Research Data Addendum

Last updated: 15 July 2026 · Effective date: 29 July 2026

This is the same document bundled in the Nexus Dex app. If you have the app, you can read it offline under Settings → Legal.

This Research Data Addendum (the "Addendum") explains how Sam Ford, trading as GrowthEngine (ABN 18 266 807 610) ("Nexus Dex", "we", "us"), uses opt-in data you contribute to our research lake, and the limits we place on ourselves in doing so.

This Addendum is part of, and incorporated into, the Terms of Service and Privacy Policy. If anything in this Addendum conflicts with the Privacy Policy on a research-related question, this Addendum prevails.

We wrote this so it can be read in one sitting. The short version is on page one.

1. The deal in plain English

Research mode is opt-in. It is off by default for every account. Nothing goes to the research lake unless you turn it on yourself, and you can turn it back off in two taps — once off, it stays off unless you turn it on again.
Default OFF and unchangeable for under-13 accounts.
We use the data to test hypotheses about the game, e.g. shiny rates around events, trade reroll bias, regional spawn patterns. We publish what we find.
We never sell your raw data. We never sell your personal information. We never license raw rows to anyone.
We may publish or share aggregate findings (e.g. "the shiny rate during Community Day 2026 was 1.3% across 412k catches"). Aggregates are constructed so an individual cannot be re-identified.
You can opt out, delete, or export your research contributions any time.
Raw GPS coordinates never leave your device. We bucket geo data to a ~80m hex (S2 cell level 17) on-device.
The lake is append-only while you're opted in. Hiding a Pokémon in the app does not remove it from past research records, but turning research mode off stops new ones from being added, and deleting your account cascades the lake.

If any of this is uncomfortable, leave research mode off. The Service works the same way without it.

2. What is the "research lake"?

The research lake is a separate, append-only table in our database (research_events) that captures the detailed history of each Pokémon record you upload, beyond what the app's user view needs. The user view (pokemon_records) is what the app shows you and is fully under your control. The lake exists because:

the app only needs "you have a 100% Hundo Salamence",
but research questions need "this Hundo was caught on this date, in this ~80m geo cell, with these weather conditions, before/after a trade reroll, etc."

The lake is the second of two tables we write to in the same transaction at upload time. If research mode is off, we skip the lake write and only write the user view.

3. What goes in the lake (and what does not)

3.1 Goes in (when research mode is on)

For each Pokémon record:

Identity: species, form, gender, shiny / lucky / shadow / purified flags, size tag.
Stats: IVs (atk, def, sta), CP, HP, level estimate.
Pre/post-trade snapshot: if a record is from a trade reroll, the pre-trade IVs/CP/HP plus the post-trade ones.
Catch context: event kind (wild catch, raid, hatch, research, lure, trade, etc.), catch date, weather boost flag and type, any event tag (e.g. "community_day_2026_03").
Geo bucket: an S2-cell-level-17 token (~80m). No raw lat/long.
Free-form preservation: a JSON blob with anything else from the source CSV we don't yet know what to do with, used so we don't have to ask you to upload again later.

3.2 Does NOT go in, ever

Your name, email, account ID, or anything that personally identifies you (the lake row is keyed by an internal account_id integer, which we sever when computing aggregates).
Your raw GPS coordinates.
Your home address, work address, school, or any address.
Your IP address.
Your trainer code.
Other people's data.
Special-category data under GDPR or the Privacy Act (health, biometrics, religion, sexual orientation, etc.).

4. Why we run a research lake

We want to answer questions about the game that no individual trainer can answer alone:

Are shiny rates higher under certain conditions, lures, incense, weather, geography, time-of-day?
Is there a hidden spawn-sequence trigger that can predict the next shiny?
Does trade reroll bias against high-IV Pokémon and toward low-IV ones?
How do 100% IV chances differ across catch methods (wild, hatch, research, raid)?
What is the actual distribution of XXL/XXS catches, and does it move with events?

We publish findings publicly, in aggregate, so the trainer community benefits whether they contributed to the data or not. This is what we mean by "by trainers, for trainers".

5. Who can contribute and consent

5.1 Adults (18+)

You can opt in or out at any time. Default is OFF — you have to turn it on yourself in Profile, and it is two taps to turn back off.

5.2 13-17 year olds

You can opt in or out at any time. Default is OFF — you have to turn it on yourself in Profile. We strongly recommend reviewing the choice with a parent or guardian. Some jurisdictions require parental consent for users under 16 (e.g. EU GDPR-K), where we detect this we apply the local rule.

5.3 Under 13

Research mode is forced off and cannot be turned on, regardless of where the account is created or who creates it. This applies whether or not the local age of digital consent is lower than 13.

Even if a parent gives consent and uses an under-13 account on the child's behalf, the lake stays off for that account. The user view still works as normal.

6. Your control, opt-out, deletion, export

You have four primary controls. All are free and immediate.

6.1 Turn research mode off

Profile → "Research mode" → toggle off. From that moment:

No new rows are added to the lake from your account.
No geo bucket is computed or sent on subsequent uploads.
The change is logged with a timestamp on your account.

You can turn it back on later. We don't ask why.

6.2 Delete your past research contributions, "research erase"

Profile → "Research mode" → "Erase my research contributions". This deletes every research_events row associated with your account within 30 days (often immediately). It does not affect your pokemon_records user view.

After research erase, you remain opted out unless you opt back in.

6.3 Delete your account

Profile → "Delete account". Cascades both pokemon_records and research_events. Within 30 days no record of your account or your contributions remains, except for legal-retention items listed in section 6 of the Privacy Policy.

6.4 Export

Profile → "Export my data". You receive a JSON file with:

your pokemon_records (user view),
your research_events rows (the lake contributions),
your priorities map and account info.

Format is machine-readable and human-readable.

7. What we promise NOT to do

7.1 Never sell raw rows

We will not sell, license, transfer, or otherwise commercialise raw row-level data to any third party. This includes:

gaming companies, including Niantic, Nintendo, The Pokémon Company,
analytics, advertising, or data-broker companies,
AI/ML training data buyers,
governments or law enforcement, except where compelled by a valid legal request as described in the Privacy Policy.

7.2 Never re-identify

We will not attempt to re-identify any individual user from aggregate research outputs. Internal staff with access to the lake operate under written confidentiality and security obligations and cannot use the data to target, profile, or contact individual users.

7.3 Never use research data for ads

We do not run targeted advertising. Research lake data will never be used to build advertising profiles, lookalike audiences, or behavioural-marketing segments.

7.4 Never weaken the geo bucket

The on-device geo bucketing to ~80m is a privacy floor. We will not lower the bucket precision (e.g. to raw lat/long, or even ~10m) without:

written consent from each affected user, captured separately and after the fact,
an updated version of this Addendum,
at least 30 days' advance notice to affected users.

7.5 Never silently change scope

If we want to add a new category of data to the lake, or a new use case beyond hypothesis testing, we will:

update this Addendum and ship the new version with a "Last updated" date,
notify affected users at least 14 days before the change takes effect,
give you a chance to opt out before the change applies to your contributions.

8. What we MAY do

8.1 Publish aggregate findings

We may publish and share aggregate analyses, charts, and conclusions (e.g. blog posts, presentations, papers, in-app insights) where the underlying data has been aggregated such that no individual user can be re-identified. Standard de-identification techniques apply, including k-anonymity thresholds (minimum group size before publication) and suppression of small cells.

8.2 Use aggregates internally

We may use aggregate findings to improve the Service, for example to tune custom-dex defaults, calibrate the storage optimiser, or surface "best time to catch X" hints in the app.

8.3 Share aggregates with researchers

We may share aggregates (NOT raw rows) with academic researchers or community partners under a written agreement that prohibits re-identification and onward sharing. We list any such partnership at https://nexusdex.ai/legal/research-partners.

8.4 Retain de-identified aggregates after you opt out

Once an aggregate has been computed (e.g. "shiny rate during Community Day 2026 was 1.3%"), the aggregate itself does not contain identifiable data. If you later opt out or delete your account, we delete your raw rows from the lake but the aggregates remain. You agree to this on contribution.

9. How aggregation actually works (for the curious)

We compute aggregates from the lake using these rules:

Minimum group size: an aggregate is only published when it covers at least 20 distinct contributing accounts. Cells smaller than that are suppressed.
Geo aggregation: the lake stores an S2-L17 token (~80m). Public aggregates use at minimum S2-L13 (~5 km), often much coarser.
Time aggregation: typically by event window (e.g. a Community Day) or month, never finer than 24 hours in publication.
No joins to identifiable tables: aggregation jobs run in a sandboxed environment that has access to research_events only, not to accounts or sessions.

If we ever need to lower these thresholds for a specific study, we will (a) get external review, (b) notify users, and (c) give you a chance to opt out before the study runs.

10. Security of the lake

The lake table lives in the same encrypted-at-rest production database as the user view, with stricter access control.
Engineering staff access is least-privilege, audited, and subject to written confidentiality.
Backups are encrypted and retained per the Privacy Policy retention table. Backups containing your data are also affected by deletion within the same 30-day window.
We run regular security testing and apply the anti-tampering controls described in section 4 of the Terms of Service.

11. Conflicts and survival

If a court or regulator finds any provision of this Addendum unenforceable, the remaining provisions stay in force. If your local law gives you rights stronger than those in this Addendum, those stronger rights apply automatically.

Sections 4, 5, 6, 7, 8.4, and 9 survive termination of your account, except that any actual deletion obligations under section 6 will be carried out as described.

12. Contact

Research-data questions and opt-out support: privacy@nexusdex.ai
Security disclosures: security@nexusdex.ai
Legal notices: legal@nexusdex.ai

By trainers, for trainers. The lake exists to answer questions we all wonder about. Your contribution is voluntary, your control is real, and your raw data is yours.