The Rise of Algorithmic Retail Option Traders¶
Bibliographic record. Follow the original-source link for the publication.
| Field | Value |
|---|---|
| Primary domain | Microstructure |
| Other domains | — |
| Methods | Financial Ml, Research Methods |
| Facets | Instrument Index Options, Horizon 0Dte, Structure Vertical, Structure Iron Condor |
| Authors | Pedro Angel Garcia-Ares, Diego Amaya, Neil D. Pearson, Aurelio Vasquez |
| Published | 2026-04-17 |
| Source | SSRN Financial Economics Network |
| Identifiers | doi:10.2139/ssrn.6480379, ssrn:6480379 |
| URL | Open original source |
Editorial synthesis¶
Why it matters¶
Shared automation schedules may cluster multi-leg orders, making execution more expensive even when leg-based quotes look tighter. This is execution research, not proof of trader identity or a profitable strategy. (full_text:S43, full_text:S71, full_text:S103, full_text:S434, full_text:S447, full_text:S380)
Main author claims¶
- The authors find brief clock-time bursts concentrated in short verticals and iron condors. Timing and standardized trade designs suggest retail automation, but OPRA cannot directly identify the accounts. (
full_text:S6,full_text:S43,full_text:S71,full_text:S477) - The authors classify burst-window membership with 24 features and no clock input. A chronological 80/20 split yields holdout ROC AUC 0.6848. The label is not bot identity or future returns, and some inputs describe executed trades. (
full_text:S392,full_text:S393,full_text:S395,full_text:S412,full_text:S414,full_text:S777) - The authors report: Table 9 shows short-call-vertical synthetic quoted half-spreads falling from 0.104 to 0.095 while effective half-spreads rise from 0.039 to 0.043. The quote measure is built from leg NBBOs, not executable combo quotes; 10:00 quoted spreads widen instead. (
full_text:S103,full_text:S434,full_text:S435,full_text:S447,full_text:S788)
Data, method, or discussion scope¶
OPRA/SpiderRock data cover Cboe SPX/SPXW 0DTE, May 2022–January 2026. Main event and classifier analyses use 2024–2026 with different controls. Q2 reflects SSRN posting on April 17, 2026; the read PDF is dated August 12 and revised online August 13, not April-time information. (full_text:S209, full_text:S212, full_text:S392, full_text:S393, full_text:S735, description:S6, description:S7)
Main limitations¶
Identity is inferred, and institutional algorithms may look similar. Exact-time grouping can split genuine packages. Noisy aggregate expiration profits do not establish dependable returns, and spread units are not full account costs. Tables 1/2 use unresolved overall complex-share denominators. Preprint; not replicated. (full_text:S241, full_text:S378, full_text:S380, full_text:S434, full_text:S477, full_text:S744, full_text:S750)