for 83.35% of accounts. Two-asset portfolios sit closest to the efficient frontier defined by their held assets, a proximity that reflects the narrowness of their opportunity set rather than deliberate optimisation. Passive market-capitalisation weighting outperforms every MPT optimisation strategy in median realised return, and entry month alone explains 70-79% of the variance in returns, far exceeding the contribution of allocation choice. Mean-variance optimisation therefore appears neither descriptive of observed behaviour nor prescriptively useful in the cryptoasset domain, even if MPT retains its value as a normative benchmark.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript reconstructs cryptoasset portfolios for over 116 million Ethereum accounts from 2015-2025 using public on-chain transfer data, measures each portfolio's distance to a constrained efficient frontier, and reports that market entry timing explains 70-79% of variance in realized returns while allocation choice contributes far less. It further finds pervasive under-diversification (83.35% single-asset holdings), that two-asset portfolios lie closest to their asset-defined frontier, and that passive market-capitalization weighting outperforms all tested MPT optimization strategies in median realized return.

Significance. If the reconstruction and variance decomposition hold, the work supplies a rare large-scale empirical test of Modern Portfolio Theory in an emerging asset class, leveraging blockchain transparency to quantify investor behavior at unprecedented scale. The finding that entry timing dominates allocation, together with the outperformance of passive weighting, would challenge both the descriptive accuracy and prescriptive utility of mean-variance optimization in crypto while preserving its normative value. The 116 M account sample and the explicit variance attribution constitute clear strengths.

major comments (2)
  1. [Abstract] Abstract: the headline result that entry month alone explains 70-79% of realized-return variance rests on the assumption that token-transfer events permit complete reconstruction of every account's holdings, entry timing, and risk exposures. This assumption is load-bearing; substantial off-chain custody (CEX balances), multi-address strategies, and non-transfer exposures (DeFi lending, liquidity provision) would systematically mis-measure portfolio weights and entry dates, thereby inflating the apparent explanatory power of timing relative to allocation.
  2. [Methods] Methods (implied by absence of detail): no description is given of the precise constraints imposed when computing the per-account efficient frontier, the rules for filtering illiquid tokens, price sources, or the treatment of accounts with incomplete histories. These choices directly affect the reported proximity metrics, the claim that two-asset portfolios sit closest to the frontier, and the comparison of passive versus optimized strategies.
minor comments (1)
  1. [Abstract] The abstract states the time window as 2015-2025 but does not specify the exact end date or how returns are computed (simple holding-period returns, log returns, or risk-adjusted).

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which help clarify the scope and limitations of our on-chain analysis. We address each major comment below and indicate the revisions we will implement.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the headline result that entry month alone explains 70-79% of realized-return variance rests on the assumption that token-transfer events permit complete reconstruction of every account's holdings, entry timing, and risk exposures. This assumption is load-bearing; substantial off-chain custody (CEX balances), multi-address strategies, and non-transfer exposures (DeFi lending, liquidity provision) would systematically mis-measure portfolio weights and entry dates, thereby inflating the apparent explanatory power of timing relative to allocation.

    Authors: We agree this assumption requires qualification. Our reconstruction relies exclusively on observable Ethereum token transfers, which capture on-chain movements but exclude off-chain custody, multi-address aggregation, and non-transfer positions such as DeFi lending or liquidity provision. These omissions could bias the timing-versus-allocation decomposition. We will revise the abstract to specify that findings apply to on-chain portfolios and add a dedicated limitations subsection discussing these data gaps, potential biases, and directions for future integration of off-chain signals where feasible. revision: yes

  2. Referee: [Methods] Methods (implied by absence of detail): no description is given of the precise constraints imposed when computing the per-account efficient frontier, the rules for filtering illiquid tokens, price sources, or the treatment of accounts with incomplete histories. These choices directly affect the reported proximity metrics, the claim that two-asset portfolios sit closest to the frontier, and the comparison of passive versus optimized strategies.

    Authors: We acknowledge the need for greater methodological transparency. The revised manuscript will expand the Methods section with explicit details: per-account efficient frontiers are computed via quadratic programming under long-only constraints with no short-selling or leverage; illiquid tokens are filtered using a minimum 30-day average trading volume of $100k and market capitalization above $1M; price data are obtained from the CoinGecko API supplemented by on-chain oracle feeds for major assets; accounts with incomplete histories are retained from their first observed transfer, with robustness checks restricted to accounts active throughout the sample window. These additions will include pseudocode and parameter tables to support reproducibility. revision: yes

Circularity Check

0 steps flagged

No circularity in the derivation chain

full rationale

The paper reconstructs account portfolios directly from public Ethereum transfer events, computes per-account constrained efficient frontiers using the specific assets held by each account, measures distances to those frontiers, and performs a statistical variance decomposition of realized returns with entry month and allocation variables as predictors. No equations reduce the key reported metrics (70-79% variance explained by entry timing, outperformance of market-cap weighting) to fitted parameters or inputs by construction. The analysis is self-contained against the observable on-chain data and uses standard MPT calculations without self-definitional loops, load-bearing self-citations, or imported uniqueness results.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Analysis rests on completeness of public blockchain records and standard MPT assumptions adapted to crypto without additional invented entities.

free parameters (1)
  • Efficient frontier constraints
    Specific limits on short-selling, position sizes, or asset inclusion used when computing the constrained frontier for each account's held assets.
axioms (1)
  • domain assumption On-chain transfers fully capture portfolio composition and risk
    Assumes no significant off-chain trading, custodial mixing, or unrecorded positions affect measured holdings.

pith-pipeline@v0.9.0 · 5822 in / 1179 out tokens · 31010 ms · 2026-05-21T06:00:23.229883+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

  1. [1]

    Barber and Terrance Odean

    1 Brad M. Barber and Terrance Odean. Trading is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors.The Journal of Finance, 55(2):773–806, 2000.doi:10.1111/0022-1082.00226. 2 Tom Barbereau, Reilly Smethurst, Orestis Papageorgiou, Alexander Rieger, and Gilbert Fridgen. DeFi, not so decentralized: The measured distribut...

    work page doi:10.1111/0022-1082.00226 2000

  2. [2]

    4 Board of Governors of the Federal Reserve System (US)

    doi: 10.1257/aer.91.1.79. 4 Board of Governors of the Federal Reserve System (US). 3-Month Treasury Bill Secondary Market Rate, Discount Basis [TB3MS],

    work page doi:10.1257/aer.91.1.79

  3. [3]

    Retrieved from FRED, Federal Reserve Bank of St. Louis. URL:https://fred.stlouisfed.org/series/TB3MS. I. Vynyavskyy 21 5 Nicola Borri, Yukun Liu, Aleh Tsyvinski, and Xi Wu. Cryptocurrency as an Investable Asset Class: Coming of Age, October 2025.arXiv:2510.14435. 6 Marie Brière, Kim Oosterlinck, and Ariane Szafarz. Virtual currency, tangible return: Portf...

    work page doi:10.2469/faj.v42.n4.39 2025

  4. [4]

    10 Nir Chemaya, Aviv Yaish, Shahar Yacouel, Dahlia Malkhi, and Lin William Cong

    doi:10.1111/j.1540-6261.2006.00883.x. 10 Nir Chemaya, Aviv Yaish, Shahar Yacouel, Dahlia Malkhi, and Lin William Cong. Quanti- fying inequality in blockchain networks, September

    work page doi:10.1111/j.1540-6261.2006.00883.x 2006

  5. [5]

    11 CoinGecko

    URL:https://papers.ssrn.com/ abstract=5540258. 11 CoinGecko. CoinGecko: Cryptocurrency prices and market capitalization. https:// www.coingecko.com. Accessed: 2026-01-19. 12 Jeffrey Dean and Sanjay Ghemawat. MapReduce: Simplified data processing on large clusters. InProceedings of the 6th Symposium on Operating Systems Design and Implementation (OSDI), pa...

    work page 2026

  6. [6]

    Optimal versus naive diversification: How inefficient is the 1/n portfolio strategy?The Review of Financial Studies, 22(5):1915–1953, May 2009.doi:10.1093/rfs/hhm075

    14 Victor DeMiguel, Lorenzo Garlappi, and Raman Uppal. Optimal versus naive diversification: How inefficient is the 1/n portfolio strategy?The Review of Financial Studies, 22(5):1915–1953, May 2009.doi:10.1093/rfs/hhm075. 15 Tatiana Didier, Roberto Rigobon, and Sergio L. Schmukler. Unexploited Gains From Inter- national Diversification: Patterns Of Portfo...

    work page doi:10.1093/rfs/hhm075 1915

  7. [7]

    17 Erigon

    URL: https://openlibrary.org/books/OL25441475M/ Modern_portfolio_theory_and_investment_analysis. 17 Erigon. Erigon (formerly turbo-geth). https://github.com/erigontech/erigon. Accessed: 2026-01-19. 18 ethereum.org. What is wrapped ether (WETH),

    work page 2026

  8. [8]

    URL:https: //ethereum.org/en/wrapped-eth/

    Accessed: 2026-04-22. URL:https: //ethereum.org/en/wrapped-eth/. 19 Kenneth R. French and James M. Poterba. Investor diversification and international equity markets.American Economic Review, 81(2):222–226,

    work page 2026

  9. [9]

    20 William N

    Papers and Proceedings. 20 William N. Goetzmann and Alok Kumar. Equity Portfolio Diversification.Review of Finance, 12(3):433–463, 2008.doi:10.1093/rof/rfn005. 21 GraphSense. GraphSense REST api,

    work page doi:10.1093/rof/rfn005 2008

  10. [10]

    23 Oliver Hellum, Theis Ingerslev Jensen, Bryan T

    doi:10.1086/260910. 23 Oliver Hellum, Theis Ingerslev Jensen, Bryan T. Kelly, and Semyon Malamud. Complex modern portfolio theory. SSRN Working Paper 6443319, March

    work page doi:10.1086/260910

  11. [11]

    24 Gur Huberman

    URL:https://ssrn.com/ abstract=6443319,doi:10.2139/ssrn.6443319. 24 Gur Huberman. Familiarity breeds investment.The Review of Financial Studies, 14(3):659–680, 2001.doi:10.1093/rfs/14.3.659. 22 Modern Portfolio Theory in the Crypto-Wilderness 25 Roger G. Ibbotson and Paul D. Kaplan. Does asset allocation policy explain 40, 90, or 100 percent of performanc...

    work page doi:10.2139/ssrn.6443319 2001

  12. [12]

    28 Olivier Ledoit and Michael Wolf

    doi:10.1017/ S0022109000004129. 28 Olivier Ledoit and Michael Wolf. Honey, i shrunk the sample covariance matrix.The Journal of Portfolio Management, 30(4):110–119, July 2004.doi:10.3905/jpm.2004.412318. 29 Qiangqiang Liu, Qian Huang, Frank Fan, Haishan Wu, and Xueyan Tang. Detecting sybil addresses in blockchain airdrops: A subgraph-based feature propaga...

    work page doi:10.3905/jpm.2004.412318 2004

  13. [13]

    Markowitz, Portfolio selection, The Journal of Finance, 7 (1952), pp

    doi: 10.1111/j.1540-6261.1952.tb01525.x. 34 Harry M. Markowitz. Foundations of Portfolio Theory.The Journal of Finance, 46(2):469–477, 1991.doi:10.1111/j.1540-6261.1991.tb02669.x. 35 Massimo Massa and Andrei Simonov. Hedging, Familiarity and Portfolio Choice.Review of Financial Studies, 19(2):633–685, 2006.doi:10.1093/rfs/hhj013. 36 Sarah Meiklejohn, Marj...

    work page doi:10.1111/j.1540-6261.1952.tb01525.x 1952

  14. [14]

    42 PietroSaggese, MichaelFröwis, StefanKitzler, BernhardHaslhofer, andRaphaelAuer

    doi:10.1016/ j.econlet.2019.01.019. 42 PietroSaggese, MichaelFröwis, StefanKitzler, BernhardHaslhofer, andRaphaelAuer. Towards verifiability of total value locked (TVL) in decentralized finance, 2025.arXiv:2505.14565, doi:10.48550/arXiv.2505.14565. 43 William F. Sharpe. Capital Asset Prices: A Theory of Market Equilibrium Under Conditions of Risk.The Jour...

    work page doi:10.48550/arxiv.2505.14565 2019

  15. [15]

    doi:10.1007/978-3-030-51280-4_33

    Springer-Verlag. doi:10.1007/978-3-030-51280-4_33. 46 Friedhelm Victor and Bianca Katharina Lüders. Measuring ethereum-based ERC20 token networks. InFinancial Cryptography and Data Security, volume 11598 ofLNCS, pages 113–129, Berlin, Heidelberg,

    work page doi:10.1007/978-3-030-51280-4_33

  16. [16]

    47 Aviv Yaish, Nir Chemaya, Dahlia Malkhi, and Lin William Cong

    Springer-Verlag.doi:10.1007/978-3-030-32101-7_8. 47 Aviv Yaish, Nir Chemaya, Dahlia Malkhi, and Lin William Cong. Inequality in the age of pseudonymity. InProceedings of the Fortieth AAAI Conference on Artificial Intelligence (AAAI- 26), AAAI’26/IAAI’26/EAAI’26, pages 17293–17301. AAAI Press, 2026.arXiv:2508.04668. 24 Acronyms A Glossary Following is a li...

    work page doi:10.1007/978-3-030-32101-7_8 2026

  17. [17]

    DeFi Summer

    109,007,240 7,174,916 30,000,000 60,000,000 90,000,000 2,000,000 4,000,000 6,000,000 Jun '20 Dec '20 Jun '21 Dec '21 Jun '22 Dec '22 Jun '23 Dec '23 Jun '24 Dec '24 Jun '25 Dec '25 Jun '26 Externally owned accounts (EOA) Contract accounts (CA) Contract accounts (CA) Externally owned accounts (EOA) Figure 11Number of accounts: the temporal evolution of acc...

    work page 2025

  18. [18]

    The top subplot shows CA, the bottom subplot EOA

    EOA CA Jun '20 Dec '20 Jun '21 Dec '21 Jun '22 Dec '22 Jun '23 Dec '23 Jun '24 Dec '24 Jun '25 Dec '25 0% 25% 50% 75% 100% 0% 25% 50% 75% 100%Share (%) Wealth bins [USD] 0−1 1−100 100−1K Figure 12Distribution of accounts acrosslow-wealthbins over time (share of accounts with non-zero wealth). The top subplot shows CA, the bottom subplot EOA. Acronyms 27 E...

    work page 2021

  19. [19]

    Figure 14 summarises the annual distribution of tokens grouped by their holder count

    This decline is driven by a compositional shift. Figure 14 summarises the annual distribution of tokens grouped by their holder count. The share of tokens with more than 10000 holders drops from28.3% to15.6%, while smaller tokens (100–1000 holders) grow from17.7% to27.5%. Mid-sized tokens (1000–10000 holders) remain the dominant group throughout, ranging ...

    work page 2025

  20. [20]

    3-month Treasury bill yield

    proxies the risk-free rate with the U.S. 3-month Treasury bill yield. Their convention has carried over to subsequent crypto-portfolio work, and our setting matches it directly: in the 2023–2024 window, which contains the bulk of our account-time observations, the FRED seriesTB3MS averaged≈5% [4]. Empirical match to on-chain stablecoin yields. Account hol...

    work page 2023

  21. [21]

    Table 12Per-snapshot frequency of beating the wETH–WBTC market index, summarised across the 72 monthly snapshots. Each cell is the unweighted median or mean across snapshots of the within-snapshot share of accounts whose realised return exceeds the market (left pair) or whose CAPMαis strictly positive (right pair). Beats market by return (%) Positive CAPM...

    work page 2025

  22. [22]

    unstructured storage

    shows predicted returns swinging by tens of percentage points across the sample period. The per-month median (red line) tracks the broader crypto market cycle. The model is not learning which accounts allocate well, rather, which months were good or bad to be in the market. −20 pp 0 pp 20 pp 40 pp May 2020 Nov 2020 May 2021 Nov 2021 May 2022 Nov 2022 May ...

    work page 2020