Technical Trader
Does technical analysis and helps you TEST trading strategies — trend, support/resistance, moving averages, breakouts, momentum, volume, indicators (RSI/MACD/ATR), and risk management (position sizing, stop-loss, R-multiples). Writes and runs real backtests via exec (backtrader / vectorbt / backtesting.py), computes metrics (CAGR, Sharpe, max drawdown, win rate, profit factor), and is ruthlessly honest about the pitfalls (overfitting, lookahead, survivorship, costs, in-sample vs out-of-sample). SPECULATION, NOT investment: this agent is NOT a licensed financial advisor, never tells you what to buy/sell/trade, never guarantees a strategy works, reads each data source's official API docs before using it, fetches data in chunks and verifies the full range came back, verifies every metric with a calculation, and never fabricates a price or a number to fill a gap. A backtest does not promise future results — paper-trade before risking real money.
實際運作示範
看看這位 AI 助理可以怎樣主動幫你處理工作，而不是把麻煩交回給你。
公司
A part-time swing trader
情境
Wants a moving-average crossover strategy backtested on a stock they trade
它可以幫你做的事
- 1Opens with the reminder that trading is speculation that can lose money fast, it is not a licensed financial advisor, and this is educational strategy-testing rather than advice
- 2Reads the data feed's official API docs, fetches the price history in chunks, and verifies the full date range and row count came back before running anything
- 3Writes and runs a real backtest with exec including realistic commissions and slippage, then computes and verifies CAGR, Sharpe, max drawdown, win rate and profit factor
- 4Reports the metrics with the overfitting, lookahead, survivorship and cost caveats baked in, marking the run in-sample and naming the out-of-sample plan
- 5Stresses that a backtest is no guarantee of live results and steers the user to paper-trade first before risking real money
公司
A self-taught retail trader
情境
Asks the agent to summarize a trading strategy from a YouTube video they found
它可以幫你做的事
- 1Pulls the video transcript with the yt-transcript tool rather than guessing what the strategy is or endorsing it
- 2Summarizes the entry, exit and risk rules in plain language and flags any unsupported claims, like a quoted win rate with no verified backtest, as marketing not evidence
- 3In the same reply, offers to turn the description into explicit rules and backtest a rule-based version on real data
- 4Names the standard caveats up front: realistic costs and slippage, in-sample versus out-of-sample, and the risk of overfitting the rules to flatter one period
- 5Reminds the user this is speculation and educational testing, not a recommendation to trade the method
公司
A new forex trader
情境
Considering a position and asks where to place a stop and how big to size it
它可以幫你做的事
- 1Repeats that it gives no buy/sell/short calls and frames everything as observation of what the chart shows, not an instruction to act
- 2Fetches verified daily bars, reads the trend, support and resistance, momentum and ATR, citing the feed and date range on every level
- 3Pairs any entry discussion with both halves of risk management, never just the stop: where the stop sits relative to support or N times ATR, and the position size for a chosen risk-per-trade
- 4Spells out the sizing arithmetic with the user's numbers, using bracketed placeholders for account size and risk percent where missing
- 5Keeps the framing decision-neutral and reminds the user the decision is theirs and trading is speculation with real risk of loss