Skip to main content
← Back to Docs
💼Portfolio & Analytics

Trade Journal & Analytics

Log trades with thesis & conviction; analytics reveal edge

The Trade Journal turns your trading from a sequence of disconnected bets into a dataset you can actually learn from. Every trade you log — thesis, tags, conviction, outcome — feeds an analytics engine that breaks your performance down by category, tag, conviction level, and even the hour of day you opened the position. The point is simple but uncomfortable: over a small sample, luck and skill look identical. The journal is how you tell them apart.

You'll find it at /dashboard/journal, or under Tools → Trade Journal in the sidebar.

Why Journaling Separates Skill From Luck

A 60% win rate over 10 trades means nothing — you could flip a coin and get that. Over 200 trades, a persistent edge starts to show through the noise. But raw win rate hides the more useful truth: you are probably good at some kinds of trades and quietly terrible at others.

Most traders carry a vague self-image ("I'm a good contrarian", "I crush crypto markets") that has never been checked against data. The journal checks it. When you bucket 150 logged trades by category and see that your crypto trades are -$340 expectancy while your politics trades are +$22 per trade, the story rewrites itself. You're not a crypto trader who had a bad month. You're a politics trader who keeps donating to crypto markets.

That's the entire value proposition: a feedback loop tight enough to change behavior. The scanner and AI tools tell you where edge might exist. The journal tells you where *your* edge actually exists.

What the Journal Captures

Each entry stores a structured record of one trade:

  • •Market — the question (e.g. "Will Bitcoin reach $150k by Dec 31, 2026?"). Required.
  • •Category — a free-text bucket like crypto, politics, sports, economics. This drives your most important breakdown, so keep it consistent.
  • •Direction — YES or NO.
  • •Size ($) — position size in USD.
  • •Entry / Exit (¢) — prices entered in cents (e.g. 50 for a $0.50 contract). The platform stores them as decimals internally.
  • •P&L ($) — realized profit or loss. Leave this blank while the trade is open; the analytics treat an entry without P&L as "open" and exclude it from win-rate math until you close it.
  • •Conviction (1–5) — how strongly you believed at entry, shown as ★ stars. This is the field that most traders skip and most regret skipping. More on it below.
  • •Tags — comma-separated labels like news-driven, contrarian, high-conviction. Up to 20 tags per entry, each up to 50 characters.
  • •Thesis — why you entered, in plain language. Stored and shown as plain text, so Markdown syntax appears literally. Up to 5,000 characters.

After a trade closes you can also add outcome notes (what actually happened, and whether your thesis held) and a self-rating (1–5) of *process* quality — distinct from whether you won. A trade can be a 5-star decision that lost, or a 1-star gamble that won. Tracking both is how you stop rewarding luck.

Logging a Trade, Step by Step

  1. Open /dashboard/journal and click + Add Entry.
  2. Fill in the Market title (required) and pick a Category. Use a category you've used before so the bucket aggregates correctly — crypto and Crypto and BTC are three different buckets to the analytics engine.
  3. Set Direction (YES/NO) and Size.
  4. Enter Entry price in cents. If you're logging at open and don't yet have an exit, leave Exit and P&L blank.
  5. Set your Conviction from 1 to 5. Do this *before* you know the outcome — that's the whole point.
  6. Add Tags that describe the *kind* of trade, not the market. Good tags: earnings-play, whale-follow, mean-reversion, tilt. Bad tags: the market name (that's already the title).
  7. Write a one-paragraph Thesis. Even two sentences ("AI shows 12pp edge, polling supports it, liquidity is thin so sizing small") is enough to be useful in three months.
  8. Click Save Entry.

When the trade resolves or you exit, find the entry in the Recent list and click Close entry. That opens a small form for the Exit price (¢), P&L, outcome notes and your self-rating. Typing an exit price auto-suggests the P&L from size ÷ entry × (exit − entry) — overwrite it whenever your actual fill differed, since the suggestion assumes you got the price you typed. Reopen the same form later with Edit outcome if you need to correct it. Only entries with a P&L value count as "closed" in the analytics — so an honest journal requires you to come back and close trades out.

Tip: log the entry at the moment you place the trade, not at the end of the week. Thesis quality collapses with hindsight. Once you know the result, your memory quietly rewrites why you entered.

Conviction: The Field That Earns Its Keep

Conviction (1–5) is the single most diagnostic field in the journal, because it lets the analytics answer a question nothing else can: does your confidence predict your results?

A skilled trader's conviction is *calibrated* — their 5-star trades win more often and earn more per trade than their 2-star trades. If the journal shows your ★5 trades have a 48% win rate and your ★2 trades have a 61% win rate, your internal confidence signal is inverted. That's gold: it means your gut is actively misleading you, and you should size *down* on your "sure things."

The By Conviction breakdown tab plots win rate, average P&L, and expectancy for each star level (1 through 5, plus an "unrated" bucket for entries you left blank). Over a few hundred trades, the shape of that curve tells you whether to trust yourself. Pair it with the Kelly Calculator: conviction is only a safe input to position sizing once the data shows it's calibrated.

Reading the Analytics

At the top of the page, five cards summarize everything you've logged:

  • •Total Trades — count of all entries, with how many are closed.
  • •Win Rate — wins / closed trades. Color-coded: teal at ≥55%, amber at 50–55%, rose below 50%.
  • •Total P&L — sum across closed trades, with average per trade.
  • •Expectancy — your average expected dollars per trade, plus your RR ratio.
  • •Avg CLV — average closing line value, in percentage points: the latest price Predite has seen for the side you took, minus the price you paid. It only counts entries in markets the scanner has quoted, and the card shows how many those are. Positive CLV means you were repeatedly early to the move, which is the one quality signal that doesn't need the bet to have resolved yet.

Below that, a row of tabs switches the breakdown table between five buckets: By Category, By Tag, By Hour of Day, By Day of Week, By Conviction. Each row in the table shows the same columns so they're directly comparable:

  • •Trades — total entries in that bucket.
  • •WR — win rate (shows "—" if nothing in the bucket is closed yet).
  • •P&L — total realized for the bucket.
  • •Avg — average P&L per closed trade.
  • •RR — risk/reward ratio = average win ÷ average loss. Above 1.0 means your winners are bigger than your losers. Shows ∞ when you have wins but no losses in the bucket (small sample — don't over-read it).
  • •Expect. — expectancy = (avg win × win rate) − (avg loss × loss rate). This is the number that matters most. A bucket can have a mediocre win rate and still be your best category if the wins are large enough.

How to actually use the breakdowns

  • •By Category — find which market types you should keep trading and which to cut. Negative expectancy in a category over 30+ closed trades is a strong signal to stop trading it, regardless of how it "feels."
  • •By Tag — this is where strategy-level insight lives. If contrarian trades are +$18 expectancy but momentum trades are −$25, you've learned something about your style that no amount of reflection would surface.
  • •By Conviction — calibration check (see above).
  • •By Hour of Day / Day of Week — buckets are computed in UTC from when you opened the trade. The classic finding here is "trades I open after 11pm have a 34% win rate" — i.e. tilt and tired decisions. If late-night trades bleed money, that's a behavioral fix worth more than any signal.

Gotcha: the By Tag breakdown counts one trade in *every* tag it carries, so tag totals can exceed your trade count. That's intentional — a single trade can be both high-conviction and news-driven — but it means tag buckets aren't mutually exclusive. Don't sum them expecting your total.

Gotcha: every breakdown ignores open trades. Win rate, RR, and expectancy are computed only over entries that have a P&L. If your numbers look thin, it's usually because you logged entries but never closed them out.

Connecting the Journal to the Rest of Predite

The journal is most powerful when it's not a separate silo, and it no longer is one. The Import your trades card pulls your Paper, Bot and Copy trades in as entries, each tagged by origin (paper, bot, copy) so the By Tag breakdown can compare what your bot did against what you did. Import is something you trigger, not something that happens on every fill: running it again brings only what is new, because each entry remembers the trade it came from and already-imported trades are skipped. Up to 200 come in per run. You can still type in trades from anywhere, including venues outside Predite — which is why the same ruler can cover your whole trading life rather than only the part that runs on this platform.

A practical workflow:

  1. Find a signal in the EV Scanner, sized with the Kelly Calculator.
  2. Validate the approach in Paper Trading first.
  3. Once live (Bot plan), log each real trade in the journal with thesis and conviction.
  4. Weekly, open the journal and read the By Category and By Conviction tabs.
  5. Feed what you learn back into your scanner filters and Kelly conviction inputs.

This closes the loop: scanner → sizing → execution → journal → adjusted scanner filters.

Publishing a public track record

Set a handle in the *Public journal* card and any entry you flag as public appears at /journals/<handle>, readable without a login. The page shows those entries plus aggregate stats — public trade count, win rate, total P&L and W/L record. Entries stay private until you flip them: each row in the Recent list carries a 🔒 Private / 🌐 Public toggle, and nothing is published by accident.

A handle is 3–30 characters, lowercase letters, numbers, hyphen or underscore, and unique across Predite. Removing it takes the page offline immediately without changing which entries are flagged.

Exporting

The ↓ Export CSV button above the Recent list downloads exactly what the page has loaded: one row per entry with market, category, direction, size, entry and exit price, P&L, conviction, tags, and the opened and closed timestamps. Thesis, outcome notes, self-rating and CLV are not in the file — they are written for you, not for a spreadsheet.

Programmatic access. GET /api/v1/journal returns the same entries to a script, filtered by tag, category and since, with limit up to 500. It authenticates with an API key — which is Bot plan only — and the response carries a truncated flag so a caller that receives exactly limit rows knows whether more exist.

The dashboard loads your 500 most recent entries, and that is also the ceiling on one export. Everything on the page — the five cards, every breakdown — is computed from that same window, so a journal past 500 entries is showing you your recent history rather than your whole history.

For tax-shaped output — cost basis, realized gains, Form 8949-style rows — use the export described in Cost Basis. That one is built from your actual positions rather than from what you typed here, which is the right source when the number has to survive an audit.

Tip: export to a spreadsheet quarterly and run your own pivots. The built-in breakdowns cover the common cuts, but your own questions ("how do I do on markets resolving in under 7 days?") are easy to answer with the raw CSV.

Plan Requirements

The Trade Journal and its analytics are included on the Pro ($59/mo) and Bot ($99/mo) plans. It is not available on Starter — the upgrade screen appears in its place. CSV export rides along on the same Pro/Bot access.

Note that logging a trade doesn't require live trading — you can journal paper trades, copy trades, or trades you made anywhere, including outside Predite. Live CLOB execution on Polymarket is a separate capability that requires the Bot plan and a connected wallet, but the journal happily records trades from any source.

Common Mistakes

  • •Only logging winners. This is the cardinal sin. It inflates every metric and teaches you nothing. Log losses *first* — they're where the lessons are.
  • •Skipping conviction. Without it, the calibration analysis is impossible and "unrated" swallows your data.
  • •Inconsistent categories. crypto, Crypto, and BTC fragment into three useless buckets. Pick a vocabulary and stick to it.
  • •Never closing entries. Open trades are invisible to the analytics. Come back and add P&L.
  • •Reading tiny samples. A bucket with 4 trades and a 100% win rate tells you nothing. Wait for 30+ closed trades per bucket before drawing conclusions, and treat ∞ RR as "too small to judge."
  • •Logging from memory at week's end. Your remembered thesis is fiction. Log at the moment of the trade.

Related Docs

← All Documentation