WaveOn

Background animation: soft teal and green light drifting slowly behind the headline. It is a brand mood, not market data.

Systematic multi-asset strategies

From hypothesis to live execution, one tested gate at a time.

WaveOn designs, tests and runs rules-based strategies across options, futures, ETFs, equities and fixed income. Every idea is written as code, judged against pass/fail rules fixed before the first backtest, and traded by the same engine that tested it.

Request the strategy briefing See the seven gates

Approach

A research lab that runs like an execution desk.

Good ideas are cheap. What separates a strategy that survives from one that does not is the path between them and a broker account. WaveOn is built around that path.

Evidence before conviction

Each idea starts as a written hypothesis with a reason to work and the test that would prove it wrong. Every result is kept and counted, including the failures.

The tested code is the trading code

The backtest, the paper account and the live account run one codebase on the LEAN engine. Nothing is rewritten between research and execution.

Risk limits come first

Every strategy carries a known maximum loss per position, hard stops and a risk budget agreed before the first order is sent.

Process

Seven gates between an idea and live capital.

Each step ends in a gate with a written pass condition. A failed gate sends the idea back to research or ends it. The right-hand column shows TB3c, our options overlay, at each step.

#StepWhat happensGate to passTB3c
1
HypothesisWhy should this earn money?
What happens

State the idea, the economic reason it should pay, who is on the other side, and the objective: income, growth or protection.

Gate to pass

A falsifiable claim and a reason it should persist after costs.

TB3cOption income on top of an existing portfolio, defined-risk spreads only, every exit by rule.
2
Pre-registrationRules before results
What happens

Write down the pass/fail rules, the test window and the comparison before any backtest runs.

Gate to pass

Rules recorded in advance, so results cannot move the goalposts.

TB3cEach rule change was tested against criteria set before the run. Every alternative tried is logged.
3
DataHistory and live feeds
What happens

Load minute-to-daily history for prices, full option chains and futures contracts. The same source feeds live trading.

Gate to pass

Stress periods covered and data errors found and fixed first.

TB3cHourly SPY option quotes from January 2012, covering 2018, 2020 and 2022.
4
BacktestReplay history with costs
What happens

The rules run through history as code, day by day, paying commissions and the quoted bid/ask on every trade.

Gate to pass

Results after costs judged against the objective in the brief; losing years shown and understood.

TB3c6,444 spreads over 2012–2025 with $81,764 of fees. Every year shown, 2022 included.
5
RobustnessStress the settings
What happens

Try neighbouring settings, other sizes and instruments, sub-periods and stressed costs, one change at a time.

Gate to pass

The result must not hinge on one lucky setting.

TB3cSeven rule alternatives tested, plus SPX index options at $5M and $10M.
6
Paper tradingSame code, simulated money
What happens

The exact code trades at the broker: live quotes and real order routing, with fills simulated by the broker.

Gate to pass

Orders, fills and recovery behave as tested; every incident logged and investigated.

TB3cInteractive Brokers paper account since 21 Oct 2026. After two broker disconnects, the position book reloaded and matched the broker on restart.
7
Live executionSmall first, then scale
What happens

Start at a fraction of target size and scale in steps as real fills and costs confirm the paper results.

Gate to pass

Real fill quality, costs and risk limits in line before each step up.

TB3cFollows the paper data-collection phase.
Monitor and report, every day

Holdings reconciled with the broker, every order logged, the position book saved after each change and reloaded after any restart. Live results are tracked against the tested range; if they drift outside it, the strategy is cut back or stopped.

3 of 170+

variants have reached paper trading. That ratio is the process working: each rejection is a flaw found in simulation instead of in a client's account.

Strategies

One process across asset classes.

The same seven gates apply to a two-ETF allocation and to a multi-leg options book. Each line shows the furthest gate that work has reached, stated plainly.

Asset classWhat we buildStage reachedHistory used

Index and ETF options

Defined-risk income overlays added on top of an existing portfolio; spread ladders; directional and volatility structures.

Paper trading

TB3c on an Interactive Brokers paper account since 21 Oct 2026. The original three-rung TB3 also runs on a simulated paper account.

From 2012

Futures: equity, rates, currencies, commodities

Trend following across 30 markets, sized by volatility; trend as a hedge for the options overlay.

Backtested

Tested as a stand-alone sleeve and as a hedge for TB3c. Hedge code built and run in simulation.

From 2010

ETFs and multi-asset allocation

Rules-based rotation between stocks, bonds and short Treasuries, with staggered rebalancing.

Backtested

In-sample result.

From 2004

US equities

Low-volatility large-cap portfolio with a market trend filter that moves to short Treasuries in downtrends.

Backtested

In-sample result that held across every nearby setting tested.

From 1999

Fixed income and cash

Treasury futures inside the trend sleeve; T-bills and short Treasuries as collateral and as the defensive asset.

Inside tested strategies

Used as building blocks of the strategies above.

From 2004

Backtested means simulated on history that was already visible while the rules were chosen. Only TB3c trades at a broker, on an Interactive Brokers paper account; the original TB3 runs on the engine's own simulated paper account. What can trade live also depends on the broker and the client's account permissions.

Case study

TB3c: the process run end to end.

From a written brief to a broker account in 13 days, with every rule tested before it traded.

TB3c is an options income overlay. Three times each trading day it sells a small, defined-risk put spread on the S&P 500 ETF, 35 to 45 days from expiry, on top of a portfolio the investor already holds. The portfolio stays invested and the option income is added on top. It earns while the market holds up and loses when the market falls hard.

  1. Brief. Objective, limits, strikes and every exit rule written down.

  2. Oct 2026

    Backtest and robustness. 2012–2025 with commissions and quoted spreads; seven rule alternatives; SPX index options at $5M and $10M.

  3. First broker trade on an Interactive Brokers paper account, 13 days after the brief.

  4. 24–26 Oct 2026

    Three upgrades, each tested first: a conditional time exit, a quote check with a pause after stops, and a limit-order ladder with assignment handling.

Risk management

Risk is designed in at five levels.

Returns are an outcome; risk limits are a design input. Every strategy is specified from its worst case outward, and each control below is code, not discretion.

PositionEach trade

  • Defined-risk structures: the maximum loss is known before entry.
  • A hard stop on every position.
  • No entry when quotes are too wide to trade fairly.

In TB3cStop at 2× credit. No entry if the legs' combined bid/ask width exceeds 30% of the credit.

StrategyThe whole book

  • A risk budget cap on the sum of all open maximum losses.
  • Many small, staggered positions instead of a few large ones.
  • Automatic pauses after losses and in market stress.

In TB3c35% of the account across 21 slices; 3 entries a day; 5-day pause after a stop; no entry if VIX is above 32, up 30% in 3 sessions, or SPY fell 2.5% the day before.

PortfolioThe client's whole account

  • A drawdown budget agreed with the client before launch.
  • Overlays sized on the whole portfolio they sit on.
  • Hedging sleeves tested for how they combine, not alone.

In researchA futures trend sleeve tested in simulation as a hedge for TB3c; not deployed.

ExecutionOrders and the broker

  • Limit orders that step toward the market before any market order.
  • Unfilled orders cancelled so nothing fills untracked.
  • Assignment and early-exercise handling built in.

In TB3cCombo limits at mid, then 1/3 and 2/3 toward the natural price every 2 minutes, then market.

ModelThe research itself

  • Rules fixed before testing; every variant tried is counted.
  • Neighbouring settings, sub-periods and stressed costs.
  • Live results tracked against the tested range.

Program170+ variants logged with their verdicts, failures included.

Technology

One stack from research to the broker.

Data, engine, broker link and monitoring are shared by every strategy. A new idea is a new module on the same stack, not a new system. Numerical results come from the calculation engine; generated text only explains them.

01Market data

Prices, option chains and futures history from minute to daily; live feeds from the same source.

02Research

Strategies written as code, backtested and scored against pass/fail rules set in advance.

03LEAN engine

One codebase runs the backtest, the paper account and the live account.

04Broker

Interactive Brokers: orders, fills and holdings; paper account first, then real.

05Client account

Each strategy trades the account it is set up for, within the limits agreed with the client.

Every day

Holdings reconciled with the broker, every order logged, positions recovered after any restart, results reported against the backtest.

No translation step

The code that was tested is the code that trades, so the backtest, the paper account and the live account follow exactly the same rules.

Built to scale out

More strategies or more clients mean more servers and broker accounts on the same stack. Each live strategy runs on its own server with its own limits.

AI-assisted, human-approved

AI tools speed up research, testing and code review. Every change that reaches a broker account is approved and deployed by a person.

Governance

Every task has a purpose, an owner and a boundary.

A human mandate creates scoped tasks. Each task carries its own permissions and compute allowance. Exceptions return to a reviewer.

Design example
  1. MandateResearch lead approves: test a rule change on the agreed dataset.
  2. TaskData check · read-only data access · small compute allowance.
  3. TaskImplementation · code repository only · no market access.
  4. TaskExperiment report · reads results, cannot change them.
  5. ExceptionReport task asks to rerun with new parameters. Blocked and returned to the reviewer.

Hard boundaries

Research autonomy is not production authority.

No self-granted access

An agent cannot grant itself broker access or new permissions.

No limit changes

An agent cannot change risk limits.

No self-promotion

An agent cannot promote its own code into a live environment.

Isolated clients

Client strategies, data and results stay separate. Only contracted general methods are reused.

Exportable work

Your evidence history stays yours, and can be exported.

Evaluated changes

New models and tools are tested on a fixed task suite before adoption.

OASIS operating model

A research ecosystem that can adapt.

Five principles, each tied to a concrete operating choice and the evidence we collect for it.

PrincipleOperating choiceEvidence we collect
OpenInteroperable interfaces, documented dependencies, exportable artifacts.Successful export and reproduction outside the original workspace.
AutonomousPurpose, permissions, budget, memory scope and escalation set per agent.Policy decisions, denials, approvals and budget use.
SymbioticHumans own intent and judgment; agents perform bounded work.Named reviewers, exception handling, accountable acceptance.
IntelligentMarket expertise, statistics, deterministic computation and AI assistance combined.Baseline comparisons, test outcomes and reviewed learning.
SustainableEconomic viability, knowledge continuity, adaptability and resource efficiency.Cost per investigation, handover quality, recovery tests.

Detailed architecture, security documents and access boundaries are shared after an NDA.

Working with WaveOn

Your objective, run through the same gates.

A client's idea, from a simple ETF portfolio to a multi-leg options overlay, goes through the same seven gates before any money is committed.

What you receive

  1. A written strategy specification, every rule in plain language.
  2. A backtest report with costs, the worst periods and what was not tested.
  3. An interactive briefing to explore the results.
  4. A paper-trading period before any real money.
  5. Live execution within agreed limits, monitored daily, with regular reports.

What we need to start

  • The objective: income, growth, protection, or a mix.
  • Capital, time horizon and the largest drawdown you can accept.
  • Instruments allowed and not allowed.
  • Broker or custodian, and any tax, regulatory or jurisdiction constraints.
  • How often you want reports.

About

Who builds and runs WaveOn.

WaveOn is founder-led, with complementary investment and technical leadership.

Albert Calosso

Co-founder

Albert builds investment strategies the way engineers build bridges: specify the load, test to failure, and only then open to traffic.

At WaveOn, Albert designed the research-to-execution stack, in which one codebase runs the backtest, the paper account and the live account, and leads a research program that has put more than 170 strategy variants through pass/fail tests fixed in advance. TB3c, WaveOn's options income overlay, went from written brief to its first broker trade in 13 days.

Career details pending biography approval.

Massimo Mistretta

Co-founder

Product strategy, architecture, AI orchestration, cybersecurity and adaptive organization design. Author of the OASIS operating model. Leads the Exentriq relationship.

Owns technical scope, AI governance and technical diligence.

Biography pending approval.

Contact

Request the strategy briefing.

The briefing covers TB3c's rules, the full backtest with costs and worst periods, live paper results, and the risks, in one interactive document.

Email[contact email]
LinkedIn[profile link]
Based in[city, country]
  1. We replyTo arrange a discovery conversation.
  2. DiscoveryWe check the objective, the data and the fit.
  3. ScopeIf it fits: NDA, then a paid, bounded scope.

Keep the first message high level. Please do not include confidential strategies, account credentials or proprietary data. If there is a fit, we offer an NDA and a secure transfer process.