Help & Documentation

Getting Started

Welcome to Nostradamus. This guide takes you from sign-up to your first AI-generated stock analysis in under 5 minutes — and then deeper into what makes the platform tick.

What Nostradamus actually does

Nostradamus is an AI-powered stock analysis platform that combines three different analytical lenses into one personalised recommendation:

• Technical analysis — indicators, chart patterns, support/resistance, multi-timeframe trends
• Sentiment analysis — financial news and social-media sentiment, weighted by recency and scored by an LLM
• Fundamental analysis — SEC-filed financials, valuation ratios, profitability, growth and health metrics

The platform then synthesises all three into a Complete Analysis that gives you a clear verdict (BUY / HOLD / SELL), a score, a confidence level, a bull case, a bear case, and a concrete trade setup with entry, stop-loss, and take-profit levels — all personalised to your Trader Bio.

1. Sign up — free, no credit card

Create an account with your email and a password (8+ characters). You'll get 300 tokens for free right after registration — that's enough for 10 Complete Analyses, no payment required. The welcome bonus is one-time, but the tokens themselves never expire.

2. Pick your first stock (onboarding)

Right after sign-up, you'll see a stock picker with six well-known tickers (AAPL, NVDA, TSLA, MSFT, GOOGL, AMZN). Click any of them and Nostradamus runs a full Complete Analysis straight away. This first analysis is paid from your welcome bonus, so it feels free.

3. Set up your Trader Bio (highly recommended)

The Trader Bio is what makes Nostradamus different from a generic stock-screener. It tells the AI:

• How long you hold positions (Swing, Position, Long-term)
• How much risk you accept
• What indicators you trust
• How aggressive your stop-loss should be
• How many take-profit targets you want

The same stock can have a "BUY" verdict for an aggressive swing trader and a "HOLD" verdict for a conservative long-term investor — that's the personalisation at work.

4. Run analyses or backtests

1. Go to Stocks → Complete Analysis
2. Type a symbol (e.g. AAPL) and hit Enter — or browse your Watchlist
3. Confirm the token spend (you can tick "Don't ask again" to skip future prompts)
4. Wait ~30 seconds — Technical, Sentiment, and Fundamental analyses run in parallel
5. Read the unified recommendation including score, confidence, bull/bear cases, and a personalised verdict

5. Build a Watchlist

Add stocks you're tracking to your Watchlist. From any analysis dashboard you can either click a single Play button on a row, tick checkboxes to analyse a subset ("Analyze Selected"), or run all of them at once with "Analyze All".

6. Manage tokens

Your token balance is always visible in the top-right corner. When you run low, the Tokens page lets you see your full history, top up your balance, and review the price list of every paid action.

What it's not

Trader Bio

Your Trader Bio is a structured profile that personalises every analysis. The same stock can show "BUY" for one bio and "HOLD" for another — the difference is in the lens, not the data.

Why it exists

A swing trader looking at AAPL wants different information than a buy-and-hold retirement investor. The swing trader needs near-term momentum, clean entry levels, and a tight stop-loss. The retirement investor needs long-term fundamentals, valuation context, and dividend safety. Generic analysis can't serve both — your bio tells the AI which one you are.

Profiles & multi-profile management

You can have up to 10 bios per account. Common patterns:

• One bio for your short-term swing trading account
• Another for your long-term retirement portfolio
• A third for paper-trading aggressive setups

Switch the active bio from the profile dropdown in the header. Each bio has its own analysis history and its own backtest results — they're isolated boxes, so switching doesn't pollute data. Your Watchlist is shared across all bios (so you don't have to rebuild a stock list for every profile).

Trading style — the most important field

Choose how long you typically hold positions. This setting drives everything else: which candle timeframe is used, the default weights for Technical / Sentiment / Fundamental analyses, the default stop-loss aggressiveness, and the backtest evaluation window.

• Swing Trading — Holding period: days to a few weeks. Uses daily candles. Technical analysis weighs heaviest (price action matters most short-term). Backtest window: 14 trading days.
• Position Trading — Holding period: weeks to a few months. Uses daily candles with a longer lookback. Sentiment matters more here (theme persistence). Backtest window: 30 trading days.
• Long-Term — Holding period: months to years. Uses weekly candles. Fundamentals dominate (valuation, growth, balance sheet). Backtest window: 26 weeks (~6 months).

Experience level

Tells the AI how much hand-holding to provide in the narratives. Four levels:

• Beginner (< 1 year) — Explanations stay plain-language, jargon is unpacked, and warnings are louder.
• Intermediate (1–3 years) — Standard technical/financial vocabulary, less hand-holding.
• Expert (7+ years) — Concise institutional-style commentary, assumes you know what RSI divergence or DCF means.

Risk tolerance

How much downside you're willing to accept. Affects stop-loss tightness, take-profit aggressiveness, and which recommendations show as "actionable" vs "wait for better entry."

• Conservative — Tight stops, conservative targets, only high-confidence trades flagged actionable.
• Moderate — Balanced. The default if you're unsure.
• Aggressive — Wider stops, more ambitious targets, lower confidence threshold for "actionable" trades.
• Very Aggressive — Maximum risk tolerance, momentum-chasing style.

Portfolio size

Used for position sizing recommendations — how many shares you should buy given the suggested stop-loss and your max-position-size percentage.

• Small — < $10k
• Medium — $10k – $100k
• Large — $100k – $1M
• Institutional — > $1M

Optionally also set Max position size % — the cap on how much of your portfolio a single trade can use. E.g. 5.0 = no single trade larger than 5% of the portfolio. Aggressive traders sometimes go to 10%; conservative investors stay under 2%.

Custom analysis weights

By default, the Complete Analysis blends Technical / Sentiment / Fundamental scores with weights that depend on your trading style:

• Swing default: 55% Technical, 25% Sentiment, 20% Fundamental
• Position default: 40% Technical, 30% Sentiment, 30% Fundamental
• Long-Term default: 20% Technical, 20% Sentiment, 60% Fundamental

If you trust news sentiment more than charts (or vice versa), enable "Use custom weights" in the bio editor and set your own three numbers (they must sum to 100%). The AI then applies your weights every time.

Favorite indicators

Pick the technical indicators you actually use — for example RSI, MACD, Bollinger Bands, EMA crossovers, ATR. The technical analysis highlights signals from your favorites in the narrative ("RSI is showing bullish divergence — one of your favorites"), and the AI summary references them by name. Indicators you didn't pick are still calculated under the hood, but they don't dominate the explanation.

Optionally also set Indicator settings — custom parameters (e.g. RSI period 21 instead of the default 14, or MACD with different EMAs). Leave blank to use community-standard defaults.

Stop-loss strategy

Six options for how the AI should suggest your exit-on-loss:

• None — No stop-loss (typical for buy-and-hold long-term investors)
• Fixed — Fixed % or $ distance below entry (e.g. always 5% below)
• ATR-based — Distance scales with the stock's volatility (Average True Range). A volatile small-cap gets a wider stop than a stable mega-cap.
• Trailing Fixed — Moves up with price by a fixed % (e.g. always 5% below the highest price since entry)
• Trailing ATR — Trailing stop with ATR distance
• Support-based — Stop placed just below the nearest technical support level

Take-profit strategy

Five options for how the AI should structure your exit-on-profit:

• None — No take-profit, hold "forever" (long-term investor style)
• Single Target — One profit target, exit fully when hit
• Multiple Targets — 2 or 3 targets, scale out a fraction of the position at each (set Target count separately)
• Trailing — Let the trend run, exit when momentum reverses
• Resistance-based — Exit at the nearest technical resistance level

Minimum risk/reward ratio

A trade with a stop-loss at −5% and a target at +10% has a risk/reward (R/R) ratio of 2.0 — you risk $1 to make $2. If you set min_risk_reward_ratio = 2.0, the AI won't recommend trades below that ratio. They'll still be shown, but flagged as "below your R/R threshold." Conservative traders often set this to 2.5 or 3.0.

Analysis quality filters (advanced)

Eight toggles that control how strict the AI is before calling a trade "actionable." Turning them off doesn't change the score itself — it just means the filter no longer blocks the trade from being recommended.

• Market Context filter — Considers the broader market (SPY trend, regime). Blocks bullish trades during clear bear regimes.
• Entry Quality filter — Checks if the current entry level is favourable (not chasing a parabolic move).
• Confidence filter — Blocks low-confidence trades from being flagged actionable.
• Momentum Exhaustion filter — Warns when RSI is >80 or volume has dried up — late-trend signals.
• MTF (Multi-Timeframe) filter — Requires daily and weekly trends to agree before raising the Technical score.
• Overbought/Oversold filter — Penalises buying overbought or selling oversold (mean-reversion guard).
• Conflict Penalty filter — Penalises Unified score when Technical / Sentiment / Fundamental disagree strongly.
• Volume Confirmation filter — Strong moves on low volume are downgraded vs strong moves on high volume.

Bio completion %

The bio editor shows a completion percentage based on how many key fields you've filled. You can run analyses with an incomplete bio (defaults kick in for missing fields), but the personalisation gets better the more you fill out. Aim for at least 80%.

Conversation mode (AI fills it for you)

If filling 13 fields by hand sounds tedious, open the Trader Bio page and use the chat interface instead. Tell the AI in plain English how you trade ("I'm a beginner, ~$30k account, hold things 2–4 weeks, prefer fundamentals, don't like aggressive stops"). The bot extracts your trading style, risk tolerance, portfolio size, etc., and asks follow-ups for anything missing. End result: a complete bio without ever touching a dropdown.

Analyses

Nostradamus offers four analysis types. The Complete Analysis is the headline product that fuses all three dimensions into a single verdict; the individual analyses let you drill into specific lenses when you want focused detail.

How analyses are scored

Every analysis produces a score in the range −100 to +100 on a continuous scale:

• +70 to +100 — STRONG_BUY signal. Multiple dimensions strongly bullish, low conflict.
• +30 to +70 — BUY signal. Bullish bias but with some caveats.
• −30 to +30 — HOLD / neutral. Mixed signals, wait for clarity.
• −70 to −30 — SELL signal. Bearish bias.
• −100 to −70 — STRONG_SELL. Strong bearish across dimensions.

Alongside the score, every analysis also produces a Confidence score (0–100%). Score tells you direction; confidence tells you reliability. A +60 score at 30% confidence is "weak buy in noisy conditions"; a +60 score at 85% confidence is "high-conviction buy." Use score to decide direction; use confidence to size the position.

Complete Analysis (Unified) — the flagship

Runs Technical, Sentiment, and Fundamental analyses in parallel, then synthesises a personalised verdict. This is what you'll run 80% of the time.

Output includes:

• Unified score — weighted average of the three component scores, with minority consensus dampening applied. If only one of three dimensions agrees (e.g. only Sentiment is bullish, Technical and Fundamental are bearish), the unified score is dampened by 40% to avoid false confidence.
• Recommendation: STRONG_BUY / BUY / HOLD / SELL / STRONG_SELL based on the unified score
• Direction: UP / DOWN / NEUTRAL — the market direction signal
• Confidence breakdown — base 24 + bonuses for signal agreement, MTF alignment, fundamentals quality, volume confirmation. Penalties for signal conflict, momentum exhaustion, low data quality.
• Personalised verdict — 1-paragraph plain-language summary tuned to your bio (style + risk + experience level)
• Bull case — 2–4 bullet points the AI uses to argue for the trade
• Bear case — 2–4 bullet points arguing against the trade (so you see both sides)
• Trade setup — concrete entry price, stop-loss, take-profit target(s). Built from your bio's stop-loss type, TP strategy, and target count.
• Snowflake chart — visual 5-axis radar showing Technical, Sentiment, Fundamental, Confidence, and Unified scores at a glance
• Style-specific dashboard — Swing view emphasises entry quality; Long-Term view emphasises fundamentals + alignment

Technical Analysis

Pure price & volume analysis. Run this when you care about chart action and want to skip news / fundamentals (e.g. day-of trading decisions).

Computes:

• Trend indicators: SMA 20/50/200, EMA 9/21/55, MACD, ADX (trend strength)
• Momentum indicators: RSI (14), Stochastic (14,3,3)
• Volatility: Bollinger Bands (20, 2σ), ATR (14)
• Support & resistance: auto-detected pivot levels (1d, 2d, 3d strong levels)
• Multi-Timeframe (MTF) trend: confirms direction by checking daily and weekly trends agree. Disagreement = lower technical score.
• Volume profile: confirms whether the move has volume backing it
• AI narrative: tells the story in plain English — what's working, what's not, and what to watch

Sentiment Analysis

News & social-media sentiment over the last 5 days, scored by an LLM article-by-article. Run this when news flow matters (e.g. earnings week, FDA decisions, M&A rumours).

• Per-article scoring — every article gets a sentiment score from −100 to +100 by the LLM
• Time-weighting — articles from the last 24h count 4× more than 5-day-old articles
• Composite score — weighted average that becomes the Sentiment score
• Dominant theme detection — what the market is talking about (e.g. "AI catalysts," "regulatory risk," "earnings beat")
• Bull / Bear case generation — the strongest bullish and bearish narratives extracted from the articles
• SEC filing context — recent 8-K, 10-Q, 10-K filings are pulled in for material event context

Fundamental Analysis

Long-term value analysis using SEC-filed financials. Critical for Long-Term and Position trading; less relevant day-trading.

Five sub-dimensions, each scored separately:

• Valuation — P/E (TTM & Forward), PEG ratio, P/B, P/S, EV/EBITDA. Compared against sector medians.
• Profitability — Gross / Operating / Net margins, ROE, ROA, ROIC. Higher = better quality business.
• Growth — Revenue YoY, EPS YoY, multi-year CAGR. Distinguishes growth stories from value plays.
• Financial Health — Debt/Equity, Current Ratio, Quick Ratio, Interest Coverage. Flags balance-sheet risk.
• Efficiency — Asset Turnover, Inventory Turnover. How well capital is utilised.

The overall Fundamental score is a weighted blend of all five. Long-term bios weight Valuation + Health higher; aggressive growth-style bios weight Growth + Profitability higher.

Reading the score chart

Caveats — when to trust the result

• Cached results are free: if the same (user, bio, symbol) was analysed in the last ~hour, the page serves the cache. Force a re-run with the Refresh button (tokens charged).
• Earnings windows: sentiment can be noisy around earnings — confidence is lowered automatically. Wait 24–48h after print before trusting the new sentiment score.
• Penny stocks & thin volume: Technical indicators get less reliable below ~$5 share price or under ~500k average daily volume. The volume confirmation filter penalises this automatically when enabled.
• News-free stocks: small-caps often have 0 news articles in 5 days. Sentiment will return NO DATA, and the Complete Analysis falls back to Technical + Fundamental only.
• Recently-IPO'd stocks: less than 6 months of public history makes Fundamental and MTF unreliable. Score is lowered and confidence dampened.

Backtesting

Backtesting answers: "If I had followed Nostradamus's recommendation on date X, how would the trade have turned out?" It's how you validate the platform's edge before trusting it with real capital.

How a backtest actually works

1. Pick an entry date in the past (e.g. 2025-01-15)
2. Nostradamus runs a point-in-time Complete Analysis using only data that was available on that date — no future information
3. The recommendation (BUY/HOLD/SELL) is simulated as an actual trade with the stop-loss and take-profit from your Trader Bio
4. The trade is then "held" for the evaluation window (14d for Swing, 30d for Position, 26 weeks for Long-Term)
5. During the holding period, the system tracks day-by-day whether the price hit your stop-loss, hit each take-profit target, or exited at window-end
6. The trade is scored: WIN / LOSS / BREAKEVEN, with P/L%, max drawdown, max profit, targets hit

The point-in-time reconstruction is what makes backtests meaningful — we don't peek at the future when generating the analysis.

Single backtest

Pick one symbol + one entry date. Useful for:

• Sanity-checking the analysis against a date you remember well
• Validating the system on a specific known event (e.g. "What did Nostradamus say about NVDA the day before earnings?")
• Tuning your Trader Bio — try different stop-loss strategies on the same date and see which would have worked best

Batch backtest — the real edge tester

Run many single backtests across a date range to generate statistically meaningful win-rate and P/L numbers.

Pricing is 1 token per evaluated entry day. For daily styles (Swing, Position), one entry is generated per trading day (Mon–Fri, skipping weekends and holidays). For Long-Term, one entry per Monday (weekly cadence).

Typical date ranges and token costs (Swing/Position styles):

• 30 calendar days ≈ 22 trading days = 22 tokens
• 90 calendar days ≈ 65 trading days = 65 tokens
• 6 months ≈ 130 trading days = 130 tokens
• 1 year ≈ 252 trading days = 252 tokens

Long-Term style is cheaper (1 entry/week): 6 months = 26 tokens, 1 year = 52 tokens. The dashboard shows the exact cost for the range you've picked, in real time, before you confirm.

What you get back from a batch

• Win rate — % of trades that closed in profit
• Average profit % — mean P/L across all closed trades
• Profit factor — total wins ÷ total losses (above 1.0 = profitable strategy)
• Max drawdown — worst single trade across the batch
• Distribution — bar chart of W/L/BE counts, profit distribution
• Trade-by-trade table — every single entry with its outcome, sortable
• Equity curve — cumulative P/L over time if you'd traded every entry

Evaluation window per trading style

• Swing — 14 trading days. Daily candles. Suits "in for 2 weeks, out" strategies.
• Position — 30 trading days. Daily candles. Captures longer mean-reversion or trend trades.
• Long-Term — 26 weeks. Weekly candles. Suits "buy & hold for a quarter or two."

The window is fixed by trading style — you can't shorten or extend it. The reason: changing the window changes what counts as a "win" and breaks comparability across batches.

What backtests don't model

• Slippage — assumed entry at the close of the entry day. Real fills can be worse on illiquid stocks.
• Commissions / fees — not subtracted from P/L. Most modern brokers are $0 commission, but options or international fees aren't modelled.
• Survivorship bias — we use Marketstack data which includes delisted tickers, but you may have selection bias in which symbols you backtest.
• News & sentiment of the entry day — Sentiment analysis on backtests uses news available on the entry date only, which is correct, but very recent news (hour-of) might not have been indexed yet.

How to interpret results

• Win rate > 60% with profit factor > 1.5 = strong strategy in that market regime
• Win rate ~50% with profit factor > 1.5 = winners are larger than losers — still profitable
• Win rate > 60% with profit factor < 1.5 = many small wins, few large losses — fragile, watch tail risk
• Win rate < 40% = the strategy doesn't fit this stock / regime; try a different bio or different symbols

One stock isn't enough — run batches across a basket of 5–10 symbols you care about to see whether the edge is real or sample-of-one luck.

Watchlist

Your Watchlist is the central hub for stocks you're tracking. It's accessible from every dashboard (Technical, Sentiment, Fundamental, Complete Analysis) and has its own dedicated page.

Adding stocks

Three ways to add a stock:

1. Watchlist page — type a symbol in the add-field, hit Enter
2. Star icon on any analysis detail page — single click adds the stock you're viewing
3. Autocomplete — type the first letters and pick from suggestions (it shows the company name to help disambiguate, e.g. AAPL vs APRN)

You can add notes to any watched symbol — useful for tagging why you added it ("earnings 11/14," "potential breakout above $250").

The dashboard view

Each analysis page shows your watchlist as a table with:

• Symbol + company name
• Price with 1-day change %
• Cached score from the most recent analysis (if any)
• Signal badge — STRONG_BUY / BUY / HOLD / SELL / STRONG_SELL
• Direction / trend badge
• Last analyzed timestamp
• Actions — view (opens detail), play (re-run analysis)

Three ways to analyze from the dashboard

• Per-row Play button — one stock at a time. Cost chip on the button shows tokens.
• Multi-select — tick checkboxes for specific stocks, then click "Analyze Selected (N) — N×cost tokens." Useful for analysing only the 3 stocks with earnings this week, not all 20 in your watchlist.
• Analyze All — one click, every stock in the list. Cost chip shows the total tokens for the entire watchlist.

Cached vs fresh

When you open a dashboard, Nostradamus pulls the most recent cached analysis for each stock (no tokens charged for the page load itself). The action buttons only spend tokens when you actually re-run the analysis. So you can browse the dashboard for free; you only pay when you ask for fresh data.

Removing & reordering

Click the trash icon on any row to remove. Drag rows (on the Watchlist dashboard) to reorder. Removal is undoable for a few seconds via toast notification.

Watchlist scope

Tokens

Tokens are Nostradamus's in-app currency for paid actions. Pay-as-you-go, never expire, fair refunds. No subscription required to use the platform — though subscribing gets you cheaper tokens and bonus discounts on top-up packs.

Why a token model?

Stock analyses cost real resources to run: API calls to Marketstack and StockNewsAPI, GPU time on LLM providers, server compute. A flat subscription would either over-charge light users or push us to limit heavy users. A token model means you pay for what you use, nothing more.

Tokens also make the cost of each feature transparent. No hidden tiers, no "premium-only" features — every action has a visible price and your balance is always one click away.

Pricing — what every action costs

Welcome bonus

Every new account gets 300 tokens free on registration. Use it however you want:

• 10 Complete Analyses (30 tokens each)
• or 30 individual analyses (10 tokens each)
• or a 1-year batch backtest of a single Long-Term style stock (52 tokens), still leaving 248 for other uses
• or a ~22-day Swing batch backtest across 13 different stocks

The bonus is one-time, automatic on sign-up, no opt-in needed.

Carry-over policy — your tokens never expire

Whether they came from your welcome bonus, a top-up purchase, or a subscription, your tokens stay in your account indefinitely.

This is intentional. Many SaaS products use "use it or lose it" monthly credits to drum up urgency, but it punishes infrequent users and erodes trust. Our position:

• You bought / earned those tokens — they're yours
• Stockpile them for burst usage (earnings season, market volatility, big-news weeks)
• Take a 3-month break, come back, your balance is still there
• Subscription tokens accumulate if you don't use them all — no pressure to "consume the allocation"

Token packages (pay-as-you-go)

Buy more tokens any time on the Tokens page. Bigger packs have a lower per-token cost:

• Starter — 200 tokens / 99 Kč (0.495 Kč/token)
• Standard — 500 tokens / 199 Kč (0.398 Kč/token, 20% discount)
• Pro Pack — 1 500 tokens / 499 Kč (0.333 Kč/token, 33% discount)
• Enterprise — 5 000 tokens / 1 299 Kč (0.260 Kč/token, 47% discount)

Subscribers get an additional discount on top of these prices (5–35% depending on tier).

Subscription tiers (optional)

If you analyse stocks regularly, a subscription is cheaper than top-up packs alone:

• Free — 0 Kč/month, no monthly tokens (just the one-time welcome bonus), 0% discount
• Basic — 149 Kč/month, 300 tokens monthly, +5% bonus discount on top-up packs
• Pro — 399 Kč/month, 1 000 tokens monthly, +15% bonus on top-up
• Premium — 899 Kč/month, 3 000 tokens monthly, +25% bonus on top-up
• Trader — 2 499 Kč/month, 10 000 tokens monthly, +35% bonus on top-up

Subscription tokens carry over month-to-month just like everything else. The bonus discount stacks with package base discount — a Trader-tier subscriber buying an Enterprise pack pays just 844 Kč for 5 000 tokens (0.169 Kč/token, the lowest price in the system).

Refunds — automatic, no manual claims

If a paid action fails (data provider down, server error, timeout), your tokens are automatically refunded within seconds. You'll see a transaction labelled "Refund" in your Recent Activity, linked to the failed task.

Important details:

• Refunds are idempotent — even if our system retries, you get refunded exactly once
• Refunds happen on technical failure, not on "the recommendation was wrong" — the value you paid for was the analysis itself, not a winning trade
• If a batch backtest fails midway, the whole batch is refunded (we don't do partial refunds)

Pre-pay model — why tokens are deducted up-front

We charge the tokens before the analysis runs, not after. Two reasons:

1. Abuse prevention — without pre-pay, a single click could queue 100 analyses for free. With pre-pay, your balance is the only thing standing between you and the bill.
2. Industry standard — OpenAI, Anthropic, Stripe API, all consumer-facing API-credit products work this way

The automatic refund on failure makes this fair: if anything goes wrong, you don't pay.

Confirmation modal

By default, a modal asks you to confirm every paid action ("Spend 30 tokens?"). It shows the cost, your balance, and what you'll have left after the action. Tick "Don't ask again" to skip future confirmations — useful once you're comfortable with the costs.

You can re-enable confirmations any time in Settings → Tokens & Spending. There's a toggle there called "Ask before spending tokens."

Cached analyses don't cost tokens

If the same (user, bio, symbol) was analysed in the last ~hour, opening the page or running a Complete Analysis on a stock that already has a recent cache serves the cached result — no tokens spent. You only pay when you click the Refresh button (which forces a re-compute).

Transaction history

The Tokens page shows every charge, refund, top-up, and welcome bonus in chronological order. Each entry includes:

• Operation type (e.g. analysis_unified)
• Symbol context (e.g. AAPL)
• Amount (positive = credit, negative = debit)
• Balance after the transaction
• Linked task ID (for refunds and analyses)
• Timestamp

Pagination: 5 transactions per page. Full audit trail — you can always reconcile what you spent and on what.

Payments — current state