Visibility graphs can make money in financial markets

Rafa{\l} Rak

Authors on Pith no claims yet

Pith reviewed 2026-05-09 13:42 UTC · model grok-4.3

classification 💻 cs.CE physics.data-anq-fin.TR

keywords tradingstrategyvgrsivisibilityanalysisassetdji30financial

0 comments

The pith

VGRSI derived from visibility graphs on asset prices generates trading signals yielding ~$340,000 total profit across DJI30, EUR/USD and XAU/USD in 2024-2025 backtests with Sharpe ratios of 2.55-3.6.

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

Financial prices form time series that can be turned into graphs by connecting points that are visible to each other without obstruction from other points. The authors build a relative strength index variant on these visibility connections instead of the usual price momentum calculation. They rescale the resulting value to the familiar 0-100 range and use it to trigger buy and sell orders in an automated strategy. The strategy is tested by repeatedly optimizing its parameters on the most recent 30 trading days and then applying them to the next 7 days, across three different markets over roughly two years. Reported outcomes include hundreds of thousands of dollars in simulated profit, moderate drawdowns, and high risk-adjusted returns. The core idea is that the geometric structure captured by visibility graphs reveals patterns traditional indicators miss.

Core claim

The strategy based on VGRSI signals generated a profit of USD~146,000 for DJI30, USD~69,000 for EUR/USD, and USD~125,000 for XAU/USD. This gives a total result of USD~340,000, which corresponds to an average profit of USD~676 per trading day, with a fixed investment of USD~1,000 to open a single trade.

Load-bearing premise

That the profitability observed in the 503-day 2024-2025 window with repeated 30-day parameter optimization will persist in future unseen market regimes without substantial overfitting or data-snooping bias.

read the original abstract

Traditional technical analysis indicators, although widely used by market participants, are often not sufficiently effective. We propose the Visibility Graphs Relative Strength Index (VGRSI), based on backward visibility relations in the price of a financial instrument. Rescaled to the 0--100 range, it can generate profitable trading signals. The performance of the indicator was evaluated using an automated trading strategy based on a 30-day optimisation window and a 7-day test window for three instruments representing different asset classes: DJI30, EUR/USD and XAU/USD over the 2024--2025 period (503 trading days). The strategy based on VGRSI signals generated a profit of USD~146,000 for DJI30, USD~69,000 for EUR/USD, and USD~125,000 for XAU/USD. This gives a total result of USD$\sim$340,000, which corresponds to an average profit of USD$\sim$676 per trading day, with a fixed investment of USD~1,000 to open a single trade. For all three assets, the strategy generated substantial profits while maintaining a moderate drawdown (10--18\% relative to a portfolio value of USD~10,000), a relatively low trading intensity (3.3--4.8 trades per day) and high Sharpe ratio values (2.55--3.6). These results indicate that VGRSI constitutes a promising technical analysis tool that goes beyond the classical trend-following approach by exploiting the geometric properties of asset price fluctuations.

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.

Desk Editor's Note private letter to a colleague

VGRSI is a new visibility-graph take on RSI that produces clean signals, but the reported profits rest on rolling 30-day parameter fits that make the edge hard to trust.

read the letter

This paper introduces VGRSI, a visibility-graph version of RSI, and reports that a trading strategy using it made around 340k USD profit across three assets in a 503-day backtest. The new element is turning visibility relations into a 0-100 oscillator for signals. That part is a fresh combination. The paper does a clear job laying out the graph construction and the rescaling step. It applies the same method to equity index, currency pair, and commodity futures, which is good for showing generality. The strategy details are explicit enough that someone could try to reproduce the signals. The soft spots are bigger. The optimization runs on 30-day windows rolling forward every 7 days, so parameters get refit to recent data before each test segment. This setup makes the high Sharpe ratios and low drawdowns partly the result of adaptation rather than a static rule based on the graphs. Without a fixed-parameter run or a longer unseen test set, the profits do not strongly support that the visibility graph itself provides a persistent advantage. Transaction costs and slippage are not addressed, and with several trades per day they would reduce the net returns. No p-values or bootstrap tests appear to check against a null of no edge. The whole evaluation sits in one recent period. This is for readers who follow technical analysis innovations or graph methods in time series. A quant looking for new indicators might experiment with the VGRSI definition. It deserves peer review because the core idea is original and the code-like description allows checking, but the authors would need to add robustness tests to make the profitability claim hold up. I would not bring it to reading group for trading advice, but the method section could spark discussion on alternative signal constructions.

Referee Report

4 major / 2 minor

Summary. The paper introduces the Visibility Graphs Relative Strength Index (VGRSI), constructed from backward visibility relations in asset price series and rescaled to the 0-100 range. It evaluates an automated trading strategy that uses VGRSI signals, with parameters optimized over rolling 30-day windows and tested on subsequent 7-day windows, on DJI30, EUR/USD, and XAU/USD over 503 trading days in 2024-2025. The strategy is reported to generate aggregate profits of approximately USD 340,000 (USD 146k, 69k, and 125k per instrument), average daily profit of USD 676 on USD 1,000 per trade, Sharpe ratios of 2.55-3.6, drawdowns of 10-18%, and trading intensity of 3.3-4.8 trades per day.

Significance. If the profitability claims prove robust after correcting for optimization bias, transaction costs, and limited sample scope, VGRSI could offer a geometrically grounded alternative to classical RSI that exploits visibility-graph properties of price fluctuations. The work provides concrete backtest numbers and Sharpe ratios on three asset classes, which is a strength for reproducibility if code and exact parameter sets were released; however, the current evaluation design does not yet establish that the edge is attributable to the visibility-graph construction rather than repeated in-sample tuning.

major comments (4)

[Evaluation procedure] Evaluation procedure (abstract and methods): The reported profits and Sharpe ratios are obtained after optimizing VGRSI signal thresholds and scaling parameters inside each rolling 30-day window before evaluating on the next 7 days. This walk-forward scheme with short windows and contiguous 2024-2025 data only means the headline USD 340k aggregate profit and 2.55-3.6 Sharpe values incorporate in-sample fitting; a fixed-parameter baseline or longer out-of-sample hold-out is required to separate the contribution of the visibility-graph construction from data-snooping.
[Results section] Backtest assumptions (results section): No transaction costs, slippage, or bid-ask spreads are included despite 3.3-4.8 trades per day. With a fixed USD 1,000 per trade and reported daily profit of USD 676, even modest per-trade costs of 0.1-0.2% would materially reduce or eliminate the net profitability; the manuscript must either incorporate realistic costs or demonstrate that the edge survives them.
[Results and discussion] Statistical validation (results and discussion): The manuscript presents raw profit, drawdown, and Sharpe figures without bootstrap tests, t-statistics against a null of no predictability, or comparisons to buy-and-hold or random-signal benchmarks. Given the short 503-day window and repeated optimization, it is unclear whether the observed performance exceeds what would be expected by chance or by standard trend-following rules.
[Abstract] Generalizability (abstract): The entire evaluation is confined to a single 503-day period (2024-2025) across three instruments. No multi-year historical tests, regime-shift analysis, or cross-market validation are provided, leaving open whether the VGRSI edge persists outside the specific market conditions of the test window.

minor comments (2)

[Methods] Notation for VGRSI construction: The precise definition of the visibility-graph degree or link count used to compute the index (and how it is rescaled to 0-100) should be stated with an explicit formula or pseudocode to allow independent replication.
[Results] Figure clarity: Performance equity curves and drawdown plots would benefit from explicit labeling of the optimization versus test segments to illustrate the rolling procedure.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The claim rests on the unstated assumption that visibility-graph geometry adds predictive power beyond classical RSI and on the empirical performance of an optimized trading rule over a limited recent window.

free parameters (2)

VGRSI signal thresholds and scaling parameters
Chosen or optimized to produce the reported trading signals and profits.
30-day window optimization parameters
Fitted inside each rolling window to maximize performance on subsequent 7-day test periods.

axioms (1)

domain assumption Backward visibility relations in price series contain exploitable predictive structure not captured by standard technical indicators.
This premise underpins the definition of VGRSI and the expectation that it will outperform classical trend-following approaches.

pith-pipeline@v0.9.0 · 5571 in / 1552 out tokens · 93583 ms · 2026-05-09T13:42:43.143466+00:00 · methodology

Visibility graphs can make money in financial markets

Core claim

Load-bearing premise

discussion (0)