Signals and copy score
Copy Score helps you see when a trade is more likely to beat the market.
We at Polycopy built Copy Score to give users a clearer answer than raw statistics and raw signals can. We use machine learning and trade profiling to test whether a trader has historically created edge in this exact kind of setup, then publish that read directly on our trade cards.
Example trade
Will BTC close above $150k by June?
Copy Score
74
Win Rate
58%
Avg P&L
+12% / +$184
Experience
412 trades
Conviction
2.1x
Why copy score matters
Why not just look at traditional, simple trading stats?
Because strong-looking surface stats can still hide weak pricing. Raw signals are complex, easy to misread, and they are not trained to deeply understand performance indicators across trade context. Copy Score exists to give users that machine-learned layer.
The core idea is simple: does this trader beat the market on trades like this one?
Traditional trading stats
A trader can post strong-looking stats and still add very little value if they mostly buy outcomes the market already prices correctly. Traditional stats are useful inputs, but on their own they leave the key question unanswered.
Copy Score
Copy Score brings price, context, and machine-learned pattern recognition back into the picture. In this study set, all trades averaged +0.0%, while the 70–100 Copy Score range averaged +5.8%.
How copy score works
What flows into Copy Score
Polycopy does not judge trades in a vacuum. The system starts with trade profiling and trader stats, then runs that context through our machine learning algorithm to produce one score.
Trade profiling helps Polycopy understand why the trade is being made and what kind of setup it belongs to. That visible layer feeds the model, but the machine learning is what turns those signals into a more intelligent final read.
Trade class
Market category
Entry band
Bet type
Trader stats
Conviction
Win rate
Experience
ROI and P&L
ML algorithm
50M+ historical trades
Trade profiling inputs
Pattern recognition
Score calculation
Copy Score
74
The visible indicators help users understand the trade, but on their own they are not as useful as Copy Score because they are not always pointing in the same direction. The model combines them with many additional dimensions to make the final read faster and clearer.
Pure edge, not market probability
Copy Score is calculated after removing the market price from the probability scoring. The entry price a trader pays is factored out so that the score reflects only the statistical edge the trader brings to that type of trade — not how cheap or expensive the position was. A high Copy Score means the trader has historically outperformed what the market price alone would predict, regardless of whether they entered at 20¢ or 80¢. This is true edge, not a reflection of market odds.
Copy score ranges
How to read Copy Score
The grouped ranges below show the historical pattern clearly. As the score rises, average trade quality improves, with the strongest outcomes concentrated in the top end of the range.
The simplest way to read the page is this: the low end of Copy Score is the weakest historical pocket, the middle needs more care, and the strongest edge sits at the top. In this study set, the cleanest outcomes show up in the 70–100 range.
Market category and trade context
Polycopy does not judge trades in a vacuum.
For every trader, Polycopy uses trade profiling to categorize markets and link those categories back to historical performance. That makes Copy Score smarter than a generic lifetime average.
Market category
A trader can be strong in one category and ordinary in another. Copy Score uses market category so a crypto trade is not judged the same way as a politics trade.
Entry band
Price matters because the market is already telling you the implied probability. A trader’s history at one price range can look very different from another.
Bet type
Different market setups behave differently. Copy Score uses the shape of the trade, not just the market label, to compare a trader against the right context.
Other trade signals
The visible indicators help users understand why a trade looks stronger or weaker.
These are not the whole model, but they make the logic more intuitive. They show how trade quality has shifted across conviction, experience, win rate, and average ROI.
Conviction
Conviction compares this bet size to the trader’s usual size. When traders size up with purpose, outcomes can improve.
| Conviction | Avg P&L | Trades |
|---|
Experience
Experience tells you how much history the model has to work with and where the strongest pockets have shown up.
| Experience | Avg P&L | Trades |
|---|
Win rate
Win rate matters only in context. This table shows why surface-level win percentage alone can be deceptive.
| Win rate | Avg P&L | Trades |
|---|
Average ROI
Polycopy looks at return quality, not just raw activity. In the product, this also sits alongside dollar P&L context.
| Average ROI | Avg P&L | Trades |
|---|
These are the visible indicators users can reason about on the page. Copy Score uses them together with many other dimensions to produce the final read.
How it's built
The score looks clean because a lot of modeling work happened behind the scenes.
We wanted real conviction in the score before putting it in front of users. That meant testing the system hard, not just building a nice-looking number.
50M+
historical trades behind the headline claim
86.2M
resolved trades in the warehouse snapshot
374k
tracked markets
3.5k
tracked traders
Rolling and shifted windows
We used rolling and shifted time windows so the score had to hold up across changing market regimes, not just one convenient slice of history.
Multiple model families tested
Polycopy tested different model types and kept pushing until the results held up well enough to trust the score in real product surfaces.
Context-rich features
The score benefits from multi-dimensional context: trader history, market category, price, bet sizing, and the structure of the trade itself.
That is why Polycopy’s signal system is a real product differentiator. The intelligence comes from connecting the trade to the right trader history, then stress-testing whether that signal still works across different windows and different modeling approaches.
Snapshot as of 2026-03-16. Backtest window on this page: 2026-01-15 to 2026-03-17 across the top 100 wallets by realized P&L in the study set.
How to use it
Copy Score should help you look at a trade faster, not blindly follow it.
It is an indicator. It is there to help users decide whether a trade deserves deeper attention.
On the trade card, Copy Score should help you decide whether the setup is worth inspecting more carefully.
A stronger score means the trader, the context, and the supporting signals are lining up in a historically better way.
The best way to use the score is alongside the visible indicators. A strong score with healthy conviction, enough experience, and believable performance context is much more useful than any one stat in isolation.
Price movement matters. If the market price drops after entry, Copy Score can improve because the trade may become more attractive. In the product, Polycopy flags trades where price has dropped more than 20% since entry so users know the score may look better than it did at the moment the trader entered.
That is why the score should guide attention, not replace judgment. It helps users see edge faster. It does not remove risk.
Performance data
Does Copy Score actually work? The data says yes.
We built a dedicated page showing full backtest results so you can evaluate the score's track record for yourself.
Methodology notes
The methodology is deliberately lighter than the modeling work behind it.
We want readers to feel informed without turning this page into a technical white paper. These are the points that matter most.
Held-out validation
The score was checked on unseen historical trades. The goal was not just a good in-sample story, but evidence that the signal generalizes.
Contextualized trader data
Trader performance is evaluated in context, not as one flat lifetime average. That is a major reason the score is more useful than surface-level stats.
Historical limits
Signals are built from historical outcomes. Markets change, trader behavior changes, and no backtest guarantees future results.
Indicator, not recommendation
Copy Score helps users evaluate trades more intelligently. It is not a buy or sell instruction.
Study snapshot generated 2026-03-16. The score ranges on this page use 0 resolved test trades from the top 100 wallets in the study set.
Start using Copy Score
Browse the feed, spot high-scoring trades, and make faster decisions on Polymarket.
Risk disclaimer
Trading carries real risk.
Copy Score and the supporting signals are informational tools. They can help users think more clearly about trade quality, but they do not guarantee performance.
Past performance does not guarantee future results. Traders with strong histories still lose, and even the best-looking setups can fail.
Copy Score can also change as price changes. A trade can look better or worse after entry if the market moves materially.
Polycopy does not provide financial advice. Use the score and the signals as tools for decision-making, not as instructions.