Can we use gender to personalize trading features?

No. Trading features, limits and pricing must not depend on predicted gender. Personalization should be based on stated preferences and product usage instead.

How accurate are predictions from usernames?

Usernames often lack a real name, so unknown results are common. The response includes result_status, confidence and source so you can filter weak results out of aggregate reports.

What does forceToGenderize do?

It checks the dataset first and then lets AI read nicknames or handles with no real name. It costs 3 credits and cannot be combined with ai_mode off or always.

Gender prediction · Crypto exchanges

Aggregate audience analytics for crypto platforms

Crypto marketing teams often want to know whether education content, referral programs and community events reach beyond a narrow audience. Finowl gender predictions estimate how names or email local parts are typically used, which can produce an aggregate view of who signs up for a newsletter or attends an online explainer session. Those numbers help plan more inclusive outreach and content.

Predictions must stay out of every regulated and financial decision. Do not use them in identity verification, trading limits, fees, fraud scoring or account access, and never treat them as a person's identity. A prediction can be wrong for some people. Keep analysis aggregate, mention it in your privacy notice, and let users state or correct their own gender if you ask at all.

Try it in the sandbox About gender

USE CASES

Where it fits in crypto exchanges

Education content reach
Estimate the aggregate gender mix of subscribers to beginner guides or market newsletters, to see whether educational content is reaching new audiences or mostly existing traders.
Referral program analysis
Compare aggregate estimates of referred signups across campaigns to understand which referral channels broaden your audience, without attaching predictions to any individual account, trading profile or verification record.
Community event planning
Estimate the overall mix of registrants for online events or community meetups, helping your team choose speakers, topics and partner communities that welcome a wider range of people.
Username-based community stats
For forums or social features where members use handles, estimate aggregate composition from usernames, accepting that many results will be unknown and reporting them as such.

HOW TO INTEGRATE

The workflow

1
Export newsletter or event registrant names separately from KYC and trading data, using opaque ids.
2
Send batches of up to 50 items to POST /v1/gender/batch, or single handles to POST /v1/gender/username.
3
Keep identified results above a confidence threshold and count unknown results as their own group.
4
Produce aggregate charts per campaign and delete per-record predictions after reporting.

POST /v1/gender/username

curl https://api.finowl.co/v1/gender/username \
  -H "Authorization: Bearer $FINOWL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "username": "hodl_maria88",
  "options": {
    "ai_mode": "fallback"
  }
}'

Estimates gender from a community forum handle for aggregate community statistics. The result is never linked to KYC, limits, fees or any decision about the account holder.

GOOD PRACTICE

Before you ship

Never use predictions in KYC, fraud scoring, trading limits, fees or access decisions. They are statistical estimates of name usage and can be wrong for individuals.
Keep marketing analytics physically separate from regulated customer data, document gender inference in your privacy notice, and store only aggregate outputs.
Handles are noisy. Expect many unknown results from usernames, and avoid AI mode for bulk reporting unless the extra 3-credit cost is justified.

Read the API reference for every field and error code. Each lookup costs 1 credit (3 when it requests AI); see pricing.

A FEW GOOD QUESTIONS

Common questions