Risk Disclosure
Read this before you trade with anything you learn or build here.
- Trading can lose you money quickly, including more than you planned. Most people who trade actively lose money. Only trade with money you can afford to lose.
- Trader Machine does not make you profitable. It is research software and education. It does not tell you what to trade, it does not know your finances, and it never places an order.
- Backtests and AI output describe the past, not the future. Every result in the Service is hypothetical (§3).
- Automation acts on its own. An exit alert closes your whole position in that symbol (§6).
1. Trading risk
- Stock prices can fall sharply and without warning. You can lose some or all of the money you trade with.
- Prices can gap — open far from the previous close — after earnings, news or overnight events. A stop order does not guarantee the price you set; in a gap it may fill much worse.
- Short-term and frequent trading magnifies costs and mistakes. Most people who trade actively lose money over time.
- Margin and short selling can lose more than you deposited. Your broker can close positions without asking you.
- Thinly traded stocks can be hard to sell at a fair price, and extended-hours and overnight trading is thinner still. Halts can stop you from exiting at all.
- U.S. brokers apply pattern-day-trader rules to margin accounts that day-trade frequently with less than the required equity. Breaking them can restrict your account.
- Trading results have tax consequences. We do not give tax advice.
2. Not investment advice
Not investment advice. We publish general research software and education. We are not a broker-dealer, investment adviser, commodity trading advisor or financial planner, and we are not registered as any of these anywhere. Nothing in the Service is a recommendation to buy, sell or hold any security for you.
Our software, data, backtests, AI output, alerts, live sessions, lessons, Direct Access calls and support replies are impersonal (the same for every subscriber), general (not about your account or your finances) and regularly published (on a schedule, not timed to your trades or to specific market moves). None of it is a view that a strategy or a security suits you. Whether to trade, and how, is your decision alone. Consider speaking with a licensed professional who knows your circumstances.
3. Hypothetical performance
Every backtest, equity curve, return figure, win rate, drawdown and other statistic in the Service is hypothetical unless it is clearly labelled as a real account statement — and we do not publish any.
Hypothetical or simulated performance results have inherent limitations. Unlike an actual performance record, simulated results do not represent actual trading. Because the trades were never executed, the results may have under- or over-compensated for the impact of market factors such as lack of liquidity. Simulated strategies are designed with the benefit of hindsight. No representation is being made that any account will or is likely to achieve profits or losses similar to those shown.
Real trading also involves things a simulation cannot reproduce — above all, whether you can keep following a set of rules through a run of real losses, and whether you have the money to do so. These factors can decide real results, and they are absent from every backtest.
There are often sharp differences between hypothetical results and the results later achieved in real trading. Past performance, actual or hypothetical, does not indicate future results.
4. Why backtests look better than reality
These are the specific ways a backtest in our software — or any software — can flatter a strategy:
- Fills. The backtester assumes orders fill at modelled prices taken from the bars, as if the market were always liquid enough. Real orders face slippage, partial fills, rejections and queue position. Inside a single bar, the order in which the high and the low happened is unknown.
- Costs. Commissions, fees, borrow costs, spreads and taxes may be modelled simply or not at all.
- Data. Historical data can contain errors, missing bars and adjustments for splits and dividends that differ by provider. Lists of stocks may leave out companies that were later delisted, which makes past results look better than they were.
- Overfitting. Rules tuned to fit past data closely tend to describe noise, and noise does not repeat. The more parameters a strategy has, the more likely its backtest is to fail going forward.
- Testing many variants. When many versions of a strategy are tested — by you, or automatically by AI QUANT — the best-looking one is partly the luckiest one. Its result is biased upward, and the bias grows with the number of variants tried. Treat a winner from a large search with more suspicion, not less.
- Small samples. A handful of trades, or a short period, can produce impressive statistics by chance.
- Changing markets. Conditions that made a rule work — volatility, trends, interest rates, market structure — change.
- Live is not the backtest. Live alerts are evaluated when our system checks your conditions, and real orders fill at the market after an alert arrives, so live results drift from the backtest even when nothing goes wrong.
5. AI QUANT and other AI output
AI QUANT is an AI — a large language model provided by Anthropic, not a person and not an adviser. It proposes strategy rules from your request; the backtest shown beside each proposal is hypothetical (§3–4) and subject to the multiple-testing bias above. Our AI support agent uses the same kind of model.
- Models can misread a request, state falsehoods with confidence, suggest rules that do not suit you, and answer the same question differently twice.
- AI QUANT does not search the internet and does not know your account, your finances, the news or company filings.
- AI output is not advice and not a forecast. Never make a trading decision on an AI result alone.
6. Automation risk
If you route your strategy's alerts through your own SignalStack account to your own broker:
- AN EXIT ALERT CLOSES YOUR WHOLE POSITION IN THAT SYMBOL, including shares you bought yourself outside the strategy.
- The automation does not see your account — not your cash, positions, open orders or fills. It cannot know whether an earlier order failed, and it will not stop because of a loss.
- Each alert is sent once and is never re-sent automatically. A failed delivery is not repaired by a second attempt; check your account.
- Pausing or deleting automation, or the end of your subscription, does not close an open position, and no further exits are sent for it.
- The quantity in an alert is the one you entered. The automation does not size positions to your account.
- Alerts can arrive late or not at all because of data delays, outages (ours, SignalStack's or your broker's), internet problems, halts or rejections. A missed exit can leave a position open while the market moves against you.
- Market orders can fill far from the last price you saw, especially at the open, around news and in thinly traded stocks.
- Automation acts without pausing to reconsider, so it can lose money faster than manual trading.
- Your SignalStack webhook address works like a key to your account. Anyone who obtains it can send orders.
- SignalStack is a separate company that bills you directly: a small free monthly allowance, then paid monthly plans. It counts a signal when an order reaches your broker, even if the order does not fill. Your broker may charge for orders too.
- Watch your account, set risk limits at your broker, and begin with small size.
7. Education, Direct Access and conflicts
- Live sessions, recordings and lessons reflect the presenter's opinions and teaching examples. A trade discussed or shown is an illustration, not a recommendation, and may be shown with hindsight.
- Direct Access calls review your research method, not your account. They never cover which securities to buy or sell, your positions or your position size, and nothing said on them is a recommendation (Terms §15).
- Our founder and staff may hold or trade securities discussed, including in the opposite direction. When our founder discusses by name a security he holds, he says so at the time. We do not announce our trades.
- How we earn: from subscriptions, and from referral commissions SignalStack may pay us for accounts opened through our link — not from your trading results.
- We do not publish customer testimonials about profits, and no one — including our affiliates — is authorised to promise you profits on our behalf. Affiliates are paid a commission when someone they refer subscribes.
8. Market data
Prices come mainly from Alpaca, and other data from other providers and public sources. Data may be delayed, incomplete or wrong. Different parts of the Service can use different sources and update at different times, so a scanner, chart, backtest and alert can disagree, and a fallback feed can show only one exchange's trades. Earnings dates and estimates change often. Confirm on the chart, in the company's own announcement and with your broker before acting (Terms §10).
9. Your responsibility
You confirm that you understand these risks, make your own decisions, and do your own research before every trade (Terms §5). If you do not understand a risk described here, do not trade until you do.
10. The notices shown in the application
This notice applies to every backtest, equity curve and AI QUANT result in the application:
Hypothetical results. These figures show how these rules would have behaved on past data. No trades were placed and no money was at risk. They were computed with hindsight, assume modelled fills and costs, and cannot reflect liquidity, slippage or decisions made under real losses. No representation is being made that any account will or is likely to achieve similar results. Past performance does not indicate future results. Not investment advice. Details
Where space is tight, this one-line version is used, with the link. It is the line the application shows today between the figures and the equity curve of every Strategy Builder result, and above the AI QUANT result cards:
Hypothetical results based on historical data; they have inherent limitations and do not represent actual trading. See risk disclosure.
Under the AI QUANT input box and on every AI QUANT result card:
AI QUANT is an AI tool and can be wrong. It proposes rules; the results beside them are hypothetical backtests on past data, and the best of many tries is partly luck. Never trade on an AI result alone. Not investment advice. Details