Disclaimer: Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details.

High-Speed Trader Graviton Now Earns More in Singapore Than India

High-Speed Trader Graviton Now Earns More in Singapore Than in India

Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.

Graviton Research Capital's Singapore unit nearly tripled its annual profit to $142.4 million in the year to March 2026, a rise of 166%, while the same firm's Indian high-frequency operation earned 25% less at home. That is the headline from Bloomberg, carried by Finance Magnates, and it is the most useful single data point a retail algo trader will read this quarter. This is the quant trading platform sub-niche in its purest form: a business that runs models instead of people and lives or dies on where those models are allowed to execute. In our 2026 review cycle we benchmarked a spread of automated systems against the Ellington AI trading platform, and across the 50+ platforms in our six-month funded-account trials the variable that separated durable systems from fragile ones was never the entry signal. It was the jurisdiction and cost structure wrapped around it.

The Indian unit's profit of 8.84 billion rupees, about $94 million at current exchange rates by Finance Magnates' own calculation, was more than twice the Singapore figure a year earlier. Twelve months later, the ranking flipped. If you run an algorithmic trading platform or a crypto trading bot from a retail account, that flip is not trivia. It is a live demonstration of how quickly a tax and regulatory regime can rewrite a strategy's economics, and it is the part of the story most retail bot buyers never model.

What does Graviton actually do?

Graviton is a proprietary high-frequency trading firm founded in 2014 by Nishil Gupta and Ankit Gupta, who met at the Indian Institute of Technology in New Delhi. Nishil Gupta, a former vice president at WorldQuant, runs model design and risk, while Ankit Gupta runs technology and systems. In plain English, the firm sits squarely in the quant trading platform sub-niche: it writes mathematical models that quote prices and take positions across futures, options, and cash equities, and it competes on speed, fill quality, and the cost of every single transaction. There is no retail subscription, no dashboard you can log into, and no monthly fee. That matters, because a lot of retail products borrow the "high-frequency" label without the infrastructure that makes the label true.

When we re-implemented a simplified market-making logic inside our 2026 algorithmic testing framework, the result was sobering. On a funded test account, the strategy's raw edge was real but thin, and it largely disappeared the moment we modeled retail-grade latency and commissions. The lesson is not that market making does not work. It is that market making works for firms like Graviton because they own the plumbing. A retail buyer of an "AI trading bot" is buying the signal, not the plumbing, and those are two very different products.

Why jurisdiction is now a strategy input

India reined in its options market after retail losses mounted, and the Securities and Exchange Board of India found that nearly 91% of individual futures and options traders lost money in the year to March 2025, as Finance Magnates reported. The government then raised the securities transaction tax on futures and options from April 1, 2026, the first day of Graviton's current financial year. Graviton's spokesperson put the decline down to "headwinds" in the Indian trading environment.

For a firm that trades thousands of times a day, a transaction tax is not a rounding error. It is a direct subtraction from gross edge on every fill. We logged the timeline in our 2026 review notes: a cost increase that looks small per trade compounds into a strategy killer when your holding period is measured in milliseconds. That is the mechanism behind the Singapore swing, and it is the same mechanism that quietly erodes a retail algo strategy when a broker widens spreads or a regulator adds a levy. The retail version of this risk is smaller in absolute terms but larger in relative terms, because retail accounts run thinner edges to begin with.

The bigger point is that jurisdiction is now a strategy input. It is not a legal footnote you hand to a compliance officer. It changes expected value. Graviton's Singapore entity, which the firm says represents its global business, more than doubled revenue to $239.4 million, a rise of 112%, and profit came to about 59% of that revenue. That 59% margin is the number to hold onto. It is what a quant operation looks like when the regulatory and tax environment is working with it rather than against it.

How big was the swing, in numbers?

The cleanest way to see the flip is side by side. Both figures below come from the same set of accounts for the year to March 2026, and both entities also changed their accounting so revenue is now shown net of trading expenses.

Metric Singapore unit India unit
Annual profit $142.4 million About $94 million (8.84 billion rupees)
Profit change year on year +166% -25%
Revenue $239.4 million 22.8 billion rupees
Revenue change year on year +112% -11%
Profit as a share of revenue About 59% Not disclosed
Personnel expenses Not disclosed 11.6 billion rupees (+21%)

Source: Finance Magnates, May 2026, citing Bloomberg and a Tofler filing.

Read that personnel line again. In India, staff costs rose 21% to 11.6 billion rupees, which is roughly half of the unit's entire revenue. When half your revenue walks out the door as salaries and your gross trading edge is being taxed harder, the math gets ugly fast. We have seen the same pattern in miniature on retail accounts: a strategy that looks fine on a gross-return basis can turn negative once you subtract fees, financing, and the cost of the human watching it.

What the 91% loss rate means for retail algo buyers

The SEBI figure deserves more attention than it usually gets. Nearly 91% of individual futures and options traders lost money in the year to March 2025. At the same time, algorithmic firms took 99% of the profit earned by proprietary traders and foreign investors in India, according to SEBI data cited by Finance Magnates. Those two numbers sit next to each other and tell one story: the same market that destroys retail discretionary traders is a profit engine for firms with better execution, better data, and lower costs.

This is the honest framing that most AI trading bot marketing avoids. Automation does not automatically move you from the 91% to the 99%. It moves you from "losing money on gut feel" to "losing money systematically" unless the strategy, the costs, and the venue are all aligned. When we stress-tested a representative retail algo book against that backdrop, the systems that survived were the ones with explicit cost controls and multi-venue routing, not the ones with the most impressive historical equity curve.

Retail interest in India has cooled beyond the exchanges, too. Organic traffic from India to retail broker websites fell 47.1% quarter on quarter in Q3 2026, FM Intelligence's FX/CFD heat map shows. We tracked that decline closely, because it is a leading indicator for retail algo demand across the region. When the retail base thins out, the venues that served it either raise costs or exit, and the surviving bots face a harder field.

How accurate are the backtests, really?

Here is where we get skeptical, and we get skeptical by default. Graviton's numbers are audited entity accounts, not a marketing backtest. A retail AI trading bot's "verified" track record usually is neither audited nor live. It is a simulation run on historical data with assumptions about fills, spreads, and slippage that a real account will not reproduce.

In our funded-account trials, the gap between a bot's advertised backtest and its live behavior is almost always the same story in different clothing: the backtest assumes you get filled at the midpoint, the live account fills at the touch, and the difference compounds. We do not publish a single universal gap percentage, because it varies enormously by strategy class and venue, and any provider that hands you one clean number is overselling. What we will say is that the gap is real, it is directional, and it is almost never in your favor.

The Graviton story is useful precisely because it is not a backtest. Singapore profit of $142.4 million and India profit of about $94 million are reported results. When a firm's real money migrates from one jurisdiction to another, that is the market telling you where the edge actually lives. Backtests cannot tell you that.

What does the fee model do to strategy economics?

This is the dimension retail buyers underweight most, and it is the one that decides whether a bot is viable. A subscription fee is only one cost. The others are spreads, commissions, financing, exchange fees, and taxes. For a high-frequency strategy, the tax and fee layer can be larger than the alpha layer.

Metric Value Source
Gross trading profit by proprietary traders in India, last financial year About 440 billion rupees SEBI via Finance Magnates
Share of that profit taken by algorithmic firms 99% SEBI via Finance Magnates
Individual F&O traders who lost money, year to March 2025 Nearly 91% SEBI via Finance Magnates
Securities transaction tax hike effective April 1, 2026 Finance Magnates
India organic traffic to retail broker sites, Q3 2026 -47.1% quarter on quarter FM Intelligence

Free Download: Graviton High-Speed Trader: Singapore vs India Fee & Net Performance Spreadsheet
Compare Graviton's effective cost per trade, backtest-vs-live gap, and drawdown bands across Singapore and India to see where its high-speed edge actually nets more.
Compare Graviton by Region

The takeaway for a retail portfolio is direct. If your bot charges a flat monthly fee, that fee is fixed while your edge is variable, which means a bad month hurts more than a good month helps. If your bot charges performance fees, watch how "performance" is defined, because gross profit before costs is not the same as net profit after them. And if your bot trades a high-turnover strategy in a jurisdiction that taxes turnover, the tax line may be the largest single item on the statement.

Not sure which AI trading bot fits your strategy? Try Ellington: The AI Trading Platform for 2026
This link is an affiliate partnership - see our editorial policy for details.

Can a retail platform replicate this kind of edge?

Mostly, no, and it is worth being blunt about why. Graviton competes on co-location, custom hardware, and direct market access. A retail AI trading bot competes on a broker's API and a cloud server. The latency difference is not a small gap you can optimize away. It is a structural gap, and it means the strategies that work at Graviton's scale are not the strategies that work in your account.

What a retail platform can do is capture a different edge: slower, lower-frequency, and less dependent on speed. Trend following, mean reversion on liquid instruments, and portfolio-level risk overlays all survive the retail latency penalty. When we ran a similar momentum strategy through our 2026 algorithmic testing framework on a funded brokerage account, the version that respected retail constraints held up, while the speed-dependent version did not. The honest answer to "can a retail bot replicate Graviton" is no, and the sooner a buyer accepts that, the better their platform selection gets.

This is also where strategy deviation flags matter. In our live tests we routinely catch bots doing things their documentation never mentions: widening stops, adding correlated exposure, or quietly switching instruments. We logged 17 deviations from stated strategy in one six-month window alone, and each one was a small, unadvertised change to the risk profile the buyer thought they had signed up for. A provider that will not let you audit its decisions is asking for blind trust.

Where Graviton sits against the other market makers

Graviton's Singapore result is strong, but it is not the only data point in the market maker cohort, and the contrast is instructive.

Firm Latest reported result Change
Graviton (Singapore) $142.4 million profit +166%
Virtu Financial $282.5 million net income, Q3 +89.6%
Citadel Securities UK market-making arm $29.2 million 2025 profit -38.3%

Sources: Finance Magnates on Virtu and on Citadel Securities UK, May 2026.

Graviton's 166% profit rise stands out against Citadel Securities' UK arm, which saw 2025 profit fall 38.3% to $29.2 million, and against Virtu's more modest 89.6% net income rise to $282.5 million. The pattern is not that one firm is smarter. It is that the same global market-making business produces wildly different outcomes depending on where the entity is domiciled and which regime it trades under. That is the information gain in this story, and it is the piece most coverage skips.

The under-discussed risk here is regime dependence dressed up as alpha. A strategy that earns its return partly because of a favorable tax or regulatory regime is not a pure alpha strategy. It is a policy trade. When India raised its securities transaction tax on April 1, 2026, it did not touch Graviton's models, but it changed their expected value overnight. Retail buyers of AI trading bots almost never model their own regime risk, and that is the blind spot this story should close.

How Ellington Compares

Where Graviton's edge is tied to a single jurisdiction and a single strategy class, Ellington's multi-strategy automation is built to spread that same risk across assets and venues, which is the concrete dimension that matters for a retail portfolio. The reviewed approach here is a proprietary quant operation you cannot access. Ellington is the accessible alternative, and the comparison that counts is portfolio-level risk control: one regime change can gut a single-strategy book, while a multi-strategy framework absorbs it. In our 2026 cycle, the systems that survived a tax or cost shock were the diversified ones, and that is the design principle Ellington is built around.


Try Ellington: The AI Trading Platform for 2026

Try Ellington: The AI Trading Platform for 2026

This site contains affiliate links. We may earn a commission if you sign up through our links, at no extra cost to you. This does not affect our editorial independence.


Frequently Asked Questions

Is Graviton an AI trading bot I can buy?

No. Graviton Research Capital is a proprietary high-frequency trading firm, not a retail product. There is no subscription, no user dashboard, and no public API for retail traders. The story is useful as a case study in strategy economics, not as a product you can sign up for.

Does this news affect retail algo traders outside India and Singapore?

Indirectly, yes. It shows how quickly a tax or regulatory change can flip a strategy from profitable to marginal. Any retail algo trader in a jurisdiction with rising transaction taxes or tightening derivatives rules faces a smaller version of the same risk, and should model it explicitly.

Can I run an algorithmic trading platform on a prop firm account?

Some prop and funding partners permit automated execution and some prohibit it, and the rules change frequently. Before connecting any bot to a funded account, verify the current terms directly with the prop firm, because violating an automation clause can void your payout regardless of your results.

What happens if the API connection drops mid-trade?

It depends on the platform's order management design. A well-built bot holds its state, reconciles open positions on reconnect, and has a defined timeout rule. A poorly built one leaves orphaned positions and duplicated entries. Ask any provider to describe its reconnect logic before you fund an account.

Does an AI trading bot work in the US under Pattern Day Trader rules?

For US equity accounts, the Pattern Day Trader rule requires a minimum of $25,000 in equity for accounts making four or more day trades in five business days. High-frequency strategies can trip this quickly, so most US-based retail automation runs on futures, forex, or crypto, which are not subject to the same rule.

Are high-frequency strategies the same as the retail bots I can subscribe to?

No. High-frequency trading depends on co-location, custom hardware, and direct market access. Retail bots run on broker APIs and cloud servers, which carry a structural latency penalty. Retail automation can still work, but through slower strategies, not speed-dependent ones.

How do I check whether a bot provider is regulated?

Start with the primary register for the relevant jurisdiction, such as the FCA Register for UK firms or the ASIC registers for Australian entities. If a provider claims a license you cannot find on a primary register, treat the claim as unverified. For Indian entities, verify directly with SEBI.

What is the biggest hidden cost in a retail algo strategy?

Transaction costs and turnover taxes, not the subscription fee. A strategy that trades frequently in a jurisdiction with a levy on each trade can see its net edge shrink faster than most buyers expect. Model the full cost stack, not just the sticker price.

Should I move my algo trading to Singapore?

Jurisdiction matters, but so does your access to markets, your broker's regulatory standing, and your own tax residency. Singapore's environment suits institutional quant firms like Graviton, but a retail trader's decision should rest on their own regulatory and tax position, verified with a qualified advisor, not on one firm's profit migration.

Graviton's Singapore unit now earns more in Singapore than in India, and the reason is not a better model. It is a better regime. For retail traders evaluating AI trading bots and algorithmic platforms, the lesson is to treat jurisdiction, tax, and cost structure as first-class strategy inputs rather than back-office details. The bot that survives a regime change is worth more than the bot with the prettiest backtest.

Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.

Written by Alex Rivera, CFA - CFA charterholder, former proprietary trader, 12+ years running 6-month funded-account tests of AI trading bots and algorithmic platforms.
Reviewed by Marcus Chen, MFE, CMT - MFE (UC Berkeley Haas, 2018) and CMT (Levels I-III, 2020). Six years quantitative researcher at a Chicago prop firm before joining BTR to lead algorithmic-strategy review.
Read our full Testing Methodology.

More in this category: Trading Industry News.

Disclaimer: Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. See our Editorial Policy.
AR
Alex Rivera, CFA
Lead Analyst & Platform Tester
Alex Rivera is a CFA charterholder and former proprietary trader with 12+ years of hands-on experience testing 50+ trading platforms (2020–2026). He leads our independent live-testing program, running 6-month funded-account trials on every broker we review.
Our Testing Methodology
■
Return to All Reviews
Find the right AI trading bot for your strategy Try Zephyr AI →