Why Did the Bot Wait? A Decision Record Standard for AI Trading Bots
Aug 18, 2026 · 8 min read
A liquidity sweep is a fast move through an obvious price area, not proof that someone manipulated the market or that price must reverse. Use a repeatable review before taking any action.
June 19, 2026 · 9 min read · TradingWizard AI
A liquidity sweep is a fast move through a price area where traders expect orders to be concentrated, such as a recent high or low. It can be useful context, but it is not proof that a large participant deliberately targeted stops, and it is not a buy or sell signal. The safer response is to check the market, timeframe, source timestamp, follow-through, invalidation, and risk before deciding whether the honest state is a defined setup or simply WAIT.
TradingWizard can help organize that review around a supported asset. Its bot scans can show a current state, plain-language reason, entry, stop, target, invalidation, confidence, and risk context when those fields are available. Every TradingWizard bot runs in paper mode with fake money against real market data. TradingWizard cannot place, modify, or close real orders.
Traders often use “liquidity sweep,” “stop run,” and “stop hunt” to describe a short-lived move through a visible high or low. That label is a market-structure observation, not an explanation of who caused the move or what happens next. A price can extend, reverse, or become too unclear to review.
The important mechanics are more modest. The CFTC’s study of stop orders in selected futures markets found that stop orders are more often executed on volatile days and can provide liquidity in the direction of a price move. The orders themselves are generally not visible in the order book. That does not prove coordinated manipulation in a particular chart or asset.
| Observation | What it can support | What it cannot prove | Safer next check |
|---|---|---|---|
| Price moves through a recent high or low | A commonly watched level has been crossed | That a “smart money” actor forced the move | Check the market, timeframe, volume context, and whether price holds beyond the level |
| Price quickly returns inside the prior range | A reversal may be worth monitoring | That the reversal will continue | Define invalidation and wait for a current state rather than treating one candle as a conclusion |
| Price keeps moving beyond the level | The move has follow-through | That the next pullback is safe to trade | Review trend, volatility, distance from the level, and the next current scan |
| The source is delayed or unclear | The observation may be old | That the current market is at the same condition | Treat the state as DATA STALE or wait for a fresh review |
| The level is obvious on several timeframes | More traders may be watching it | Where every order sits or who will act next | Use it as one input, not a complete setup |
Want to review a possible sweep without risking money in the workflow? Choose a supported asset and deploy a paper-mode bot in Wiz. Start with its first current state and timeline rather than a single candle.
A stop order and a stop-limit order are not interchangeable, and their behavior depends on the market and brokerage rules. The SEC’s Investor Bulletin on stop, stop-limit, and trailing stop orders explains that a stop order becomes a market order once triggered. In a fast market, the execution price can differ materially from the trigger price. A stop-limit order sets a price boundary, but it may not execute at all.
That is why a fast move through a level should not become a story about a certain reversal, a certain fill, or a hidden actor. The trader needs to distinguish the observed move from the decision that follows.
| Review item | Question to ask | A useful answer | Warning sign |
|---|---|---|---|
| Market and venue | Which supported asset and market are being reviewed? | The asset and context are explicit | A chart label or quote source is ambiguous |
| Timeframe | Is this a brief move or a broader structural break? | The timeframe is named and fits the observation | A one-minute move is used to make a long-horizon claim |
| Recency | When did the data and scan update? | Timestamp and data condition are visible | A delayed screenshot is treated as current |
| Follow-through | Did price hold beyond the level or return? | The observation is described without predicting the next move | One wick is presented as a complete trade plan |
| Invalidation | What would make the interpretation wrong? | A boundary exists before any paper test | The only instruction is to “buy the reversal” or “sell the break” |
| Risk | Are stop, target, and risk notes available when a setup exists? | The fields are shown together or the bot stays out | A confidence label substitutes for a boundary |
Use the same sequence every time. It reduces the temptation to invent a precise narrative after price has moved.
| Checkpoint | Pass condition | Stop and investigate when |
|---|---|---|
| Identity | Supported asset and timeframe are clear | Symbol, venue, or timeframe is unclear |
| Current data | Timestamp and data condition are visible | Older data appears current |
| State | A concrete setup state or no-trade state is visible | A vague bullish/bearish label replaces a decision |
| Boundaries | Entry, invalidation, stop, and target are shown when available | A trade idea has no condition that would make it wrong |
| Timeline | Later scans can be compared with the original observation | The latest view hides prior waits or changes |
| Paper label | Simulated activity is clearly marked | Paper activity is framed as real execution |
A valid outcome can be WAIT, AVOID, EXPIRED, DATA STALE, or RISK BLOCKED. A bot does not need to manufacture an entry just because price briefly crossed an obvious level.
TradingWizard is not an order-book feed, a liquidation oracle, or a live broker. It is a bot-first AI trading product for active traders. The useful path is to choose a supported asset, deploy a paper-mode bot, inspect its current state and reason, then review its timeline as conditions change.
When a setup exists, bot scans can provide relevant fields such as entry, stop, target, invalidation, confidence, and risk context. When the evidence is late, incomplete, or blocked, the useful answer may be a visible no-trade state. For the field-by-field standard, see Why Did the Bot Wait? A Decision Record Standard for AI Trading Bots. For a broader risk checklist, see How to Evaluate an AI Trading Bot. For a related crypto-context workflow, see Crypto Liquidation Heatmaps: From Crowded Risk Zones to an AI Setup Review.
Every TradingWizard bot uses fake money against real market data. Paper activity can help test whether the workflow shows current state, boundaries, and no-trade behavior. It cannot prove future performance or reproduce every live-market condition. Alpaca’s paper-trading documentation lists limitations such as market impact, information leakage, latency-driven slippage, order-queue position, price improvement, fees, and dividends.
A move through a prior high or low is observable. The claim that an institution or algorithm deliberately “hunted” a particular group of stops usually is not. Keep the article, chart review, and trading record anchored to what is actually available.
A short-lived move may reverse, extend, or be followed by more volatility. A wick does not supply an entry, an invalidation, a stop, a target, or a realistic fill assumption by itself.
Stop orders can execute at a price different from their trigger in fast markets. Broker and venue rules also vary. Review the order type with the relevant broker rather than assuming a chart level guarantees a fill.
A review that records only reversals hides the cases where price continued, data was stale, or no setup was justified. Keep waits, failed ideas, revisions, and simulated losses in the same timeline.
Paper mode is a test environment. TradingWizard cannot place, modify, or close real orders. Any real-money decision and order happens outside the product.
A liquidity sweep is context, not a command. Treat a move through a visible level as a reason to check recency, follow-through, invalidation, and risk—not as proof of manipulation or a certain reversal. Keep the workflow in paper mode, record the no-trade outcomes, and keep every real-money decision outside TradingWizard.
FAQ
Aug 18, 2026 · 8 min read
Jul 10, 2026 · 7 min read
Jul 7, 2026 · 9 min read
Jul 6, 2026 · 8 min read
Free. No card.
Trading involves risk. Every bot starts in paper mode: no real money.