How to Track a Whale Wallet and Mirror Its Trades

Find a profitable Polymarket trader with the Data API, verify their record properly, then mirror their fills with configurable sizing and risk controls.

Overview

Copying a wallet is one API call. Choosing the right wallet is the entire problem, and it is the part a leaderboard actively works against you on.

A 30-day PnL ranking answers “who made the most money recently”. It does not distinguish a trader with a repeatable edge from someone who made one correct large bet, and the second group ranks higher, because one enormous win beats many steady ones on a leaderboard sorted by total.

This guide is about the verification step in between. Everything here is read-only until the final section, so you can complete the whole selection process without funding anything.

Read time: about 13 minutes. If you want the automated version of this, see Build a copy-trading bot.

TL;DR

• GET /leaderboard ranks by realized PnL, /leaderboard/volume by USDC traded. They surface different people.
• Verify before subscribing: trade count, equity curve shape, and category concentration.
• A trader good in one category is worth copying only in that category.
• Subscribe with simulation: true and leave it there until the numbers are dull.
• Copytrade costs 25 bips per side against 10 on the Trade API, so a busy leader is expensive.
• Pause with PATCH /status, which keeps configuration. DELETE discards it.

What you will do

• Pull candidates from two different leaderboards and understand what each is telling you
• Check a wallet’s real record across three dimensions
• Identify the market category where the edge actually lives
• Subscribe with sizing and filters that reflect what you found
• Paper trade it against live prices
• Go live, and know what to watch afterwards

What you will need

Knowledge

• Comfort with REST APIs. No prediction market background required.

Tools and access

• A Bravado API key. Selection and simulation need no collateral.
• trade.execute only for the final live step

export BRAVADO_API_KEY="your-bearer-token"
export BASE="https://bravado-api-k7kaq.ondigitalocean.app"

Step 1: pull candidates

Two rankings, two different questions:

# who made the most money
curl -G $BASE/leaderboard \
  -H "Authorization: Bearer $BRAVADO_API_KEY" \
  -d window=30d -d limit=25

# who trades the most
curl -G $BASE/leaderboard/volume \
  -H "Authorization: Bearer $BRAVADO_API_KEY" \
  -d window=30d -d limit=25

Ranking	Surfaces	Fails to surface
PnL	Profitable traders	Whether the profit was one bet or a hundred
Volume	Active traders	Whether the activity is profitable at all

For copy trading, pull both and look at the intersection. A wallet high on both is active and profitable, which is a much stronger starting signal than either alone.

Every read here works on any public address with no funded account and no permission from the wallet owner.

Step 2: verify the record

Three checks, each catching a different kind of false positive.

Was it one trade or many?

curl $BASE/traders/0xWALLET/metrics \
  -H "Authorization: Bearer $BRAVADO_API_KEY"

Look at trade count and win rate together. A wallet with 12 trades and a large PnL made one good call. Copying it is a bet that lightning strikes twice, and you will be paying 25 bips a side for the privilege of finding out.

A useful floor is somewhere around 50 trades before a record means anything.

Is the equity curve steady or one spike?

curl $BASE/traders/0xWALLET/pnl \
  -H "Authorization: Bearer $BRAVADO_API_KEY"

Cumulative PnL from the wallet’s first ever trade, not from the start of your window. This is what the leaderboard cannot show you.

Two wallets with identical 30-day PnL can look completely different here: one climbing steadily, the other flat for months with a single vertical jump. Only the first is a strategy you can copy.

Where does the edge actually live?

curl $BASE/traders/0xWALLET/categories \
  -H "Authorization: Bearer $BRAVADO_API_KEY"

This is the check that changes outcomes most, and the one people skip.

Traders are rarely good at everything. Someone who reads politics markets well may be guessing at sports, and their aggregate PnL hides that: the politics profits fund the sports losses and the total still looks good. Copy them everywhere and you inherit both halves.

cats = read(f"/traders/{addr}/categories")["categories"]
for c in sorted(cats, key=lambda c: Decimal(c["realized_pnl"]), reverse=True):
    print(f'  {c["category"]:<14} {Decimal(c["realized_pnl"]):>12,.2f}')

  politics          18,420.50
  crypto             2,105.20
  sports            -6,880.10

That wallet is worth following in politics. Following it everywhere means voluntarily taking the sports losses.

PnL alone is a poor selection signal. A wallet that made one correct large bet outranks a consistently profitable one on a 30-day board. Check the curve and the trade count before you subscribe to anything.

Step 3: subscribe

curl -X POST $BASE/v2/trade/copytrade \
  -H "Authorization: Bearer $BRAVADO_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{
    "leader_address": "0xWALLET",
    "sizing": { "mode": "proportional", "value": "0.10" },
    "filters": { "categories": ["politics"] },
    "simulation": true
  }'

The filters block is where step 2 pays off. Without it you are copying the whole wallet, sports losses included.

The leader needs no Bravado account and is never notified. Any public wallet can be followed.

Sizing modes

Mode	Behaviour	Use when
proportional	A fixed fraction of the leader’s size	Their account is much larger than yours
fixed	The same stake on every mirrored fill	You want predictable per-trade exposure
capped	Proportional with a per-trade ceiling	Their sizing is erratic

proportional scales with the leader’s conviction, which is usually what you want if you trust the conviction. capped is the hedge when you like their picks but not their position sizing.

Step 4: paper trade it

curl $BASE/v2/trade/copytrade/simulation \
  -H "Authorization: Bearer $BRAVADO_API_KEY"

Simulation runs the whole subscription against live prices without touching the chain. Compare against the leader’s own PnL over the same period, and look for three things:

Divergence from the leader's own returns

Some divergence is expected, since you are filtering by category and sizing proportionally. Large divergence usually means the filter is doing more than you intended, not that the leader changed.

Fill counts higher than expected

At 25 bips per side, a leader who trades constantly can turn a profitable strategy into a break-even one on fees alone. Count fills across a week before committing capital.

Insufficient funds events

If your collateral cannot keep pace, fills get skipped, and skipped fills are not randomly distributed. Either raise collateral or lower the proportional fraction.

Step 5: go live

curl -X PATCH $BASE/v2/trade/copytrade/{id} \
  -H "Authorization: Bearer $BRAVADO_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{"simulation": false}'

Copytrade is charged at 25 bips per side, against 10 bips on the Trade API. That difference is why fill count matters more than it first appears: a leader trading twenty times a week costs materially more to follow than one trading twice.

Step 6: monitor

curl $BASE/v2/trade/copytrade/pnl     -H "Authorization: Bearer $BRAVADO_API_KEY"
curl $BASE/v2/trade/copytrade/history -H "Authorization: Bearer $BRAVADO_API_KEY"

Pause without losing configuration:

curl -X PATCH $BASE/v2/trade/copytrade/{id}/status \
  -H "Authorization: Bearer $BRAVADO_API_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{"status": "paused"}'

Prefer pausing to deleting while you work out whether a drawdown is noise or a broken thesis. PATCH /status keeps sizing and filters; DELETE discards them and you rebuild from scratch.

Wrapping up

The subscription is one call. Everything before it is the work, and it comes down to a single question the leaderboard cannot answer: is this record repeatable, and where?

Trade count tells you whether there is enough evidence. The equity curve tells you whether it accumulated or arrived all at once. Categories tell you where it came from. Filter to that, size it deliberately, and simulate until it is boring.

Frequently asked questions

Why not just copy the top wallet on the leaderboard?

Because a 30-day PnL ranking rewards one enormous win over many steady ones. The top wallet is frequently someone who got a single large call right, which is not a strategy you can inherit.

Does the leader know I am copying them?

No. They need no Bravado account and receive no notification.

Why filter by category rather than copying everything?

Aggregate PnL hides the composition. A trader can be strongly profitable in one category and steadily losing in another, and the total still looks good. Filtering means you inherit the edge without the leak.

How long should I simulate before going live?

Long enough to see a representative number of fills, which depends entirely on the leader’s cadence. The test is not elapsed time but whether the results have stopped surprising you.

What happens if I run out of collateral?

Fills that cannot be funded are skipped. Because skipped fills are not random, your returns will diverge from the leader’s in ways that are hard to reason about after the fact.

Can I follow several leaders at once?

Yes, each is a separate subscription with its own sizing and filters. Remember they draw on the same collateral pool, so two leaders active in the same market compete for it.

Resources

• Build a copy-trading bot, automating this selection process
• Copytrade API reference, sizing, filters, and risk controls
• Build a PnL leaderboard, the discovery surface
• Data API, the wallet analytics used for verification
• Simulation endpoints, paper trading reference
• Subscription endpoints, create, patch, and delete