Zerodha Kite Connect vs Nubra for India Algo APIs 2026

Why this broker API debate is trending

Retail algo trading conversations in India have shifted from just brokerage to workflow. Reddit threads repeatedly compare Zerodha, Dhan, Fyers, Upstox, Kotak Securities, Shoonya, and Angel One for API access. The common checklist is automated execution, real-time data, and reliability. Another theme is developer experience, because documentation and SDK quality decide how quickly a strategy can ship. Users also flag operational friction like daily logins and token expiry. Speed matters more for intraday and scalping styles than for positional systems. Cost sensitivity is visible, especially when APIs are free at some brokers and paid at others. The latest layer in the discussion is SEBI compliance for retail algos, where registration and approvals are now central. Nubra enters these comparisons as an SDK-first alternative focused on reducing developer workload.

What traders say matters most for algo APIs

Posts consistently frame broker choice around execution, data, and integration flexibility. Execution speed and stability show up as the first filter for active strategies. API flexibility is discussed as rate limits, WebSocket support, and order types like basket orders. Pricing is not just brokerage, but also monthly API fees and data subscriptions. Developer support is treated as a feature, because better docs reduce implementation mistakes. Traders running on VPS care about unattended authentication and the ability to recover sessions cleanly. Data quality is debated in practical terms like tick structure, symbol mapping, and whether analytics are built-in. Platform integrations with tools like Tradetron, Streak, and AlgoTest also influence choices for non-coders. Finally, compliance support is now a gating item, since retail algos need to be registered and exchange-approved via the broker.

Zerodha Kite Connect: stable stack, paid access

Zerodha’s Kite Connect is described as the most mature trading API in these discussions. It offers REST APIs plus WebSocket streaming for real-time automation. Users highlight its documentation quality and the size of the developer community. The same threads call it widely adopted by traders and fintech platforms. Pricing is a repeated drawback, with Kite Connect commonly cited at ₹2,000 per month. Separate market data and historical data subscriptions are also mentioned, often around ₹500 per month per API key. Another recurring point is the authentication loop, where a browser login and request token are used. The request token expires daily, so many workflows require re-authentication each trading day. Rate limits are discussed too, with some comparisons listing 3 orders per second as typical for Kite Connect.

Nubra API: SDK login and analytics-enriched ticks

Nubra is discussed as a contrast to the Zerodha-style build-it-yourself approach. The biggest difference highlighted is authentication without a browser loop. Posts claim Nubra can authenticate directly in code through its SDK, with support for TOTP-based login for server deployments. That matters for traders who want fully automated morning restarts. Data is positioned differently as well, with Nubra described as streaming tick-level data enriched with analytics. Reddit summaries mention live option Greeks, implied volatility, order book details, and market depth in structured objects. The comparison frames Zerodha ticks as raw exchange-level data that needs preprocessing and symbol mapping. Nubra is also described as supporting multi-leg execution via a single basket-style order. The posts describe this as reducing redundant API calls and lowering workflow complexity. These claims are presented as product capabilities discussed online, and the threads do not provide Nubra pricing details.

DhanHQ, Fyers, Angel One, Upstox: popular retail choices

DhanHQ is repeatedly described as automation-focused with modern workflows and webhook-friendly setups. Some posts call the API free, while other discussions mention a ₹500 per month charge and separate data costs around ₹500. Fyers is positioned as a free API option with a strong backtesting angle. Specifically, users cite free minute-level historical data for 1 to 2 years through the API. Angel One SmartAPI is often cited as a strong free alternative with reliable WebSocket streaming and practical rate limits. Comparisons mention 10 orders per second for SmartAPI and note that free historical data can be limited in some cases. Upstox API v2 is framed as execution-focused, with posts citing 40 to 80 ms order speed in one table. At the same time, some users note occasional lag and frequent product changes. Across these brokers, brokerage is commonly described as ₹20 per order under the discount model.

Kotak Neo and Shoonya: cost-led positioning

Kotak Securities appears in these conversations as a full-service broker that has upgraded its API ecosystem. Posts reference Kotak Smart API and Neo API with free access. A specific claim repeated in social posts is sub-50 ms execution latency and the ability to handle millions of orders daily. Kotak is also discussed for zero brokerage on API orders under specific Trade Free plans, with statutory and exchange charges still applying. Shoonya by Finvasia is brought up primarily for its low-cost model. Discussions describe Shoonya API as free and suitable for cost-sensitive algo traders. A comparison table lists brokerage as ₹5 or 0.03% in some segments, contrasting with ₹20 per order elsewhere. These points are shared as community summaries, and most threads add a caveat that broker features and pricing can change.

Snapshot table: what the threads are comparing

The tables below consolidate the specific points repeatedly mentioned in posts.

SEBI rules: legality and the new retail algo approval flow

Posts are clear that algo trading is legal in India through SEBI-registered broker APIs. A frequently cited reference is the 2012 SEBI circular permitting algo trading on NSE and BSE through brokers. The bigger change discussed is the 2023 framework for retail algos. Threads state that retail algos must be registered and approved by the stock exchange through the broker. That means you can write your own strategy, but it should not go live without registration. Brokers are described as responsible for ensuring algo compliance and risk controls. Users warn that running unregistered algos can lead to account suspension and potential SEBI action. Approval timelines are described as evolving, with some posts citing 3 to 7 business days for brokers like Fyers, Dhan, and Upstox. The practical implication is that broker support for the approval process becomes a key selection factor.

Real-world setup costs and the daily operations checklist

Several threads break down monthly costs beyond brokerage and capital. A minimal setup commonly cited is Zerodha Kite Connect plus a local PC, starting at ₹2,000 per month. A cloud setup is described as API plus a server like AWS EC2 plus a data feed, often estimated at ₹5,000 to ₹10,000 per month. Traders also call out operational chores like daily re-authentication for APIs with one-day access tokens. For VPS trading, automated login options like TOTP-based flows are discussed as a meaningful advantage. Rate limits show up in planning, especially when running multiple strategies concurrently. Historical data access also changes the stack, because limited free data can force third-party data spend. WebSocket stability is treated as a reliability baseline rather than a premium feature. Many users recommend paper trading for 2 to 4 weeks before going live. A common risk-control suggestion is starting with one lot and scaling after 50-plus live trades validate behavior.

Choosing between Zerodha, Nubra, and the low-cost APIs

The comparisons suggest there is no single best broker API for every algo trader. Zerodha is repeatedly recommended for documentation quality, community support, and a stable, widely used stack. The trade-off is cost and the daily manual authentication loop that some strategies need to operationalize. Nubra is discussed as aiming to remove friction with SDK authentication and analytics-ready market data. That positioning appears most relevant for options traders who need Greeks and multi-leg execution logic. Among low-cost APIs, Angel One SmartAPI is framed as a strong free baseline for many retail systems. Fyers stands out in these threads for historical minute data depth, which is central for backtesting workflows. Dhan is positioned as a newer, automation-friendly broker, although posts conflict on whether the API is fully free or effectively ₹500 with data charges. Upstox and Kotak are discussed mainly on execution and cost structure, with Kotak highlighted for Trade Free plans and Upstox for speed claims. Across all choices, SEBI retail algo registration and broker support for approvals is now part of the selection checklist.