Disclosures and the desk rulebook

What this is not

Quantum Advisors is an educational and analytical platform. It is not a registered investment adviser, a broker-dealer, or a financial planner, and it does not provide personalised investment advice. Every number this platform produces is a model output computed from data you supplied or connected: it describes the past behaviour of a series, not the future behaviour of a market. Models are wrong in ways that are hardest to see exactly when they matter most. Nothing here is a substitute for a licensed professional who knows your full financial position, your tax situation, and your obligations.

If money has stopped being just money

Trading losses are not only financial events. If any of this is familiar — chasing losses, trading with money you cannot lose, hiding activity from people close to you, or feeling that the account outcome and your worth as a person have become the same thing — the people below help with exactly that, free and confidentially.

Immediate danger Dial 911.
988 Suicide & Crisis Lifeline Call or text 988 (24/7, free, confidential).
National Problem Gambling Helpline Call or text 1-800-522-4700 (24/7, free, confidential).
National Foundation for Credit Counseling Call 1-800-388-2227 for free or low-cost non-profit debt counselling.
Crisis Text Line Text HOME to 741741.

This software is not a substitute for any of them, and it will say so rather than try to handle a crisis itself.

What this desk discourages

  • Taking a position with no pre-defined risk budget On a desk, size is derived from a risk number, not the other way round. If you cannot say in advance what you lose when you are wrong, you do not have a position, you have an exposure.
  • Increasing size after a loss to make it back Martingale sizing converts a survivable drawdown into a terminal one. Risk-of-ruin is not linear in position size, and doubling down inverts the one relationship that keeps you in the game.
  • Averaging into a losing position with no written thesis Adding to a loser is a legitimate strategy only when the thesis said in advance where you would add and why. Added after the fact, it is a way of refusing to mark your own mistake.
  • Trading with money you need for living expenses Capital that has to come back on a schedule cannot be risk capital. Forced liquidation at the wrong time is not a market outcome, it is a funding failure - the same thing that kills leveraged funds.
  • Borrowing to trade Credit cards, personal loans, home equity, or money from family. Leverage on top of leverage removes your ability to survive a normal drawdown, and normal drawdowns are certain.
  • Trusting a backtest that was never held out of sample A strategy tuned on the same data used to evaluate it is a description of that data, not a model of the market. Every desk separates the sample used to fit from the sample used to judge, and treats the second as spendable once.
  • Ignoring transaction costs, slippage, and financing Most retail 'edges' are real gross and negative net. Costs are the first thing an institutional backtest models, not the last.
  • Treating correlated positions as diversification Ten technology names is one position with a decimal point in it. Correlations rise towards one in exactly the drawdown you were diversifying against.
  • Acting on a single unverified data source A desk reconciles at least two independent sources before a number is allowed to drive size. Bad ticks, split adjustments, and survivorship bias are ordinary, not exotic.
  • Revenge trading, or trading to relieve a feeling If the reason for the trade is the way the last one felt, the trade is about you rather than about the market. Desks impose cool-down periods for this reason and so should you.
  • Hiding trading activity or losses from the people it affects Concealment is the reliable early indicator that risk has outgrown the capital behind it. It is also the point at which the problem stops being financial.

What this desk encourages

  • Write the thesis and its falsifier before you size the trade One sentence for what you believe, one for the observation that would prove you wrong, one for the exit. If the falsifier cannot be written, the thesis is not a thesis.
  • Derive size from volatility, not from conviction Target a risk contribution, measure realised volatility, and let the position size fall out of the arithmetic. Conviction belongs in whether you take the trade, not in how big it is.
  • Run the pre-mortem Assume it is six months later and the position lost badly. Write down why. The reasons you can list now are the risks you can actually manage.
  • Separate the sample you fit on from the sample you judge on In-sample, out-of-sample, and then a live paper period. Each one is spent the moment you look at it, so look at them in that order and once.
  • Model costs before you model returns Spread, commission, market impact, borrow, financing, taxes. An edge that survives a pessimistic cost assumption is worth researching further.
  • Judge the process, not the outcome A good decision can lose and a bad decision can win. Grade the decision against what you knew when you made it, which is the only thing you controlled.
  • Keep a decision journal you actually write in What you expected, why, how confident, and what you would need to see to change your mind. Reviewed monthly, this is the highest-return activity available to an individual trader.
  • Know your risk of ruin before you know your target return Survival is a precondition, not a goal. Every institutional risk framework starts from the loss it must not exceed and works backwards to the position.
  • Reconcile your data, and store it point-in-time Keep the vintage of every number as it was known on the date it was known. Restated fundamentals and back-adjusted prices are the most common source of an edge that does not exist.
  • Assume the other side of your trade is better informed Ask what someone would have to believe to take the opposite position. If the only answer is that they are stupid, you have not found the answer.

How the AI here is constrained

Every generated surface — the mentor, the risk note, the audit diagnostic, the journal review — is wrapped in the same written framework before it sees a single word of your input. That framework forbids it from naming an instrument to buy or sell, predicting a price, telling you a position size as an instruction, evaluating a specific investment, or presenting model output as certainty. It also requires it to disagree with you when it has reason to, rather than agreeing to be pleasant.

Before any of that runs, a deterministic screen checks your text for signs of financial crisis or personal distress. That screen is ordinary code, not a model judgement, so it cannot be talked out of firing.

About the numbers

  • Every analytic is computed on your machine from price data. None of it is a forecast, and none of it is a recommendation.
  • Where price data is simulated rather than real, it is labelled as synthetic everywhere it appears. Simulated data produces correct arithmetic on fabricated inputs — useful for learning mechanics, useless for a decision.
  • Volatility, correlation, beta and VaR are sample statistics. They describe the window measured and are not promises about the next one. Correlations in particular rise sharply in a drawdown, which is when you were relying on them.
  • The optimiser is classical simulated annealing over a QUBO encoding. It is quantum-inspired. No quantum hardware is involved and none is claimed.
  • Past performance does not indicate future results. Losses can exceed deposits in leveraged and short positions.

Your data

Your positions, journal, audits and transcripts live in a database on the server running this application. Text you send to the mentor is transmitted to Anthropic's API to generate a reply; nothing else is. You can download everything held about you at any time from your data export, and deleting your account removes it.