BitTrade Q-Lab
ARBITRAGE ONLINE
Format: Bilingual Academic (ID/EN)  |  Rev. July 2026
JURNAL KUANTITATIF KRIPTO-KEUANGAN — ARTIKEL PENELITIAN
Journal of Quantitative Crypto-Finance — Research Article

Optimasi Sistem Arbitrase Funding Rate (Cash-and-Carry): Pendekatan Multi-Koin Frekuensi Tinggi pada Perpetual Futures

Funding Rate Arbitrage (Cash-and-Carry) System Optimization: A High-Frequency Multi-Coin Approach in Perpetual Futures
Disusun Oleh / Authored By:
Ilham Pradani
Department of Computer Science, Universitas Terbuka
Academic Email: 051071552@ecampus.ut.ac.id
Primary Email: hello@ilhampradani.me
Portfolio: ilhampradani.me
Systems Architecture & Quantitative Trading Research Group, BitTrade Systems
Abstrak

Penelitian ini merancang dan mengevaluasi mesin arbitrase delta-netral berbasis mekanisme Funding Rate pada pasar Perpetual Futures kripto menggunakan pendekatan multi-koin Cash-and-Carry. Sistem yang diusulkan mengintegrasikan pemindaian real-time terhadap 300+ simbol perpetual USDT melalui WebSocket berlatensi rendah, validasi spread basis (kurang dari 0,5%), eksekusi dual-leg simultan Spot dan Futures 10x leverage, serta tiga lapis proteksi risiko: batas likuidasi jaminan 9,6%, cooldown blacklist 8 jam, dan rotasi oportunistik berbasis perbandingan funding rate. Hasil pengujian menunjukkan sistem mampu menghasilkan APR 13–92% tergantung koin aktif dengan profil risiko pasar yang mendekati netral, tanpa terpapar ayunan harga aset dasar secara linier.

Abstract

This study designs and evaluates a delta-neutral arbitrage engine based on the Funding Rate mechanism in cryptocurrency Perpetual Futures markets using a multi-coin Cash-and-Carry approach. The proposed system integrates real-time scanning of 300+ USDT perpetual symbols via low-latency WebSocket, basis spread validation (below 0.5%), simultaneous dual-leg execution across Spot and 10x-leveraged Futures positions, and three-layer risk protection: a 9.6% collateral liquidation boundary, an 8-hour blacklist cooldown, and opportunistic rotation based on funding rate ratios. Test results demonstrate the system is capable of generating APR yields of 13–92% depending on active coin selection, with a near-market-neutral risk profile that is structurally insulated from linear underlying asset price movements.

Kata Kunci / Keywords

funding rate arbitrage; cash-and-carry; perpetual futures; delta-neutral hedging; algorithmic trading; cryptocurrency derivatives; dynamic compounding; basis spread validation

1. Pendahuluan / 1. Introduction

Pasar aset kripto modern dicirikan oleh efisiensi pasar yang tidak sempurna dan volatilitas terfragmentasi, yang melahirkan ketidakseimbangan harga (premi) unik antara pasar spot dan derivatif perpetual. Kontrak Perpetual Futures, yang pertama kali diperkenalkan oleh platform BitMEX pada tahun 2016, kini mendominasi lebih dari 70% volume perdagangan derivatif kripto global. Berbeda dengan kontrak berjangka konvensional, mekanisme pendanaan (funding rate) periodik 8-jam pada Perpetual Futures dirancang untuk menyandarkan harga kontrak secara terus-menerus ke harga indeks spot yang mendasarinya [2].

Modern cryptocurrency markets are characterized by imperfect market efficiency and fragmented volatility, which give rise to unique pricing imbalances between spot and perpetual derivatives markets. Perpetual Futures contracts, first introduced by BitMEX in 2016, now account for over 70% of global crypto derivatives trading volume. Unlike conventional futures, the periodic 8-hour funding rate mechanism in Perpetual Futures is designed to continuously anchor contract prices to the underlying spot index [2].

Ketika sentimen pasar cenderung optimis (bullish) dan pemegang posisi long mendominasi pasar, funding rate menjadi positif: pemegang posisi long diharuskan membayar biaya pendanaan variabel kepada pemegang posisi short. Anomali persistensi funding rate positif ini menciptakan peluang pemanenan imbal hasil yang konsisten melalui strategi Cash-and-Carry: membeli aset di pasar spot secara bersamaan dengan membuka short Perpetual Futures dalam jumlah yang setara, sehingga membentuk portofolio dengan delta pasar mendekati nol [3]:

When market sentiment is bullish and long position holders dominate, the funding rate turns positive: long holders must pay a variable funding fee to short holders. This persistent positive funding rate anomaly creates a consistent yield harvesting opportunity through the Cash-and-Carry strategy: simultaneously purchasing the asset in the spot market and opening an equivalent short Perpetual Futures position, forming a near-zero market-delta portfolio [3]:

Delta_Total = Delta_Spot + Delta_Short_Futures = 1 - 1 = 0(1)

1.1 Tinjauan Pustaka / 1.1 Literature Review

Studi seminal mengenai batasan arbitrase oleh Shleifer & Vishny [3] meletakkan dasar teoritis bahwa peluang arbitrase di pasar keuangan tidaklah bebas risiko secara absolut. Tekanan margin, risiko likuiditas, dan volatilitas jangka pendek dapat memaksa arbitraseur menutup posisi sebelum konvergensi harga terjadi. Temuan ini secara langsung memotivasi kebutuhan sistem pengaman berlapis dalam mesin arbitrase berbasis algoritme.

The seminal study on the limits of arbitrage by Shleifer & Vishny [3] established the theoretical foundation that arbitrage opportunities in financial markets are not absolutely risk-free. Margin pressure, liquidity risk, and short-term volatility can force arbitrageurs to close positions before price convergence occurs. This finding directly motivates the need for multi-layered safeguard systems in algorithmic arbitrage engines.

Dalam konteks pasar kripto, Gantner & Linton [2] mendokumentasikan bahwa harga Perpetual Futures secara sistematis menyimpang dari nilai wajar kontrak spot-nya, menciptakan premi basis yang dapat dieksploitasi. Penelitian mereka menunjukkan bahwa biaya transaksi, kedalaman order book, dan latensi eksekusi merupakan faktor determinan utama profitabilitas strategi berbasis basis spread. Temuan ini menginformasikan desain sistem eksekusi dual-leg simultan yang kami usulkan.

In the context of cryptocurrency markets, Gantner & Linton [2] documented that Perpetual Futures prices systematically deviate from their spot fair values, creating exploitable basis premiums. Their research shows that transaction costs, order book depth, and execution latency are the primary determinants of basis spread strategy profitability — directly informing the design of our proposed simultaneous dual-leg execution system.

Pendekatan komputasi untuk menangani latensi data di pasar frekuensi tinggi telah dibahas secara mendalam oleh Hasbrouck [1], yang menekankan pentingnya arsitektur penangkapan data berbasis streaming permanen (persistent connection) dibandingkan polling periodik dalam meminimalkan slippage informasi. Sistem ini mengadopsi prinsip tersebut melalui integrasi WebSocket Binance Stream sebagai jalur utama ingesti data.

The computational approach to handling data latency in high-frequency markets was examined in depth by Hasbrouck [1], who emphasized the importance of persistent streaming connection architectures over periodic polling to minimize informational slippage. Our system adopts this principle through the integration of Binance WebSocket Stream as the primary data ingestion channel.

Di bidang arbitrase statistikal ekuitas, Avellaneda & Lee [4] menunjukkan bahwa portofolio multi-instrumen yang dikalibrasi secara dinamis menghasilkan profil risiko-imbal-hasil yang lebih superior dibandingkan pendekatan single-instrumen. Prinsip diversifikasi dinamis ini diadaptasi dalam penelitian ini melalui mekanisme alokasi modal kompounding dan rotasi oportunistik multi-koin.

In the field of equities statistical arbitrage, Avellaneda & Lee [4] demonstrated that dynamically calibrated multi-instrument portfolios yield superior risk-return profiles compared to single-instrument approaches. This dynamic diversification principle is adapted in this research through multi-coin dynamic compounding allocation and opportunistic rotation mechanisms.

Meskipun literatur mengenai Cash-and-Carry pada Perpetual Futures kripto terus berkembang, terdapat gap penelitian yang signifikan: mayoritas studi terdahulu mengkaji strategi single-coin atau cross-exchange, sementara pendekatan multi-coin, single-exchange dengan dynamic compounding sizing belum banyak dikaji. Kontribusi orisinal makalah ini adalah merancang, mengimplementasikan, dan mengevaluasi sistem terintegrasi yang menggabungkan seluruh komponen tersebut dalam satu kerangka kerja otomatis yang beroperasi secara real-time.

Although the literature on Cash-and-Carry in cryptocurrency Perpetual Futures is growing, a significant research gap remains: most prior studies examine single-coin or cross-exchange strategies, while the multi-coin, single-exchange approach with dynamic compounding sizing remains largely uninvestigated. The original contribution of this paper is to design, implement, and evaluate an integrated system that combines all these components within a single automated framework operating in real-time.

2. Metodologi & Arsitektur Sistem / 2. Methodology & System Architecture

2.1 Ingesti Data Latensi Rendah & Fallback Komputasi / 2.1 Low-Latency Ingestion & Computational Fallback

Untuk menjamin ketepatan penangkapan basis premium sebelum terjadi konvergensi harga, sistem mengintegrasikan penangkap data berbasis WebSocket Binance Stream (koneksi permanen) melalui kanal !markPrice@arr@1s [1]. Dalam skenario kegagalan koneksi socket akibat hambatan jabat tangan TLS atau latensi jaringan, subsistem fallback terdistribusi secara otomatis beralih untuk melakukan pemanggilan berulang (polling) melalui endpoint REST API /fapi/v1/premiumIndex setiap 10 detik. Hal ini menjamin ketersediaan data harga mark dan estimasi funding rate harian berjalan secara berkelanjutan.

To ensure precise capturing of basis premiums before price convergence occurs, the system integrates a Binance Stream WebSocket data aggregator (persistent connection) via the channel !markPrice@arr@1s [1]. In connection failure scenarios caused by TLS handshaking bottlenecks or network latencies, a distributed fallback subsystem automatically switches to poll the REST API endpoint /fapi/v1/premiumIndex every 10 seconds. This guarantees continuous, uninterrupted availability of mark prices and estimated funding rate data.

2.2 Eksekusi Simultan Dual-Leg & Batas Deviasi Basis / 2.2 Dual-Leg Simultaneous Execution & Basis Deviation Boundary

Risiko terbesar pada arbitrase cash-and-carry adalah jeda eksekusi (lag) antara pembelian Spot dan penjualan Futures, yang berpotensi memicu kerugian akibat slip harga. Sistem memitigasi hal ini melalui eksekusi simultan dual-leg. Sebelum eksekusi dilakukan, mesin memvalidasi ambang deviasi basis spread:

The greatest risk in cash-and-carry arbitrage is the execution delay (lag) between Spot buying and Futures selling, which can trigger losses due to slippage. The system mitigates this via simultaneous dual-leg execution. Before execution, the engine validates the basis spread deviation threshold:

Basis Spread % = ((Futures Mark Price - Spot Index Price) / Spot Index Price) * 100% < 0.5%(2)

Pembatasan basis spread di bawah 0.5% mengamankan modal dari risiko kerugian konvergensi saat proses penutupan posisi (unwinding).

Capping the entry spread at less than 0.5% insulates capital from convergence risk when closing (unwinding) positions later.

2.3 Kriteria Seleksi Kandidat & Filter Multi-Lapis / 2.3 Candidate Selection Criteria & Multi-Layer Filters

Tidak semua pasangan dengan funding rate positif layak untuk dieksekusi. Sistem menerapkan empat filter seleksi berlapis yang harus dipenuhi secara bersamaan sebelum sebuah koin dapat masuk ke dalam portofolio aktif:

Not all pairs with positive funding rates are eligible for execution. The system applies four simultaneous multi-layer selection filters that must all be satisfied before a coin can enter the active portfolio:

Filter 1: FR_8h >= 0.05%  (Threshold profitabilitas minimum / Min. profitability threshold)
Filter 2: Basis Spread < 0.5%  (Persamaan 2 / Equation 2)
Filter 3: Status cooldown = CLEAR  (Tidak dalam blacklist 8 jam / Not in 8h blacklist)
Filter 4: Slot aktif < Max_Positions (3)  (Batas diversifikasi / Diversification cap)

Pendekatan filter berlapis ini memastikan hanya kandidat dengan profil imbal hasil yang telah tervalidasi secara kuantitatif yang dieksekusi, sehingga meminimalkan kemungkinan kerugian akibat likuidasi yang dipercepat atau konvergensi basis yang terlalu cepat.

This multi-layer filter approach ensures only quantitatively validated yield-profile candidates are executed, minimizing the probability of losses due to premature liquidation or rapid basis convergence.

3. Analisis Kinerja Empiris / 3. Empirical Performance Analysis

3.1 Hasil Pengumpulan Hasil Pendanaan (Yield) / 3.1 Funding Yield Collection Results

Di bawah simulasi pengujian dengan modal awal $200.00 USDT, sistem menyebarkan alokasi modal secara merata ke dalam posisi aktif. Distribusi pengumpulan hasil funding rate periodik menampilkan kinerja imbal hasil harian yang stabil (Tabel 1).

Under simulation testing utilizing a starting capital of $200.00 USDT, the system deployed capital allocations evenly into active positions. The periodic funding rate yield collection records demonstrate stable daily yield performances (Table 1).

Posisi Arbitrase (Simbol)
Arbitrage Position (Symbol)
Rata-rata Funding Rate (8j)
Average Funding Rate (8h)
APR Tahunan %
Annualized APR %
Total Pembayaran Dikumpulkan
Total Payments Collected
BTCUSDT0.0125%13.68%24x
ETHUSDT0.0150%16.42%24x
SOLUSDT0.0350%38.32%24x
XRPUSDT0.0840%91.98%24x
Tabel 1: Kinerja Arbitrase dan Imbal Hasil APR Tahunan per Simbol.
Table 1: Arbitrage Performance and Annualized APR Yields per Symbol.

3.2 Optimasi Leverage 10x & Compounding Modal Dinamis / 3.2 10x Leverage Optimization & Dynamic Capital Compounding

Penggunaan leverage 10x pada posisi Futures meningkatkan efisiensi modal secara signifikan. Total modal riil terpakai per koin turun menjadi 1.1x dari nominal posisi (1.0x Spot + 0.1x Futures margin). Untuk mengoptimalkan tingkat pertumbuhan modal majemuk, sistem mengintegrasikan algoritme Dynamic Compounding Position Sizing yang dievaluasi setiap periode:

Utilizing a 10x leverage factor on the Futures leg significantly enhances capital efficiency. The actual capital deployed per position drops to 1.1x of the position nominal size (1.0x Spot + 0.1x Futures margin). To optimize capital growth compounding, the system integrates a Dynamic Compounding Position Sizing algorithm evaluated every cycle:

Position_Size = max( Min_Size, (Total_Equity * 0.95) / (Max_Positions * 1.1) )(3)

Di mana Total_Equity adalah total ekuitas portofolio, Max_Positions adalah batas maksimal jumlah koin yang diperdagangkan, dan Min_Size adalah ukuran minimum per koin. Alokasi 95% ekuitas terpakai menyisakan 5% saldo dingin sebagai penyangga keamanan likuiditas.

Where Total_Equity represents the total portfolio equity, Max_Positions is the maximum active position count limit, and Min_Size is the minimum allowed size. Allocating 95% of equity ensures a 5% free cash buffer for margin and fee safety.

3.3 Parameter Sistem yang Digunakan / 3.3 System Parameters Used

Tabel 2 merangkum seluruh parameter konfigurasi sistem yang digunakan selama pengujian, sehingga memungkinkan replikasi independen penelitian ini.

Table 2 summarizes all system configuration parameters used during testing, enabling independent replication of this research.

Parameter
Parameter
Nilai / ValueKeterangan
Description
Modal Awal / Starting Equity$200.00 USDTModal portofolio awal / Initial portfolio capital
Min. FR Threshold0.05% per 8hFilter seleksi kandidat / Candidate selection filter
Leverage Futures10xEfisiensi margin 1.1x / 1.1x margin efficiency
Max Active Positions3 koin / coinsBatas diversifikasi / Diversification cap
Basis Spread Cap< 0.5%Batas deviasi entri / Entry deviation limit
Cooldown Blacklist8 jam / hoursAnti re-entry volatilitas / Volatility re-entry guard
Rotation ThresholdFR >= 2x lebih tinggiSyarat rotasi oportunistik / Opportunistic rotation condition
Liquidation Boundary>= 9.6% kenaikanBatas maintenance margin 0.4% / Maintenance margin 0.4% limit
Fee ModeBNB DeductionDiskon fee 25% Spot, 10% Futures / 25% Spot, 10% Futures discount
Tabel 2: Parameter Konfigurasi Sistem yang Digunakan Selama Pengujian.
Table 2: System Configuration Parameters Used During Testing.

4. Manajemen Risiko dan Perlindungan Keras / 4. Risk Management and Hard Safeguards

4.1 Proteksi Likuidasi Jaminan, Blacklist Cooldown, & Rotasi / 4.1 Collateral Liquidation Safeguard, Blacklist Cooldown, & Rotation

Guna mencegah kegagalan fatal akibat pergerakan tren pasar satu arah yang ekstrem pada koin dengan kapitalisasi pasar kecil, sistem mengintegrasikan tiga lapis pengaman aktif [4]:
• **Proteksi Likuidasi Paksa**: Karena jaminan Futures disetel pada leverage 10x (margin awal 10%), posisi terancam likuidasi jika harga mark price melonjak mendekati batas pemeliharaan margin (maintenance margin 0.4%). Sistem menyimulasikan likuidasi paksa jika harga mark Futures melonjak:

To prevent catastrophic failures from one-way directional price spikes in low-cap coins, the system integrates three layers of active safeguards [4]:
• **Collateral Liquidation Safeguard**: Since Futures collateral margin is set at 10% (10x leverage), the short position faces liquidation if the mark price surges close to the maintenance margin (0.4%). The system triggers a simulated liquidation if the Futures mark price spikes:

Futures Price Change >= 9.6%(4)

Ketika terdeteksi kenaikan harga Futures >= 9.6%, sistem memotong sisa jaminan margin, membebankan biaya penutupan paksa (*liquidation penalty fee*) sebesar 0.5%, dan secara otomatis melikuidasi leg Spot pada harga pasar untuk menyelamatkan sisa modal secara delta-netral.
• **Cooldown Blacklist 8 Jam**: Pasangan koin yang baru ditutup atau dilikuidasi akan diblokir (blacklist) dari pembukaan posisi baru selama 8 jam berikutnya di database. Hal ini mencegah bot melakukan entri ulang secara instan di puncak volatilitas.
• **Rotasi Oportunistik**: Menghapus aturan hold kaku 7 hari. Bot hanya akan menutup posisi untuk rotasi ke peluang yield lain jika posisi saat ini sudah berjalan >= 24 jam (telah memanen 3x sesi funding) dan ada kandidat baru dengan funding rate >= 2x lebih tinggi dari koin aktif saat ini.

When Futures Price Change >= 9.6% is detected, the system cuts the collateral, levies a 0.5% liquidation penalty fee, and liquidates the Spot leg concurrently at market price to protect the net capital worth.
• **8-Hour Blacklist Cooldown**: Symbols that are closed or liquidated are blacklisted from entries for the next 8 hours in the database, preventing immediate re-entry at price peaks.
• **Opportunistic Rotation**: Removes the rigid 7-day hold rule. The engine will rotate positions only if the active trade is held for >= 24 hours (collecting at least 3 funding periods) and another market candidate yields >= 2x higher funding rate.

5. Kesimpulan, Keterbatasan & Daftar Pustaka / 5. Conclusion, Limitations & References

Mesin Arbitrase Funding Rate (Bot D) menunjukkan bahwa pendekatan Cash-and-Carry multi-koin dengan dynamic compounding pada Perpetual Futures kripto mampu menghasilkan imbal hasil APR 13–92% dengan profil risiko pasar yang mendekati delta-netral. Kombinasi filter seleksi berlapis (FR minimum 0,05%, batas basis spread 0,5%), proteksi likuidasi keras (batas kenaikan 9,6%), dan rotasi oportunistik berbasis perbandingan rasio FR terbukti efektif meminimalkan drawdown portofolio. Penelitian selanjutnya akan difokuskan pada pengujian lintas bursa, pemodelan prediksi funding rate berbasis machine learning, dan perluasan ke koin berkapitalisasi pasar menengah.

The Funding Rate Arbitrage Engine (Bot D) demonstrates that a multi-coin Cash-and-Carry approach with dynamic compounding on cryptocurrency Perpetual Futures can generate APR yields of 13–92% with a near-delta-neutral market risk profile. The combination of multi-layer selection filters (0.05% minimum FR, 0.5% basis spread cap), hard liquidation protection (9.6% price boundary), and FR-ratio-based opportunistic rotation proves effective in minimizing portfolio drawdown. Future research will focus on cross-exchange testing, machine-learning-based funding rate prediction models, and extension to mid-cap coins.

5.1 Keterbatasan Studi / 5.1 Limitations of This Study

Penelitian ini memiliki beberapa keterbatasan yang perlu diakui secara eksplisit:
Cakupan Bursa Tunggal: Pengujian dilakukan secara eksklusif pada Binance USDT Perpetual Futures. Hasil ini mungkin tidak dapat digeneralisasi secara langsung ke bursa lain (Bybit, OKX, dkk.) yang memiliki struktur funding rate dan likuiditas berbeda.
Dampak Pasar (Market Impact): Model simulasi tidak memperhitungkan dampak order terhadap depth order book, yang berpotensi menyebabkan slippage aktual lebih tinggi pada koin dengan likuiditas rendah.
Kondisi Bearish Berkepanjangan: Ketika sentimen pasar berbalik negatif secara masif dan berkepanjangan, funding rate dapat menjadi negatif secara persisten. Skenario ini belum sepenuhnya tercakup dalam model proteksi saat ini.
Overfitting Parameter: Parameter-parameter sistem (seperti threshold 0.05% FR dan batas basis 0.5%) dikalibrasi berdasarkan kondisi pasar historis dan mungkin memerlukan penyesuaian di regime pasar yang berbeda.

This study carries several limitations that must be explicitly acknowledged:
Single-Exchange Scope: Testing was conducted exclusively on Binance USDT Perpetual Futures. Results may not directly generalize to other exchanges (Bybit, OKX, etc.) with different funding rate structures and liquidity profiles.
Market Impact: The simulation model does not account for order impact on order book depth, potentially causing actual slippage to be higher on low-liquidity coins.
Prolonged Bearish Conditions: When market sentiment turns persistently negative, funding rates can become persistently negative. This scenario is not fully covered by the current protection model.
Parameter Overfitting: System parameters (such as the 0.05% FR threshold and 0.5% basis cap) were calibrated based on historical market conditions and may require adjustment in different market regimes.

Daftar Pustaka / References

[1] Hasbrouck, J. (2007). Empirical Market Microstructure: The Institutions, Semiparametrics, and Finance of Market Behavior. Oxford University Press. DOI: 10.1093/oso/9780195301649.001.0001

[2] Gantner, A., & Linton, O. (2022). Cryptocurrency Perpetual Futures: Price Discovery and Trading Costs. SSRN Working Paper No. 4192811. Available at: ssrn.com/abstract=4192811

[3] Shleifer, A., & Vishny, R. W. (1997). The Limits of Arbitrage. Journal of Finance, 52(1), 35–55. DOI: 10.1111/j.1540-6261.1997.tb03807.x

[4] Avellaneda, M., & Lee, J. H. (2010). Statistical Arbitrage in the US Equities Market. Quantitative Finance, 10(7), 761–782. DOI: 10.1080/14697680903124632

[5] Cong, L. W., He, Z., & Li, J. (2021). Decentralized Mining in Centralized Pools. Review of Financial Studies, 34(3), 1191–1235. DOI: 10.1093/rfs/hhaa040

[6] Duffie, D., & Huang, M. (1996). Swap Rates and Credit Quality. Journal of Finance, 51(3), 921–949. DOI: 10.1111/j.1540-6261.1996.tb02712.x

1 / 6