CoChat
Sign Up
Cited by 103Financial Markets and Investment StrategiesStock Market Forecasting MethodsBlockchain Technology Applications and Security

The profitability of technical trading rules in the Bitcoin market

Open access
Gerritsen, Dirk F., Bouri, Elie, Ramezanifar, Ehsan, Roubaud, David·2020·Finance Research Letters
DOI 10.1016/j.frl.2019.08.011
Open this paper in CoChat — free accountOpen PDF

Abstract

We apply seven trend-following indicators to assess the profitability of technical trading rules in the Bitcoin market. Using daily price data from July 2010 to January 2019, our main results show that specific technical analysis trading rules, mainly trading range breakout, contain significant forecasting power for Bitcoin prices, allowing the outperformance of the buy-and-hold strategy through the Sharpe ratio computed via the bootstrapping method. Results from various sub-periods, representing normal and boom markets, generally confirm our main finding and show that the added value of the trading range breakout rule delivers outperformance in strongly trending markets.

AI snapshot

Written from the paper’s full text.

TL;DR

The authors test seven families of well-known technical trading rules on daily Bitcoin prices and find that only the trading range breakout rule reliably beats a buy-and-hold benchmark, mainly in strongly trending markets. Countertrend indicators such as the relative strength index and Bollinger bands significantly underperform.

Key findings

  • Over the full sample, the buy-and-hold daily Sharpe ratio is 0.056 (annualized 1.07), while the trading range breakout rules average roughly 0.08; for SUP/RES150 the Sharpe ratio is 0.093, a difference of 3.77 basis points with a bootstrapped p-value of 0.06, and several breakout variants are statistically significant across strategies.
  • Moving average rules generally do not differ significantly from buy-and-hold (e.g., MA1-200 in Strategy 1: Sharpe 0.047, difference −0.83 basis points, p = 0.65). MACD and rate-of-change rules show significant outperformance only under the long-or-out strategy (e.g., rate of change 0.092, +3.67 basis points, p = 0.01).
  • The relative strength index and Bollinger bands significantly underperform and sometimes produce negative Sharpe ratios (RSI −0.110, difference −16.61 basis points, p < 0.01; Bollinger bands −0.060, −11.55 basis points, p < 0.01). On-balance volume lags the benchmark but not significantly.
  • In subsamples, breakout rules outperform in 2011–2012, 2013–2014 and 2017–2018 (with significance in the latter two), but underperform in the stable 2015–2016 period, though not significantly.

Methods

  • Seven rule families (moving averages, trading range breakout, MACD, rate of change, on-balance volume, RSI, Bollinger bands) drawn from Gerritsen (2016) and restricted to rules publicized before Bitcoin exchange trading to limit data snooping.
  • Daily Bitcoin price data from July 17, 2010 to December 31, 2018 (3084 observations), using Mt Gox then Coinmarketcap, excluding the June 2011 Mt Gox hack period; trading rule application begins February 2011 because 200 prior days are needed for the MA-200. Risk-free rate is the 3-month US T-bill; volume data only available from end-2013.
  • Each rule is applied under three strategies (long/short, long-or-out, and double-long/long/out), and performance is assessed by comparing daily Sharpe ratios against buy-and-hold using the Ledoit and Wolf (2008) bootstrap with 1000 resamples.

Limitations

  • The paper states that future work should examine other leading cryptocurrencies and assess the breakout rule's profitability after including transaction costs (Hudson et al., 1996); transaction costs are not incorporated in the reported results.
  • The paper notes that short positions are not feasible on many exchanges, which motivates the alternative strategies; the long/short results are therefore not directly implementable.
  • Evident from the text: volume-based rules could not be tested in the earlier subsamples because volume data start only at year-end 2013, and the subsample conclusions rest on a small number of four-year windows.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.frl.2019.08.011

Cites

28 of 28 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 9 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    For MACD rules and the rate-of-change rule, significant outperformance over buy-and-hold occurs only in Strategy 2.

    “For MACD rules and for the rate-of-change rule, we see significant outperformance only in Strategy 2.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Taking short positions in Bitcoin is not possible across many exchanges.
    • The authors state this as an assumption: When not holding Bitcoin, the investor invests in an asset that yields the risk-free rate.
  • Findingown data

    The buy-and-hold daily Sharpe ratio for the full sample period is 0.056.

    “The lowest row of the table indicates the daily Sharpe ratio of the buy-and-hold strategy for the sample period, which equals 0.056”
    Quote in the paperNot a substantial claim
  • Findingown data

    Other technical trading rules do not outperform the buy-and-hold strategy.

    “Other technical trading rules do not outperform the buyand-hold strategy.”
    Quote in the paperSubstantial claim
  • Findingown data

    For most trading range breakout rules and applied strategies, the difference in Sharpe ratio with the buy-and-hold benchmark is statistically significant.

    “For most trading range breakout rules and applied strategies, the difference in Sharpe ratio with the buy-and-hold benchmark is statistically significant.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Taking short positions in Bitcoin is not possible across many exchanges.
    • The authors state this as an assumption: When not holding Bitcoin, the investor invests in an asset that yields the risk-free rate.
  • Findingown data

    The trading range breakout rule consistently delivers higher Sharpe ratios than a buy-and-hold strategy for the full sample period.

    “Our main results show that the trading range breakout rule consistently delivers higher Sharpe ratios than a buy-and-hold strategy for the full sample period”
    • Also stated in: § 3.1. Full sample results · plain
      “The trading range breakout rule provides consistently higher Sharpe ratios than the buy-and-hold strategy.”
    Quote in the paperSubstantial claim
  • May combine two claims. 5
  • The quote only partly states this claim. 3
  • Classified as probably not a substantial claim. 8

  • May combine two claims.

    Findingown data

    The relative strength index and Bollinger bands significantly underperform the buy-and-hold benchmark, and in some instances yield negative Sharpe ratios.

    “Interestingly, the countertrend indicators relative strength index and Bollinger bands both not only significantly underperform the buy-and-hold benchmark, but in some instances even yield negative Sharpe ratios.”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingown data

    The trading range breakout rule delivers higher Sharpe ratios than a buy-and-hold strategy during periods of strongly trending markets.

    “the added value of the trading range breakout rule delivers outperformance in strongly trending markets”
    • Also stated in: § 3.2. Subsample results for trading range breakout · hedged
      “In strongly trending markets, the trading range breakout rule seems to deliver outperformance.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For on-balance volume, Sharpe ratios lag the buy-and-hold benchmark, but the differences are not statistically significant.

    “For on-balance volume, the Sharpe ratios lag those of the buy-and-hold benchmark, but the differences are not statistically significant.”
    Quote in the paperNot a substantial claim

  • The quote only partly states this claim.

    Findingown data

    The rate of change rule significantly outperforms buy-and-hold in the first two sub-periods (2011-2012 and 2013-2014).

    “The rate of change significantly outperforms in the first two sub-periods, but loses its outperformance during the last two periods.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The Bitcoin price data from Mt Gox and Coinmarketcap are accurate and representative of the Bitcoin market.
    • Rests on an assumption: The Sharpe ratio is an appropriate measure of risk-adjusted performance for comparing technical trading rules and buy-and-hold.

  • Classified as probably not a substantial claim.

    Premisereasoning

    The extent to which technical trading rules perform when applied to Bitcoin prices remains unclear.

    “Furthermore, the extent to which technical trading rules perform when applied to Bitcoin prices remains unclear.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    The comparison of Sharpe ratios is based on statistical inference using the bootstrapping method of Ledoit and Wolf (2008).

    “The comparison of the Sharpe ratios must be based on statistical inference; therefore, we use the bootstrapping method of Ledoit and Wolf (2008) to measure the statistical significance of the difference between the Sharpe ratios”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Ledoit and Wolf (2008) source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    Volume data was not available for the 2011-2012 period.

    “Volume data was not available for this time period.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The Bitcoin price data from Mt Gox and Coinmarketcap are accurate and representative of the Bitcoin market.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductcitation

    The study used the Ledoit and Wolf (2008) bootstrapping approach based on 1000 simulations to test whether the differences in Sharpe ratios between technical analysis strategies and buy-and-hold are zero, reporting bootstrapped p-values.

    “Following Ledoit and Wolf (2008), we use the bootstrapping approach based on 1000 simulations to test whether the differences in the Sharpe ratios of a technical analysis strategy and a buyand-hold strategy are zero, the bootstrapped p-value is reported.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Ledoit and Wolf (2008) source

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductreasoning

    The application of the trading rules starts from February 2011 only, because the study relied on price data from prior 200 days to compute the first value of the MA-200.

    “Although we used the entire price history as of July 2010, the application of the trading rules starts from February 2011 only.This is because we relied on price data from prior 200 days to compute the first value of the MA-200.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The profitability of specific technical trading rules, such as the trading range breakout rule, can consistently exceed that of a buy-and-hold strategy for the full sample period and during strongly trending markets.

    “In this paper, we show that the profitability of specific technical trading rules, such as the trading range breakout rule, can consistently exceed that of a buy-and-hold strategy for the full sample period and during strongly trending markets.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). The quote only partly states this claim.

    Study design and conductown data

    Volume data was not available for the 2013-2014 period.

    “Volume data was not available for this time period.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The Bitcoin price data from Mt Gox and Coinmarketcap are accurate and representative of the Bitcoin market.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The Bollinger bands (20,2) is the traditional Bollinger bands method.

    “The BB (20,2) is the traditional method (Lento, 2007).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Lento, 2007 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The 12-day EMA and the 26-day EMA are the most frequently used exponential moving averages.

    “According to Murphy (1999), the 12-day EMA and the 26-day EMA are the most frequently used.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Murphy (1999)

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

13 of 13

  • The paper relies on this without support.

    Taking short positions in Bitcoin is not possible across many exchanges.

    “In the second, we acknowledge that taking short positions in Bitcoin is in reality not possible across many exchanges.”

  • The authors state this as an assumption.

    When not holding Bitcoin, the investor invests in an asset that yields the risk-free rate.

    “we assume that the investor invests in an asset that yields the risk-free rate.”

  • The paper relies on this without support.

    The Bitcoin price data from Mt Gox and Coinmarketcap are accurate and representative of the Bitcoin market.

    “Sources: Mt Gox, Coinmarketcap.”

  • The paper relies on this without support.

    The Sharpe ratio is an appropriate measure of risk-adjusted performance for comparing technical trading rules and buy-and-hold.

    “This table shows the Sharpe ratio on daily basis of all trading rules for three trading strategies applied to daily Bitcoin prices from 2011 to 2018.”
4 of 4

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • The introduction denies outperformance reported for MACD and rate of change

    Minor

    The introduction says no other rules outperform buy-and-hold, but the results explicitly report significant outperformance for MACD and rate of change under Strategy 2.

  • Sharpe-ratio differences are mislabeled as basis points

    Minor

    The methods say the raw Sharpe-ratio differences are multiplied by 100. That converts them to hundredths of a Sharpe-ratio unit, not basis points, which require multiplication by 10,000.

  • Two tables disagree on the same 2017-2018 breakout Sharpe ratio

    Minor

    For the 2017-2018 SUP/RES200 rule under Strategy 3, Table 2 gives 0.03 while Table B4 gives 0.07. Both rows report the same difference, 2.33, and p-value, 0.04.

  • The stated OBV significance summary overlooks a significant result

    Minor

    The full-sample summary says no on-balance-volume difference is statistically significant. Table 1 marks the Strategy 1 OBV_MA1-50 difference significant at its stated 10% threshold (p = 0.06).

  • Table B2's RSI difference disagrees with its benchmark

    Minor

    The Strategy 1 RSI Sharpe ratio is −0.14 against a buy-and-hold ratio of 0.06. Their difference is about −0.20, or −20 after multiplication by 100, not the reported −13.63; rounding cannot account for the gap.

  • Table B1's Bollinger-band differences disagree with its benchmark

    Minor

    Table B1 reports a Strategy 1 Bollinger-band Sharpe ratio of −0.10 and a buy-and-hold ratio of 0.07. Their difference is about −0.17, rather than the reported −9.69 after the paper's stated multiplication by 100; the discrepancy is too large to be rounding.

6 of 6

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Gerritsen, D. F., Bouri, E., Ramezanifar, E., & Roubaud, D.
Affiliations
  • Utrecht University (NL)
  • Holy Spirit University of Kaslik (LB)
  • MBS School of Business (FR)
Corresponding author
Ramezanifar, Ehsan
ORCID
Venue
Finance Research Letters
Publisher
Elsevier BV
Published
2020-05
Volume
34
Pages
101263
ISSN
1544-6123 (ISSN-L)
DOI
10.1016/j.frl.2019.08.011
OpenAlex
W2967069906
Semantic Scholar
107ab78c2c6122e6e15185b4521b36921df04d4b
Cited by
103
References
25
Citations by year
2026: 5 · 2025: 15 · 2024: 10 · 2023: 25 · 2022: 15 · 2021: 16 · 2020: 14 · 2019: 3
FWCI
14.55
Type
journal article
Language
en
Open access
Licence
Locations
  • Finance research letters (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • Utrecht University Repository (Utrecht University) (submittedVersion, other-oa)
  • Utrecht University Repository (Utrecht University) (submittedVersion)
Topics
  • Social Sciences › Economics, Econometrics and Finance › Finance › Financial Markets and Investment Strategies
  • Social Sciences › Decision Sciences › Management Science and Operations Research › Stock Market Forecasting Methods
  • Physical Sciences › Computer Science › Information Systems › Blockchain Technology Applications and Security
Keywords
Bitcoin, technical trading rules, cryptocurrency price prediction, buy-and-hold strategy, Sharpe ratio, market trends

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-11.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-11.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review