Trade Analytics Guide

Overview

The analytics page sits alongside the main dashboard and operates on the same data — the trades your strategies have already closed. Where the dashboard is optimised for what's happening right now, analytics is optimised for what patterns explain my results.

Everything on the page updates together when you change the strategy filter, the paper / live toggle, or the date range. A typical session is: pick a range, scan the key stats, look for any red in the trait breakdowns, and zoom in on the outliers.

What counts as a “closed trade”?

Analytics operates on complete round-trips, not individual fills. A round-trip is one opening fill (copy or fade) paired against the next closing fill on the same token (strategy exit, stop loss, take profit, manual close, or market resolution).

This means:

• Only positions that have actually been closed appear in analytics. Open positions are invisible here by design — they're shown on the dashboard instead.
• Every trait (entry price, hold time, P&L, source) is computed from the single pairing — not averaged across multiple fills.
• Changing the date range clips by exit time. A trade opened before the range but closed inside it is included; the reverse is not.

The nine cards at the top of the analytics page. Here they are computed from a fixed demo dataset so you can see them in context:

Percentage drop from the running equity peak. Zero means you're at a new high; -5% means you're 5% below the best equity you've ever had. The equity curve is built by cumulatively summing per-day P&L from the closed trades in the current filter.

The headline feature. Every closed trade is bucketed along a trait — entry price, hold duration, side, source, exit reason, day of week, hour of day — and each bucket gets its own n, win rate, average P&L, total P&L, and profit factor. The segmented control in the top-right of each panel switches which metric is plotted.

Green bars are net-profitable buckets; red bars are net-losing. Buckets with fewer than three trades are dimmed to flag low confidence.

Entry price bucket

Probably the single most useful trait. Polymarket trades behave very differently at the tails (prices under 0.10 or above 0.90) than in the middle. Finding an entry-price band that consistently loses is permission to add a filter to your strategy.

Signal source

Which source is pulling its weight? If one source has a consistently lower win rate and negative expectancy, you can disable it on the strategy without touching anything else.

Hold duration

If the “under 1h” bucket consistently outperforms everything else, your strategy is really a scalper and your exit rules should reflect that. If long holds win but short ones lose, the opposite is true.

Side

Do BUY and SELL sides of the detected trades actually perform equally? Often they don't — markets near resolution move asymmetrically, and one side may dominate your edge.

Exit reason

How your trades end. Lots of manual_close in your losing bucket is a hint that the strategy's own exit rules are not firing in time. Lots of resolution in the losing bucket means you're holding losers all the way to market close — a classic “hope” pattern.

Three mini line charts tracking win rate, average P&L per trade, and trade count over a 14-day rolling window. For each day in the range, the metric is computed on the slice of trades whose exit time falls within the trailing 14 days.

Three bar charts that show the full shape of your trades — not just the averages. Averages hide skew; distributions don't.

P&L

How realised P&L is spread across eight buckets. A healthy strategy looks like a right-skewed distribution: small losses cluster near zero and a fat positive tail drives the return. A symmetric distribution means your edge is mostly variance.

Hold time

How long you typically hold positions. Pair this with the hold-duration trait breakdown: if most trades sit in the “6–24h” bucket but that bucket loses money, your natural hold time isn't working.

Entry price

Where on the probability curve you tend to enter. Concentrated entries in one price band are common (and usually fine), but they also mean your results are exposed to anything specific about that band — for example, a strategy that only ever trades between 0.30 and 0.50.

If you're new to the page, run through this checklist once a week per live strategy:

1. Filter to one strategy, 30-day range. Analytics aggregated across strategies can hide individual issues.
2. Check the sample size. If you have fewer than ~30 closed trades, every other stat on the page is noisy. Come back in a week.
3. Read the key stats in order: expectancy → profit factor → max drawdown. Expectancy tells you if there's an edge at all; profit factor tells you how durable it is; drawdown tells you what the worst day looked like.
4. Scan trait breakdowns for red bars. Every red bar is a potential filter you could add to the strategy. Dimmed (small-sample) bars are suggestions, not conclusions.
5. Look at rolling metrics for trend. If recent weeks are worse than the aggregate, that's your cue to investigate before the drawdown gets ugly.
6. Test the change in paper first. Clone the strategy, apply the filter, run it in paper mode for a couple of weeks, then compare analytics side-by-side before you promote.

• Small-sample distortion. Everything on this page is empirical — if you have 12 closed trades, your 83% win rate is not a number you should trust. We dim small buckets, but we can't dim the overall stats, so you have to be careful yourself.
• Unclosed positions are invisible. Analytics never incorporates unrealised P&L. A strategy with one large position that's -40% and still open will look clean here until it actually closes.
• Exit time determines membership. Range filters clip by when trades closed, not when they opened. A 7-day range shows trades closed in the last seven days, even if some were open for weeks.
• Analytics is not a backtest. It tells you what already happened on this data. For forward-testing strategy changes, clone the strategy and run it in paper mode. See the Strategy Builder Guide for details.