Comparisons

OpenCandle is best compared with a general chatbot. The difference is that OpenCandle gathers provider-backed evidence through explicit finance tools, discloses when a source was missing, stale, or degraded, and synthesizes from that evidence trail.

OpenCandle is read-only research software. It does not place trades, route orders, or provide financial advice.

OpenCandle vs General Chatbots

OpenCandle runs blind head-to-head comparisons against generic frontier-model agents answering the same prompts without finance tools. A judge scores the answers without knowing which agent wrote them. Testing and Evals describes the method. The patterns below come from those runs, including the ones OpenCandle lost.

Where no-tool chatbots fail

They fabricate current data. Asked what changed in a company's most recent 10-Q, a no-tool agent produced specific, plausible, invented filing changes. Asked for the current social mood around a volatile stock, another invented current posts and sentiment. The fabricated answers read better than honest ones, and judges rewarded that specificity until the scoring penalized unverifiable claims explicitly. OpenCandle fetches the actual filings and the actual posts, and says when a source is unavailable.

They cannot see today's market. "Why did this stock move today" needs a current quote. In benchmark runs, no-tool agents either declined or guessed; with a live quote, the answer can state the move, the session, and how fresh the number is.

They lose track of position math. Options prompts exposed recurring slips: quoting per-contract premiums as per-share, and dropping a stated share count or cost basis when sizing a hedge or covered call. OpenCandle reads the option chain and carries the stated position through the answer.

They cannot review your portfolio. Whether your holdings are overexposed to interest rates depends on your actual lots and watchlist. OpenCandle stores that state on your device and reuses it; a chatbot needs it re-pasted every conversation.

Where chatbots win

OpenCandle's benchmark losses cluster in two places:

• Conceptual education. Questions like how to use P/E ratios without over-relying on them need no live data. A well-structured chatbot explanation can beat a tool-driven answer.
• Judgment-heavy personal finance. Brokerage selection, mortgage-versus-investing tradeoffs, and tax-loss-harvesting mechanics turn on stable facts and judgment rather than live market data.

Both prompt classes stay in the benchmark set so the comparison stays honest.

Capability	OpenCandle	General chatbot
Finance tools	Built in: market, macro, options, filings, sentiment, fundamentals, crypto, portfolio	Depends on plan, connectors, or custom setup
Missing or stale data	Disclosed in the answer	Usually silent, sometimes covered by fabrication
Local portfolio state	Stored on your device, reused across sessions	Re-pasted every conversation
Web app (no install), local GUI, and terminal	Yes	No

When OpenCandle Is Not The Right Tool

Use something else when you need order routing, regulated financial advice, high-frequency trading infrastructure, proprietary terminal data, or fully automated investment decisions. OpenCandle is built for research and evidence inspection, not execution.