How Trench Trials works.

Trench Trials is decision-bias training for crypto traders. It measures how revealing a token’s identity changes a trader’s decision.

Each trial shows three anonymous assets with pre-cutoff Nansen signals. You make a blind choice, see the identities, and can stick or switch before the future is revealed. The result measures what recognition gained or cost. Prediction games ask whether you were right; Trench Trials asks what the ticker changed.

Crypto traders have more data than ever, but familiar tickers, reputation and narratives still influence their decisions. Ordinary prediction games and backtests record only the final choice, so they cannot tell whether the token’s identity changed the decision.

Trench Trials records the choice before and after the reveal, on the same data, and measures the difference against the real market outcome.

Individual crypto traders and analysts who want to test whether they follow market signals or familiar narratives, and to measure and reduce that narrative bias through repeated blinded decisions.

Trading communities can use the shared Daily Trial as a common decision-bias calibration exercise: everyone plays the same round on the same UTC day.

• Points. A final pick that is a winner earns 100 points; any other final pick earns 0. Returns within 0.01 percentage points of the best share the win, and each earns 100.
• Switch impact. Final return minus blind return, in percentage points. It is 0.0 when you keep your pick.
• Accuracy. Blind accuracy counts how often your blind pick was a winner; final accuracy counts your final pick. Both use the first time you play each round.
• What never scores. A repeat of a round you already played, an abandoned Blind Pick, a Live prediction until its round resolves, and any invalid round.

Ticker Tax is what changing your mind after the reveal cost or earned you: your blind pick’s return minus your final pick’s return, in percentage points. Keeping your pick contributes exactly zero. A positive average means the names pulled you away from better reads; a negative one means they helped.

Timeouts are excluded because they are not decisions. Your Ticker Tax reads “insufficient evidence” until you have made five explicit keep-or-switch decisions on rounds you are playing for the first time.

Replay rounds are built from a past cutoff whose outcome is already known. You are never served a round you have already played. When you have played every available round, you see how many you completed and when the next Daily arrives.

The Daily is one round assigned to a UTC calendar date, the same for every player that day. You get one measured attempt: once you have played it, the Daily page shows your result and when the next one opens. A round used as today’s or an upcoming Daily is not served as a Replay until its day has passed. Sharing your Daily result reveals what kind of decision you made, never the answer.

A Live trial is measured in real time. The candidates and their signals come from Nansen’s current token screener at the moment the round opens (trailing 24 hours and 7 days), and every candidate must qualify under the eligibility policy. Entry stays open for five minutes, and the outcome is measured over the next 24 hours between two fixed five-minute candle boundaries. All Live times are shown in UTC.

After you lock, your prediction is pending until the window ends and the closing candles are available from Nansen. Live appears in the menu only while a Live trial is open for entry or has a verified result.

You play as a guest — no wallet, no signup. Your first locked pick sets a signed, HttpOnly cookie on this browser that links it to your decisions on our server. History lists every decided trial with your blind accuracy, final accuracy, switches helped and hurt, average switch impact and Ticker Tax.

Only decided trials count. An abandoned Blind Pick is never counted, and replays of a round you already played are listed but excluded from the metrics.

Your trial history is saved to this browser as an anonymous profile. Clearing site data or using another device starts a new anonymous profile; there is no account, so it cannot follow you across devices.

• What is stored. A random guest ID in a signed, HttpOnly cookie (one year), and on our server the decisions made under it: picks, the times they were locked, and optional recognition answers. No name, email, wallet or payment details are ever requested.
• In your browser. The cookie above, and a note that you have seen the onboarding sheet. There are no advertising or analytics trackers.
• Limits. History belongs to this browser: another browser or device, or clearing cookies, starts a new one, and it cannot be moved or recovered. There is no self-service way yet to delete the decisions stored for a guest ID.

Nansen supplies the candidate universe and the point-in-time market data used to construct every blind trial. Trench Trials deterministically selects eligible assets from that data, derives the blind clues, and uses Nansen price candles to calculate the outcome. Without Nansen, no new Replay, Daily or Live round can be produced.

Nansen screener → eligibility → blind clues → commitment → player choices → Nansen OHLCV → verdict → Ticker Tax

If the Nansen data a round needs cannot be retrieved or checked, the round is not created. No other source is substituted and nothing is estimated. See the evidence

Historical inputs are frozen and hash-sealed before a player enters. This prevents the evidence from changing after a decision has been made and makes every verdict reproducible. That is why playing a round does not call Nansen again.

Existing rounds can survive a temporary Nansen outage, but the product cannot create new content without Nansen: candidate discovery, blind signals, outcomes, Replay generation, new Daily rounds and Live capture and resolution all require it. Frozen data is the real Nansen data the round was built from, never a mock or substitute.

Each round keeps a receipt for every Nansen response it was built from: the endpoint, the exact request, the SHA-256 of the response, Nansen’s request ID, the credits it cost and when it arrived. Every request attempt — including failures and retries — is also written to a call log before the next one is sent.

The verification panel under each verdict lists those receipts and lets you recompute the round’s commitment in your browser. Raw responses are kept privately by the operator as evidence and are not republished, in line with Nansen’s redistribution terms.

A round is verified only when four separate checks hold. Each is a different claim:

• Provenance. Every input traces to a logged Nansen response, by SHA-256 and request ID, and is sealed in the round’s commitment before play.
• Calculation. Each return is exit close ÷ entry close − 1 on the committed candles, and anyone can recompute it from the revealed manifest.
• Eligibility. The round was reviewed and approved under the current eligibility policy: freely traded tokens (no stablecoins, wrapped or bridged majors, staking or tokenized assets), liquidity and trading-volume floors, complete price candles, and no near-duplicate of another round. Tokens the policy cannot positively classify are excluded.
• Availability. Only approved rounds are offered for Replay, the Daily or Live. A round can later be withdrawn: it leaves play and scoring, stays listed in the History of anyone who played it, and nothing is deleted.

The server decides what each stage may contain. During Blind Pick, responses carry only the slot letters and signal buckets — no names, addresses, exact dates, prices or returns. Names are added only after your blind pick is locked, and prices and returns only after your final choice is locked. Your browser never receives information early and hides it; it simply does not receive it.

A round is invalid when the data it committed to cannot be verified — for example, when a closing price candle is missing at the exact measurement boundary. An invalid round is never scored and no winner is manufactured. It stays visible in the audit record so the failure is transparent.

Before a round opens, everything that decides it — the candidates, their signals and the outcome rules — is written into a manifest and sealed with a SHA-256 hash together with a secret random value. The hash is published; the manifest stays private.

At the verdict, the manifest and the random value are revealed, so anyone can recompute the hash and confirm the round was not changed after players started choosing.

Trench Trials measures decision-making on past and time-boxed market data. Nothing here is a recommendation to buy or sell anything. Token prices are volatile and past returns say nothing about future returns.