Answer · confidence: partial
Do crypto trading robots actually make money after fees and price slips?
Research question: What do academic papers find about whether automated or algorithmic crypto strategies earn returns after fees and slippage?
The records show that the evidence on whether automated crypto strategies earn returns after fees and slippage is thin and mixed 8. A 2022 neural-network study reports that its strategies beat buy-and-hold on Bitcoin, Ethereum, and Litecoin, but it does not say whether those returns were net of fees or slippage 7. A 2019 master's thesis reports a 27.8% net return in Euro terms, yet it also finds that not all individual strategies were profitable and that automated investing carries a high risk of loss 912. Work on Uniswap-style markets finds that slippage limits the arbitrage profit traders can actually obtain 5. The newest fact in this set is from 2022, roughly four years before October 2026, so the picture may be dated 11.
Checked 2026-10-18. Only documented and supported facts are used.
Dig deeper on this questionSources
[1] supported
In the study's commodity futures test, the optimal dynamic strategy that accounts for transaction costs performed best net of those costs, though the paper does not examine crypto markets.
"Our optimal portfolio performs the best net of transaction cost"Dynamic Trading with Predictable Returns and Transaction Costs (2009)
[2] supported
Empirical data from Uniswap V2 show that the MaxMax and ConvexOptimization arbitrage strategies capture nearly the same, maximal monetized profit.
"both strategies get the maximal and almost equal"Profit Maximization In Arbitrage Loops
[3] supported
The paper's empirical and example analysis concludes that the traditional arbitrage strategies and the MaxPrice strategy are not reliable for capturing maximal arbitrage profit.
"strategies and the MaxPrice strategy are not reliable"Profit Maximization In Arbitrage Loops
[4] supported
The paper's optimal strategy deliberately does not trade all the way to the target portfolio, because doing so would incur transaction costs that are too high.
"The optimal strategy is not to trade all the way to the aim portfolio, since this entails too high transaction costs."Dynamic Trading with Predictable Returns and Transaction Costs (2009)
[5] supported
Slippage limits the arbitrage profit that traders can actually obtain on Uniswap-style automated market makers.
"slippage that limits the arbitrage profit that traders can get."Profit Maximization In Arbitrage Loops
[6] supported
Transaction costs limit how far arbitrageurs can exploit return predictability, which the authors use to motivate the model.
"Arbitrageurs’ ability to do so is limited, however, by transaction costs, and our model provides a tractable and flexible framework for the study of the dynamic implications of this limitation."Dynamic Trading with Predictable Returns and Transaction Costs (2009)
[7] supported
In backtests on Bitcoin, Ethereum, and Litecoin, the authors report that their proposed algorithmic trading strategies outperformed a buy-and-hold strategy on both stability and returns. The source does not say whether these returns were net of fees or slippage.
"We show that the proposed trading algorithms are better when compared to the traditional buy and hold strategy in terms of both stability and returns."Stochastic Neural Networks-Based Algorithmic Trading for the Cryptocurrency Market (2022)
[8] supported
The review concludes that evidence for durable net performance of AI-driven strategies is thinner than evidence for upstream progress in prediction and related tasks.
"Evidence is thinner for durable net performance."Artificial Intelligence in Equity and Crypto Markets:
[9] supported
A master's thesis finds that algorithmic cryptocurrency trading can be profitable, with its self-developed trading system reaching a net return of 27.8% in Euro terms.
"The results show that algorithmic cryptocurrency trading can be profitable, as the developed trading system achieves a net return of 27.8% in Euro terms."Algorithmic Cryptocurrency Trading (2019)
[10] supported
Academic studies find that momentum and investor attention predict crypto returns, but the review stresses that these structured signals do not imply riskless profit that machines can extract.
"These findings support the existence of structured signals and frictions; they do not imply riskless machine-extractable profit."Artificial Intelligence in Equity and Crypto Markets:
[11] supported
A survey reports that some experiments found specific technical trading rules generate excess returns for cryptocurrency traders, but the text does not say whether these returns hold after fees and slippage.
"Some experiments showed that the use of specific technical trading rules allows generating excess returns, which is useful to cryptocurrency traders and investors in making optimal trading and investment deci-"Cryptocurrency trading: a comprehensive survey (2022)
[12] supported
The same thesis finds that not all of the individual strategies were profitable, and that automated investing carries a high risk of loss.
"However, not all strategies win, which shows that an automated investment approach also entails a high risk of loss."Algorithmic Cryptocurrency Trading (2019)
[13] supported
Cong et al. (2023) found extensive wash trading on less regulated crypto exchanges, so volume-based signals can reflect manipulation rather than real investor demand.
"Cong et al. (2023) find extensive wash trading across less regulated exchanges, and Aloosh and Li (2024) provide direct evidence of wash trading in Bitcoin markets."Artificial Intelligence in Equity and Crypto Markets:
[14] supported
The CryptoTrade study beats some learned baselines but does not show superiority over all traditional trading signals.
"Beats selected learned baselines; does not establish superiority to all traditional signals"Artificial Intelligence in Equity and Crypto Markets: