Systematic, regime-aware capital management. No discretion. No emotions. Pure mathematical precision applied to gold and G7 currency markets. Absolute returns fund with superior drawdown controls. We do not benchmark — we target absolute returns.
Markets are not random. They exhibit measurable statistical properties — persistence, mean-reversion, volatility clustering — that change over time as underlying regimes shift. The challenge is not finding these properties. The challenge is detecting regime transitions in real time, before they complete.
Bruce Tech was built on a single conviction: that quantitative methods, properly applied, can identify the current market regime and deploy capital with a defined, mathematically-grounded edge. Not every bar. Not every day. But consistently, with compounding precision over time.
Our 0/35 fee structure is not a marketing decision. It is a statement of alignment. If we cannot generate returns, we should not be paid.
A Singapore-based quantitative asset management firm founded by an institutional practitioner with more than two decades of cross asset-class, cross exchange live market experience on-pit and electronic.
Bruce He brings over 20 years of institutional finance experience to Bruce Tech — spanning hedge fund management, quantitative strategy development, and financial education at NUS, NTU, and AIT.
His background includes lobbying experience at parliamentary, embassy, and exchange levels across multiple jurisdictions, providing a rare combination of market mechanics and regulatory fluency. He was among the first to unify all 32 moving average types into a single machine-level coding framework for stock exchange software.
Bruce Tech's flagship system, Manifest · 昭明 · v10, is built on a single conviction: price is water. Where conventional analysis treats price as geometry — lines drawn on a chart after the fact — Manifest treats it as a fluid with a level, a flow, a velocity, currents and turbulence. The engine takes the market's shape, then moves with it. Twelve distinct scientific disciplines are synthesised into the water model — from control & estimation theory to fluid dynamics, extreme value theory, survival analysis, and Bayesian sequential inference — deployed through three self-calibrating trade types: mean-reversion, trend-following, and trend-reversal. A live phantom book runs continuously beside the engine, out-of-sample, so only currently-proven edge is allowed to touch capital. Live-tested on gold (XAUUSD M5); instrument-agnostic by design, extending to G7 FX and indices.
The firm operates under a singular mandate: deploy capital only when a quantifiably defined edge exists, and exit the moment it does not.
Three distinct engagement models — designed for fund allocators, high-net-worth individuals, and institutional broker partners.
A single, instrument-agnostic architecture. Tested on gold and engineered for any liquid market — forex, metals, indices, crypto.
Bruce Tech currently operates managed accounts through IC Markets, a globally recognized, regulated broker with multi-jurisdiction licensing. Client accounts are opened directly with IC Markets — custody is never held by Bruce Tech, unless specifically requested at mutual agreements.
For institutional enquiries regarding co-branded arrangements or dedicated execution infrastructure, contact our CQO directly. We are open to evaluating partnerships with regulated brokers who meet our execution quality and slippage standards.
Managed-account and sub-advisory terms for institutional partners and fund-of-funds allocators.
* Minimum term extends to 12 months once firm AUM exceeds S$5M. Early redemption is subject to a penalty; no redemption fees apply on normal exit.
Capacity is finite and stated plainly. Per instrument class, the strategy is expected to scale to approximately USD 200M. Mild alpha decay begins around USD 50M as size interacts with available liquidity; beyond it, incremental capital earns a gradually lower marginal return, and the firm manages allocation accordingly rather than pursuing size for its own sake.
On a USD 200,000 account, a 10% month is USD 20,000 of profit. Our fee is 35% of that, USD 7,000. You keep USD 13,000. And the fee accrues only above your prior equity peak: if the account is below its high-water mark, we earn nothing until it recovers and makes new profit.
The firm's growth is staged and sequenced, not opportunistic. Each stage clears its predecessor before the next begins.
The relationship is built so the allocator keeps custody and control throughout. We hold trading authority only. Here is the full lifecycle.
Custody never leaves the allocator. We hold trading authority only, report constantly, and can be switched off in a single action. We are paid last, monthly, and only on genuine new profit.
Most automated strategies decay because they react — they confirm a move only after it has happened, enter late, and surrender their edge to the crowd. Manifest is built on the opposite principle: anticipate the move, qualify the opportunity, and enter at a price worth holding.
GENERATION 5.0 · ARCHITECTURE FINALIZED · STAGED VALIDATION TO LIVE
In systematic trading, the exit gets the attention but the entry decides whether a trade can profit at all. Enter at a poor price and even a flawless exit only chooses between a small loss and a larger one. Manifest v10 is engineered so that every position begins from a position of strength — and so that being wrong costs a missed trade, not lost capital. The result is a strategy designed to stay accurate, stay adaptive, and resist the performance decay that retires most automated systems.
"Nothing in the world is softer than water, yet nothing is better at overcoming the hard and strong." — Lao Tzu
"Empty your mind. Be formless, shapeless, like water. Be water, my friend." — Bruce Lee
Conventional technical analysis treats price as geometry — lines and levels drawn on a chart after the move has already happened. We reject that. We treat price as a fluid: a body of water with a level, a flow, a velocity, currents and turbulence. The system does not impose a shape on the market. It takes the market's shape, then moves with it. Every construct in the engine is a measurable property of that water.
The equilibrium the fluid returns to. Not a fixed moving average, but a slow-moving latent level estimated beneath the noisy surface of ticks by a real-time flow estimator. Price is always measured against this level.
The engine's unit of measurement is Price Displacement Velocity (PDV): how far and how fast the water is actually moving. It replaces static volatility (ATR), self-calibrates continuously, and does not lag.
A surge of directional flow is a wave with a life cycle: it ignites, sustains, exhausts, and quiesces. The engine tracks this cycle explicitly rather than reacting to price alone.
When price is stretched far from the mean, a modelled restoring pressure pulls it back. A true reversal is the tide turning — detected by the current losing energy and turning against itself, not by a line being crossed.
In water, entering the current is easy; knowing when a wave has spent its energy is the rare and valuable skill. We concentrate the edge in the exit — in reading exhaustion, not in chasing entries. Entries select regime and direction. The exit decides whether a trade is allowed to profit.
The "price is water" framing is the firm's own conceptual model; the underlying methods are the established mathematics of the twelve disciplines listed on the Track Record page.
Manifest is not a collection of indicators. It is a closed loop in which each stage exists to make the next one work — and the integration, not any single component, is the advantage.
Each stage only works because the one before it did its job: clean sensing enables early anticipation; early anticipation enables disciplined pricing; disciplined pricing makes the exit winnable; and the learning loop keeps the whole system from decaying. It is a self-reinforcing loop, not a chain of disconnected signals — far harder to replicate than any single technique, and the reason the edge is built to last.
Manifest does not claim to predict prices — over short horizons, no system honestly can, and we will never market one that pretends to. What it does predict is the change in market conditions, earlier than conventional tools, and its self-auditing layer makes that early read accurate. Its patient-entry design makes being early safe. Earlier, accurate, and safe by construction — that discipline is the entire edge, and it is the kind that endures.
Manifest never chases. An entry is a precision shot taken only when every condition aligns and price comes to a pre-calculated level. A patient, limit-style fill from concealment. If the market does not offer the price, there is no trade. Being early costs a missed opportunity; it never costs capital. One clean shot, or no shot.
The same precision governs the exit, and this is where the edge concentrates. A position is not a wager between two fixed destinations. The exit engine reads the move continuously — every bar — and closes the instant the move has spent its energy or the thesis is broken. At the right moment, not at an arbitrary line drawn in advance.
Most systems are binary: a trade runs until it hits a full stop-loss or a full take-profit. Two outcomes, both large. Manifest is non-binary. Because it exits on exhaustion and on thesis-break, the large majority of trades close in between: a small controlled loss cut early when the thesis fails; a partial gain banked before a reversal can claw it back; an exit near breakeven when conviction fades. The hard stop-loss and take-profit stay in place as outer backstops, but they are the exception, not the rule.
Losses are cut before they mature; gains are protected before they evaporate. The outcome distribution clusters tightly around small, controlled results instead of splitting between two large extremes. The smooth, shallow drawdown the strategy targets is not luck. It is the mechanical consequence of never letting a single trade travel to its full adverse bound.
You do not wait for the wave to crash on the rocks, nor gamble that it carries you all the way to shore. You read when it has spent its force, and step off. Distribution shown is illustrative of the mechanism, not a performance projection.
The engine's default answer is no. Every candidate must clear a sequence of gates, and failing any one of them ends the trade — almost always at zero cost. The discipline is not in the trades it takes, but in the far larger number it refuses.
Cost is incurred only on the rare candidate that survives every gate and the market confirms by trading to our price. Rejection is free; that is what protects capital.
Every generation of Manifest earns live capital the same way a trade earns a position — by clearing gates. None of these stages is skipped, and each must pass before the next begins. Capital is the last thing committed, never the first. Prior generations have cleared this protocol and trade live today, independently verified via Myfxbook. v10 is the latest architectural generation progressing through the same standing roadmap.
Manifest 昭明 is an engine designed to be patient, selective, and durable — anticipating opportunity, declining the marginal, entering only at prices worth holding, and protecting capital by construction at every level.
It is the execution core of Bruce Tech's systematic platform — and every generation reaches live capital the only way we allow anything to: proven against itself, then scaled on evidence.
Confidential — prepared for discussion with prospective partners only. Not an offer or solicitation to buy or sell any security or to invest in any fund. Prior generations of Manifest 昭明 have traded live and are independently verifiable via Myfxbook; v10 is the latest architectural generation progressing through the firm's standing staged validation roadmap. Past performance is not indicative of future results. Forward-looking statements are subject to risk and may not be realized. AI Softworks LLP · UEN T25LL1150A · Singapore.
All performance data is independently verified via Myfxbook. No hypothetical backtests. No simulated results. Live accounts only.
The strategy's first generation was deployed on a live client account from inception in 2023. Non-simulated, non-demo, real monetary markets. It established the core approach that Manifest v10 re-architects and tightens. The audited account statement is summarised below.
| Metric | Value | Metric | Value |
|---|---|---|---|
| Initial capital | 1,200,000 SGD | Total trades | 4,319 |
| Ending balance | ~3,000,000 SGD | Profit trades | 3,589 (83.1%) |
| Total net profit | 3,022,919.87 | Loss trades | 730 (16.9%) |
| Gross profit | 3,539,661.58 | Short won % | 80.62% |
| Gross loss | 516,741.71 | Long won % | 85.98% |
| Profit factor | 6.85 | Largest win | 13,472.08 |
| Expected payoff | 699.91 | Largest loss | -7,412.79 |
| Maximal drawdown | 195,915.87 (10.94%) | Relative drawdown | 10.94% |
Version 1 achieved its returns with a 10.94% maximum drawdown. Version 10 re-architects the engine around a tighter risk envelope — a 5% drawdown target and a 7% complete halt — with the unified real-time exit law and full self-calibration through the phantom book. The objective of the redesign is the same edge at materially lower risk.
Version 1 results reflect a single client account statement from 2023 (initial deposit basis). Provenance and attribution of the original deployment available for compliance review. Past performance is not indicative of future results.
The real client account balance across 2023 to 2025, set against spot gold and the S&P 500 over the same period. All three indexed to 100 at the start. Gold ran hard and violently. Equities climbed and lurched. The account compounded in a near-straight line through both, capturing the moves without inheriting the volatility.
Over the same three years, spot gold delivered strong but jagged returns and the S&P lurched through its own cycles. The account climbed in a smooth, shallow-drawdown line to roughly four times its starting equity, indifferent to the swings around it. This is the visual signature of an intraday, exhaustion-driven strategy: it harvests the moves through sniper entries and surgical exits, then stands flat, so its equity curve does not inherit the market's volatility.
Client account line is the reported real-money balance from the original 2023 deployment, indexed to 100 and mapped across the reported 2023–2025 period; the source statement is trade-sequenced, so the time axis is representative. Spot gold and S&P 500 are month-end market levels, indexed to 100 over the same window. Shown to illustrate the relationship between the strategy and traditional markets; past performance is not indicative of future results.
The account against spot gold and the S&P 500 over the same three years. Higher return, comparable volatility, and a correlation to both that sits near zero. The returns do not come from the same place traditional assets draw theirs.
| Metric (2023–2025) | Strategy | Spot Gold | S&P 500 |
|---|---|---|---|
| Total return | +324% | +92% | +69% |
| Annualised return (CAGR) | 64% | 25% | 20% |
| Annualised volatility | 11.7% | 13.9% | 11.5% |
| Max drawdown | 9.8% * | 7.5% | 8.6% |
| Calmar (CAGR ÷ maxDD) | 6.5 | 3.3 | 2.3 |
| Correlation to strategy | · | −0.18 | +0.18 |
Near-zero correlation to both gold and equities. An intraday, flat-overnight strategy has no structural beta to either.
Over 2023–2025 the account returned several times what gold or equities did, at comparable volatility and a shallower path, while moving independently of both. For an allocator, that is the definition of a diversifier: a return stream that adds to the book without adding to its existing risk.
* Indicative comparison. Spot gold and S&P 500 are month-end market levels. The strategy series is derived from the reported 2023 client-account balance curve (digitised to a monthly series), not from tick-level returns; correlations are computed on monthly returns and are indicative rather than an audited statistic. Drawdowns are month-end-based and understate intra-month extremes. A measured correlation pack from the live series will be published as the record matures. Past performance is not indicative of future results.
Manifest v10 is under live, independently verifiable tracking on USD 200,000 accounts. Two representative tracking windows are shown below. Both meet the ~0.5%/day return objective while holding drawdown under 3%.
Windows A and B are short, discrete live-tracking periods (Jun–Jul 2026) on USD 200k accounts, verifiable on request via independent third-party tracking. Short-period results are not annualised here and are not indicative of future results; extended live history is accumulating under the firm's staged validation roadmap.
| Trade Type | The Water | The Trade |
|---|---|---|
| A · MR Mean-Reversion |
A stretched surge, far from the water level, losing energy. | Fade the extreme back toward equilibrium; exit as the level is regained. |
| B · TF Trend-Following |
A sustained, laminar current with measurable persistence. | Ride the current from a patient entry; hold until exhaustion signals fire. |
| C · TR Trend-Reversal |
The tide turning. A mature current that has measurably lost energy and begun to reverse. | Enter the reversal early, but only on genuine, corroborated exhaustion — never a mere pullback. |
Every position, regardless of type, obeys one real-time principle: HOLD while thesis valid and not exhausted · CLOSE if exhausted or broken · FLIP only if the prediction reverses against the position. A pullback does not close a trade; loss of a tracking signal does not close a trade. Only genuine exhaustion or a prediction reversal will.
The depth of a system shows in the breadth of the intellectual tradition it draws on correctly. Manifest synthesises twelve distinct fields of rigorous science, each contributing a different way of reading the water. The specific models, parameters and combination logic are proprietary and not disclosed; what follows is the intellectual provenance and the role each discipline plays.
| Discipline | Role in the Water Model |
|---|---|
| Control & Estimation Theory | Recover the true current beneath noisy ticks: the slow-water level and flow state. |
| Dynamical Systems & Topology | See through nonlinear price behaviour to anticipate the flow, not merely react to it. |
| Fluid Dynamics | Isolate the dominant coherent currents from surface turbulence. |
| Statistical Physics | Quantify the restoring pressure that returns stretched price to its level. |
| Point-Process Statistics | Model how activity clusters, how one surge raises the odds of the next. |
| Market Microstructure | Sense informed-flow pressure beneath the surface before it reaches price. |
| Complex-Systems Science | Early warning that a regime is about to turn: the water going still before the tide. |
| Bayesian Sequential Inference | Detect structural breaks in the flow in real time and update conviction as evidence arrives. |
| Extreme Value Theory | Respect the fat tails. Size stops and targets to real extremes, not to a bell curve. |
| Survival Analysis | Model how long a move is likely to survive: the hazard of exhaustion. |
| Information Theory & Optimal Betting | Size to the measured edge; stand down automatically when the edge fades. |
| Statistical Forecasting Theory | Hold every learned quantity to an out-of-sample standard before it touches capital. |
Analytical orthogonality is the objective: each discipline answers a structurally different question, so their agreement is meaningful and their disagreement is itself information. The disciplines are named for provenance only. The models, parameters and the way they are combined are trade secrets, and are not published.
Field observations, market commentary, and methodology notes — published in chronological order.
For investor enquiries, broker partnership discussions, or due diligence requests. All communications are handled directly by the founding team.