nerivox METHODS
Open workspace ↗
THE WORK BEHIND THE WORKSPACE

Know what the
numbers mean.

A research desk for crypto and tokenized stocks. Each observation has a source. Each model has assumptions you can change.

01 / Markets, with their context.

Crypto prices and candles refer to Binance spot markets against USDT. Stock-token reference prices come from the LI.FI catalog for Robinhood Chain. These are separate markets; a crypto exchange quote is not a Robinhood execution quote.

DatasetSourceCoverage / update
Prices, closed candles, order bookBinance public market API ↗8 crypto markets · 25–30s cache · 499–500 closed bars
Token catalog / reference pricesLI.FI tokens ↗Chain 4663 · 60s cache · available tokens only
Funding / open interestOKX public data ↗USDT perpetual swaps · 40s cache
Fear & GreedAlternative.me ↗Published daily value · 15min cache
Contract / balances / transfersRobinhood Chain RPC ↗Read block shown · 30–40s cache

Equity labels describe tokenized-stock entries in the source catalog. They do not certify the issuer, redemption rights or contract safety. The token contract address identifies the asset. Stock-token prices have no invented 24h changes or historical candles.

02 / A closer look at the chain.

Token resolution

Chain Lens reads ERC-20 name, symbol, decimals and total supply directly from the selected contract. All balance calls in one result use the same block. Holder supply share is balance ÷ total supply × 100.

Observed holder balances

When a full holder index is unavailable, the desk scans recent ERC-20 transfers across 5,000 blocks. It checks balances for up to 80 participants and sorts the returned balances. This scoped result is not the token's complete holder ranking. Quiet holders outside the window are absent. Index-backed results, when configured, identify Blockscout as their holder source.

Wallet inventory

Wallet lookup reads native ETH plus up to 60 ERC-20 contracts from the available catalog, prioritizing the listed stock tokens. The result displays the number checked, failed reads and the read block. NFTs, unlisted tokens and assets outside the checked subset are excluded. An empty result does not establish that an address has no assets.

Transfer tape

Choose a token, a 1,000–10,000 block window and a minimum token amount. The tape reads Transfer logs, returns up to 1,000 events and identifies capped windows. Transfers alone do not establish a buy, sale, exchange deposit or liquidation. Contract links let you inspect the underlying transaction.

03 / Work through the numbers.

Trend & structure

EMA(20/50), Wilder RSI(14), ATR(14), MACD(12/26/9) and 20-bar Bollinger bands use complete candles. RSI uses exponentially smoothed positive and negative close changes. ATR uses the maximum of the intrabar range and each gap from the previous close.

EMAₜ = α · closeₜ + (1 − α) · EMAₜ₋₁    α = 2 / (n + 1)
RSI = 100 − 100 / (1 + average gain / average loss)
σ annual = std[ln(closeₜ / closeₜ₋₁)] · √(365 · 86400 / interval seconds)

Pivots require three candles on both sides. Levels group nearby pivots within 0.35 ATR. Breaks compare a close with the previous 20 candle highs/lows. The volume profile distributes each candle's volume evenly across the bins it overlaps; it is an approximation, not trade-by-trade volume at price. The regime label is a transparent heuristic based on ATR, EMA spread and directional efficiency.

Position sizing

Risk budget = capital × risk percentage
Quantity = min(risk budget / |entry − stop|, capital × leverage / entry)
R / R = |target − entry| / |entry − stop|

Capital, entry, stop, target and leverage are editable assumptions. The calculated loss excludes gaps, fees and liquidation rules. Default entries use an observed close and ATR-derived distances; they are not a personal holding or an order.

Monte Carlo paths

Sₜ₊Δ = Sₜ · exp[(μ − σ²/2)Δ + σ√Δ · Z]
Z ~ N(0, 1)    Δ = horizon / (365 × 64)

The seeded engine computes 2,000 paths by default, with 64 steps. Volatility starts from the selected historical candle window; drift defaults to zero. You can change volatility, drift, horizon and seed. Stop and target events are detected at the simulated steps. The first barrier reached sets the modeled exit; otherwise the terminal value is used. Intrastep crossings, gaps, slippage and fees are excluded. Displayed probabilities belong to this model and are not forecasts.

Kelly allocation

f* = p − (1 − p) / b
Selected allocation = max(0, f*) × chosen Kelly fraction

Win probability p and net win/loss payoff b are supplied by you. Negative estimated edge produces a zero allocation. Estimates can be unstable; changing an assumption changes the result.

Liquidation cascade study

The model allocates assumed exposure across 5×, 10×, 20× and 50× buckets with weights 30%, 30%, 25% and 15%. Each bucket activates when the modeled move reaches 100/leverage percent. Activated notional adds price impact according to your dollar-depth-per-1% assumption. Maintenance margin, hedging and actual positions are not observed; this is a stress calculation, not a live liquidation map.

Historical SMA strategy

The backtest uses crossover signals from a completed candle and executes at the following candle's open. Fee and slippage inputs apply on both entry and exit. It is long-only and uses the loaded candle window. Open positions at the end are marked to the final close. Closed-trade win rate excludes that open position.

Five perspectives

The analyst combines trend, momentum, volatility, structure and risk calculations. These are five calculation-based perspectives, not five independent AI agents. If a language model is configured, the optional model response is labeled separately and receives public market context only.

04 / Keep a condition in view.

Create a price, 24h change or RSI rule for a supported crypto asset. Armed rules are checked by the server approximately every 30 seconds, including while the tab is closed. RSI uses 1h candles and period 14. A trigger fires when its condition becomes true; it will not repeatedly fire until the condition first becomes false again.

Checks pause on unavailable or stale source data. The rule card displays its last successful check. Alerts stay inside your workspace and appear when you return. They are not browser push messages, orders or guaranteed real-time delivery.

05 / Keep the reasoning.

Research, watchlists, journal entries, missions and rules are stored on the server in a separate workspace. A random access key is kept in this browser's local storage. Anyone with that key can access the records; connecting a wallet does not authenticate or recover the workspace. Removing browser storage loses access unless you exported a backup.

Use Settings to export your records as JSON. Import reviews the file before adding records and pauses imported rules. The access key is not included in the export. Journal positions are entered by you; they are not imported trades. Portfolio combines those entries with current prices and can separately query a connected wallet's scoped balances.

Connecting a wallet shares its public address and allows read requests. Nerivox's research tools do not approve tokens, broadcast transactions or move funds. The $NRVX ticker identifies this project's brand; no project token contract has been published here.

Open Chain Lens ↗