How to research a stock with ZadVest AI Analyst
Follow narrated AAPL backtests, including mixed AND/OR rules, then explore SEC filing checks and matched comparisons.
Check the original sources, compare like-for-like backtests, and keep the result and its limits together in your research note.
Follow the video: Run and interpret one AAPL backtest
Open the workspace from the ZadVest homepage. In AI Analyst, choose Open full analyst desk, start a New conversation, and select Deep report.
Try this prompt to reproduce the setup: “Propose one simulated daily AAPL backtest from January 2 through June 30, 2025. Enter when RSI(14) is below 40. Use a stop 2 ATR from entry and a profit target twice that distance. Assume a simulated account with 10,000 US dollars. Compare with buy and hold and explain the sample size limits.”
In the suggested backtest, select Review setup. Check the symbol, dates, entry rule, stop, target, and estimated credit cost. The recording shows a one-credit estimate. Choose Run backtest, then View run to follow its status and see the results.
The run recorded six simulated trades. Buy and hold returned −17.3% over the available January 2 to June 27 price window. Select Load details to see the chart and simulated account. Set Starting capital to 10,000 US dollars to match the prompt. The simulated account showed a −2.1% return. In the Trade ledger, review the individual trades and choose Export CSV to save them.
Reopen the original Analyst report and choose Ask about this run. Review and send the prepared follow-up. The Analyst explained that six trades are too few for a strong conclusion and that the −12.7% sum of individual trade returns uses a different calculation from buy and hold. The simulated account return above reflects how those trades were sized in the account.
The following sections show how to preserve mixed AND/OR rules, check Apple’s SEC filings and compare two versions of a strategy.
Keep AND/OR groups intact while editing
Use parentheses when one condition is mandatory and either of two alternatives may complete the entry. This example means: RSI must be below 40, and either the close must be above SMA 200 or MACD must cross above its signal. An OR at the root would describe a different rule.
The narrated production walkthrough follows one fresh request through the proposal, editable setup, review, submission and completed results. Clicks are highlighted; account details are blurred and processing waits are shortened. Follow the exact prompt and steps below to try it yourself.
Read the mixed-rule result with its limits
The November 9 production recording uses AAPL daily bars, a final RSI threshold of 40, a 2.5 × ATR stop, a 2R target, 5 bps exit-only slippage and no holdout. The nested grouping remains the same in the proposal, edited review and submitted run. No commissions are modeled.
That recorded run produced 10 trades: three wins and seven losses, a 30% win rate and an average of −0.13R per trade. Its trade-return sum was −1.02%; the 18.91% maximum drawdown uses a different, compounded one-position basis. Neither figure is a sized account return. Buy and hold returned 48.50% on adjusted closing prices before trading costs. The actual price window ended December 30, 2025, although the request specified December 31.
The report settled at five AI credits, and the submitted simulation cost two backtest credits; setup review and editing did not require another AI request. Costs and processing times vary. Ten trades, no holdout and same-close fills limit interpretation: this example demonstrates a complete research workflow, not a validated trading edge.
Research Apple’s business risks using SEC filings
Open AI Analyst, turn on Deep report, and try this prompt:
“Using AAPL’s latest SEC 10-K and prior 10-K, identify two material business risks. Separate disclosed facts from interpretation and suggest a checkable follow-up question. Highlight a year-over-year change only if both filings support it.”
Open Full report and expand Claims and filing sources. The Filing fact entries describe what Apple disclosed. The Analyst interpretation entries explain what the Analyst inferred. Follow the SEC links to read each passage in context.
In the saved example, both filings discussed supplier safety risks. The wording changed, but the passages did not establish a change in safety practices. Under Questions to investigate, the Analyst suggested checking Apple’s supplier responsibility disclosures. Draft follow-up opens that question in the composer for you to review before sending.
Check for new filings before refreshing
Open the report from Saved Reports. Under Resume research, find SEC filing updates and select Check for new filings. ZadVest checks SEC EDGAR for filings within the report’s scope using zero AI credits.
If ZadVest finds no newer filing in scope, the report shows when it checked.
When it finds a newer filing, the report shows the filing type, date, and an SEC link. Read the filing first if you want to decide whether a fresh analysis is useful.
Refresh report becomes available when the check finds new evidence. The screen shows the credit reserve before you submit. Unused credits are returned. When the refresh completes, choose Open updated report. You can still revisit the original.
If EDGAR is unavailable, try again later. If ZadVest cannot compare this report’s filing history, check the issuer’s filings directly.
Set up two backtests that differ by one rule
Start a separate Analyst report and try this prompt:
“Suggest a matched daily backtest for AAPL from January 3, 2023 through December 31, 2025. Compare RSI crossing above 30 with RSI crossing above 35. Keep the other rules, costs, and holdout split the same. Judge the result using held-out trade returns and report the sample size.”
Choose Review both setups. Check the dates, exit slippage, stops, target, and holdout split. In this example both runs used 5 basis points of exit slippage and a 30% holdout. The review shows the estimated backtest credits. If your plan runs one backtest at a time, run the baseline first and submit the variant when it finishes.
Compare the runs and save your conclusion
When both runs are ready, choose Compare baseline and variant. Confirm the change under Run settings. Read the holdout return alongside the number of trades and drawdown.
The baseline recorded 7 trades and the variant recorded 9. The sum of holdout trade returns was 13.15% for the baseline and 16.01% for the variant. Full-run max drawdown was 9.23% and 22.85%, respectively. With only two holdout trades per run, ZadVest flagged Insufficient evidence. The buy-and-hold benchmark tracks the observed price window and uses a different return basis from the sum of individual trade returns.
Choose Back to report, open My research note, and record the result and its limits. Select Save note. When you return later, Check linked runs shows the status of the saved backtests without using AI credits.
These examples illustrate a research process. Historical simulations do not establish future performance.
Optional paper setup and agreement help
General stock research and backtesting do not need a connected brokerage account; your ZadVest plan determines paid tool access. When you want simulated execution, open Settings, choose Brokerage, and set up an Alpaca paper account. The API agreement is separate from signup terms and provider authorization. Accepting an agreement does not place an order or enable automation.
A complete existing signature keeps its original acceptance date. Checking its receipt uses no AI credits and does not sign again. If the check is temporarily unavailable, retry it; disconnecting and protective account controls remain available.
Older installed apps bundle their interface and may not support the new first-time acceptance endpoint. Use the web agreement settings link in the links below to open ZadVest in a browser, sign in to the same account, review the unchanged API agreement and sign only if it is missing. Then reopen the installed app: its existing account record receives the accepted status. Reloading an older installed app alone does not update its interface. This legal setup does not require buying a new plan.