Quick Start: Paper Trading

A 15-minute playbook for standing up your first paper copy-trade automation, reading the signal, and tuning the filter until it makes money on paper. No capital at risk. Once the numbers look good, the same filter plugs straight into a Real Copy Trade automation.

The loop is: filter → estimate → paper automation → trade flow check → analytics → whitelist/blacklist → repeat.

Step 1: Build a starter filter

This walkthrough uses a Cohort Filter as the starting shape (criteria-based population). For other starting shapes (Whitelist, Hybrid, or Specialist), see Filter Setups first; the rest of this loop applies to any of them.

Go to the Filters tab on the Polymarket dashboard and click +.

Name it something like Starter paper v1, then set these sensible defaults under Trader Criteria:

Under Trade Criteria, set:

Leave Market Criteria empty for the first pass. You’ll narrow by category later using analytics.

Sortino is available as an analytics breakdown dimension (By Trader Sortino). Filter on Sharpe first, slice by Sortino after you have results.

Step 2: Read the estimate

Click the Estimate coverage button at the bottom of the builder. It runs a one-shot estimate against the criteria you’ve entered and shows the result:

1,234 traders   89 markets   (rough estimate)

This tells you how much of Polymarket’s universe currently matches. Target ranges for a starter filter:

The estimate does not auto-refresh. If you change a criterion after running an estimate, the old result is shown with a “stale, click Estimate coverage to refresh” hint. Click the button again to recompute. Tune the knobs until the numbers feel right, then click Save Filter.

Step 3: Attach a Paper Copy Trade automation

Go to the Automations tab and click +.

1. Pick Paper Copy Trade as the automation type
2. Select the filter you just saved (locked after creation, use a new automation to change)
3. Use the default capital settings to start:
   • Starting Capital: $10,000
   • Position Size: 5% (so each new trade is ~$500)
   • Max Exposure: 25% (single position can’t exceed $2,500)
4. Keep the default exit behavior (Follow sells ON, Proportional exits ON, Hold until resolution ON)
5. Leave Active on creation ON and click Create

Your automation starts listening immediately. New matching trades fire paper buys in real time.

Step 4: Check trade flow

Give it 24 to 48 hours, then open the automation detail page and look at Trade Activity.

The goal is signal density that’s actually tradeable:

If you widen or narrow, edit the filter directly. All automations on that filter pick up the new criteria immediately, no need to create a new automation.

Step 5: Read the Analytics tab

Once you have roughly 50 to 100 closed positions (typically 1 to 2 weeks for a healthy filter), open the Analytics tab on your automation.

Top-level summary metrics to focus on:

Then scroll to Performance Breakdown and flip the dimension dropdown. This is where you find what’s actually working and what isn’t.

Step 6: Use breakdowns to find your edge

Switch the breakdown dimension to answer specific questions:

Sort each table by Total PnL descending and then ascending. The extremes tell you everything:

• Top of PnL, high Trades count → that trader / tag / bucket is your real edge. Promote them.
• Bottom of PnL, high Trades count → that trader / tag / bucket is bleeding you. Cut them.
• Small Trades count, either direction → noise, ignore for now.

Step 7: Tighten with whitelists and blacklists

Open the filter in the builder and use the component dropdowns to add whitelists or blacklists:

• Trader Blacklist: add the bottom-PnL traders from the By Trader breakdown. They never match again no matter what other criteria say.
• Trader Whitelist: to go concentrated, add the top-PnL traders and remove the Trader Criteria component. Only those addresses can match now.
• Market Blacklist: add specific markets or category tags (Politics, Sports) that consistently lose money in the By Market Tag breakdown. The blacklist accepts both shapes.
• Market Whitelist: restrict matches to a handpicked list of specific markets. To restrict by category instead, use Market Criteria → Tags (OR logic, market matches if it carries any selected tag), not the whitelist.

Blacklists always win. If a trader or market is on a blacklist, the trade is rejected even if everything else matches.

Other knobs to retune from what the breakdown showed you:

• If By Trader Trade Size says bigger copies win, raise Trade Size min in Trade Criteria
• If By Market Volume says deep markets win, add Volume min in Market Criteria
• If By Entry Price says cheap outcomes win, set Last Traded Price max in Trade Criteria (e.g., 0.7)

Save the filter. The live automation picks up the new criteria on the next matching trade.

Step 8: Rinse and repeat

Give each filter change another 50 to 100 closed positions before judging. Changing too frequently turns the feedback loop into noise: you can’t tell whether the last edit helped or hurt.

A typical rhythm:

1. Week 1: starter filter, 50 to 100 trades, first read
2. Week 2: one or two tightenings based on analytics
3. Week 3: validate the tighter filter, maybe split into two filters for different regimes
4. Week 4+: if Sharpe holds > 1.5 and Max DD stays under 15% on paper, consider a small Real Copy Trade automation pointed at the same filter

When you’re ready for real capital, see Real Copy Trading Setup. The filter is identical, just swap the automation type.

Common mistakes

• Changing the filter every day. You’re not measuring the filter, you’re measuring noise. Hold each version for ~100 closed trades.
• Chasing raw PnL without checking Max Drawdown. A filter can print 30% then vaporize. Always look at both.
• Starting too tight. Zero trades for a week teaches you nothing. Better to start loose, see the full distribution, then cut.
• Ignoring the breakdowns. The summary metrics tell you if it’s working. The breakdowns tell you why, which is how you improve.
• Trusting a single trader. Even great traders have streaks. Three or more traders carrying most of the PnL is a lot safer than one.