Build1 distinct publisher3 min readUpdated
Two-factor authentication on automated orders and an audit trail of every automated decision took effect on April 1, 2026, and the broker carries the obligation.
The Engineer · Build desk

Compiled by The EngineerSomething wrong?How this is made
SEBI's algo trading norms took effect on April 1, 2026, requiring brokers to implement two-factor authentication for algo orders and to maintain audit trails of every automated decision [1]. The rules separate the retail participant who codes a strategy from the broker who executes it, and they put the broker on the hook for the plumbing [2] - which means the log you kept for your own debugging is now somebody's regulatory evidence.
One caveat on provenance before the mechanics. The account used here is a dev.to post by Shakti Tiwari dated 2026-08-15, published as educational material and explicitly not investment advice, which attributes the substance of the rules to livemint and Liquide coverage rather than to circular text [1][7]. Read what follows as reported, not as read off the regulator's page.
The 2FA requirement, as described, means an automated strategy cannot fire without an authenticated session behind it, with the stated intent of stopping runaway bots and unauthorized scripts [3]. The engineering consequence is that session lifecycle becomes a production dependency with the same blast radius as your market data feed. A strategy that assumed a long-lived credential now has a component that can expire mid-session, and an expiry during an open position is not a failed login, it is an unmanaged book. If you run algos, the practical checklist from the same source is to confirm your broker supports the new flow, keep your own logs alongside the broker's, and treat the 2FA session as a pre-trade check rather than a setup step [5].
The audit-trail mandate is the more interesting half, because it specifies a schema. Every algo order is to be logged with its inputs, time, and outcome [4], which is the minimum record needed to reconstruct why an order existed rather than merely that it did. That is the difference between knowing a strategy lost money and knowing which input it was looking at when it decided to [4]. Firms that already keep this for backtest reproducibility absorb the requirement by existing; firms that log fills and nothing else are now building a data pipeline under a deadline that has already passed.
What the source does not carry matters as much. There is no stated log retention period, no enforcement or penalty schedule, and no threshold defining which orders count as algorithmic [9], and no enforcement action is described [10]. Those three gaps determine cost: retention sets storage and export obligations, the algo definition sets scope, and the penalty schedule sets how carefully the first two get read. The same post frames the FY26 agenda - loss curbs, an STT hike, and the algo rules - as one theme of making uninformed trading visible and frictionful [6], which is the author's reading rather than a regulatory statement.
As of the article's date the rules had been in force 136 days [8]. Watch three things: whether the trail is expected to cover rejected, modified, and cancelled orders or only accepted ones, since the described schema mentions outcomes but not order lifecycle events [4]; whether brokers implement 2FA as a session that survives a trading day or one that must be renewed inside it [3]; and what retention window emerges in practice, because that number sets the real build [9].
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
SEBI's algo trading norms took effect April 1, 2026, requiring brokers to implement two-factor authentication for algo orders and maintain audit trails of every automated decision, per livemint and Liquide coverage as relayed by a dev.to post by Shakti Tiwari.
The rules separate the retail participant who codes their own strategy from the broker who executes it, and put the broker on the hook for the plumbing.
Two-factor authentication on algo orders means an automated strategy cannot fire without an authenticated session behind it; the stated intent is to stop runaway bots and unauthorized scripts.
The audit-trail mandate means every algo order is logged with its inputs, time, and outcome, so a losing run can be reconstructed to establish what the strategy saw and why it acted.
The author advises traders running their own algos to confirm their broker supports the new flow, keep their own logs alongside the broker's, and treat the 2FA session as part of the pre-trade checklist.
The author characterises SEBI's FY26 agenda as loss curbs, an STT hike and the algo rules, describing them as a single theme of making the cost of uninformed trading visible and frictionful.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Single secondhand relay, no primary document
All substantive content traces to one dev.to explainer that attributes the rule description to livemint and Liquide without quoting or linking a SEBI circular, clause, or effective-date notification. No second publisher, no regulator text, and no independently checkable figure appears in the cluster. The rule's headline requirements are internally coherent and uncontested, but the specification-level facts an assessor would need — retention period, algo-order threshold, penalty schedule — are absent, and the author's own disclaimer ('every figure is attributed to a reported source') is itself the only provenance guarantee offered.
No implementation or usage evidence
The supplied source establishes only that the norms are in force; it reports no broker that shipped the 2FA flow, no audit-trail deployment, no algo-order volumes, no compliance filings, and no enforcement or supervisory action in the 136 days since the effective date. Nothing in the cluster measures real-world uptake, so no adoption value is inferred.
Interpretive framing outruns the sourcing
The substantive claims are modest and plausible, but the presentation is inflated relative to what is evidenced: the topic is called a 'front-page regulatory story', the mandate is elevated to a systemic 'single theme' of FY26 policy, and each rule is mapped onto a layer of the author's own quant stack as 'confirmation of the method taught here'. That interpretive superstructure sits on one secondhand paragraph of rule description with zero implementation, enforcement, or cost evidence. The gap is moderate rather than severe because the underlying factual assertions are narrow and the piece carries an explicit educational-only disclaimer instead of a trading claim.
Self-referential methodology promotion
The source has a visible and undisclosed promotional interest: it repeatedly routes the regulatory story back to 'the governed pipeline this site already teaches', cross-sells named companion articles on XGBoost workflows, volatility surfaces and risk limits, and states that 'the entity behind this work is defined by doing it first, voluntarily, and citing the rules as confirmation rather than surprise'. The regulator is cast as validating the author's method. This is content-marketing incentive rather than commercial conflict over a traded instrument, and the educational-only, not-investment-advice label is present, which keeps the score below the top band.
Low: one publisher, no corroboration, no primary text
Confidence is constrained by cluster structure rather than by internal inconsistency. One publisher, one item, secondhand attribution, no primary regulatory document, no adoption or enforcement observations, and a self-promotional framing all limit how much weight the assessment can carry. The direction of the story — 2FA plus audit-trail duties placed on brokers from 1 April 2026 — is coherent and uncontested within the cluster, which supports a non-trivial floor.
build
Force the tool call, then hand Lightsail a long-lived key1 distinct publisher
build
AI-written code fails the same four ways, and every gate you own reports green1 distinct publisher
build
CSA's 2026 threat list is a flat line, so ask which threats a config snapshot can prove1 distinct publisher
build
An empty array is a claim about your query: verify identifiers before you trust the metric1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
dev.to
1 article · August 14, 2026