Crypto markets move fast—and they don’t close. Prices, funding rates, and order books shift second by second. A dataset that’s missing candles, misaligned timestamps, or polluted by token redenominations can generate signals that look great in a spreadsheet but fail live. Clean data aligns your strategy with market microstructure, protects you from false edges, and keeps your P&L honest in both CAD and USD.

Market structure quirks you must account for

• 24/7 trading: No weekend gaps, but plenty of exchange maintenance windows that can create data holes.
• Token changes: Rebrands, redenominations, and contract migrations can distort historical prices and volumes if not adjusted.
• Multiple quote currencies: BTC/USDT, BTC/USD, BTC/CAD—mixing them without normalization leads to bad comparisons.
• Exchange fragmentation: Liquidity and fee schedules differ; backtests should reflect your actual venue(s), not a hypothetical composite.

Real costs of dirty data

• False alpha: A single incorrect wick or missing candle can flip a breakout into a fake winner.
• Poor risk sizing: Volatility estimates computed on bad bars produce undersized stops and oversized losses.
• Compliance headaches: Sloppy records complicate CRA reporting and reconciliation across Canadian crypto exchanges.

Start with a minimal, dependable dataset that reflects how you actually trade. Add granularity only when you can maintain it.

1) Price and volume (OHLCV)

• Granularity: 1m/5m for day trading, 1h/4h/daily for swing trading. Keep raw candles and your resampled versions.
• Time standard: Store timestamps in UTC, and only convert to local time in your UI or reports.
• Venue specificity: BTC/USDT on one exchange is not the same as BTC/USDT on another. Keep venue identifiers.

2) Trades and order book snapshots

• Tick data: Useful for order flow tools like CVD (Cumulative Volume Delta) and accurate slippage modeling.
• Depth data: Snapshots at multiple price levels to estimate impact and to test liquidity‑aware entries.

3) Derivatives metadata

• Funding rates & open interest: Essential for perp strategies and monitoring crowded positioning.
• Contract specs: Multipliers, tick sizes, risk limits, and maintenance margin schedules for each venue.

4) On‑chain reference signals

• Supply changes, active addresses, gas fees, and stablecoin flows can be valuable regime filters and risk triggers.
• Be consistent with how you align on‑chain timestamps with exchange candles.

5) Token reference data (“corporate actions” for crypto)

• Rebrands and tickers (e.g., XYZ → NEW), redenominations (1:1000), contract migrations, and chain merges/forks.
• Track airdrops and distributions as cashflow events; treat as P&L, not price distortion.

6) FX rates and CAD normalization

• Maintain daily CAD/USD rates and, if relevant, CAD/USDT proxies for translating P&L to your base currency.
• Record the conversion rate used at each trade’s timestamp when fills are in non‑CAD quotes.

Think in stages: ingestion → validation → cleaning → normalization → storage → monitoring. Keep it simple and repeatable so your backtests can be reproduced.

// Example pseudo‑rules for a 1m candle validator
assert(PriceOpen > 0 && PriceHigh > 0 && PriceLow > 0 && PriceClose > 0)
assert(PriceHigh >= max(PriceOpen, PriceClose))
assert(PriceLow  <= min(PriceOpen, PriceClose))
assert(VolumeBase >= 0 && VolumeQuote >= 0)
assert(timestamp == floorToMinute(timestamp))
// Flag outliers for review (don’t delete blindly)
if (abs(return) > 8 * rollingStd(returns, 1440)) flag("outlier")

Record‑keeping the CRA will appreciate

• Keep complete trade logs: Date/time (UTC), asset, side, quantity, price, fees, and the CAD equivalent at the time of the trade.
• Adjusted Cost Base (ACB): Track cost basis per asset in CAD for accurate gains/losses when you dispose or swap.
• Staking, airdrops, and funding: Treat as separate income or cashflows in your ledger rather than price distortions.
• Documentation: Retain exchange statements, wallet records, and internal reports. Organize by tax year in immutable folders.

Platform selection and compliance

Choose Canadian crypto exchanges and platforms that follow domestic rules, including registration with Canadian securities regulators and AML compliance as reporting entities with FINTRAC. Beyond security and fees, prioritize clear statements, downloadable trade histories, and stable API/CSV exports—your data hygiene depends on it.

CAD exposure and FX conversion

• Unified base currency: Even if you trade BTC/USDT or ETH/USD, convert fills to CAD for portfolio and tax views.
• Fee transparency: Track spreads and conversion fees when moving between CAD and USD books.
• Time alignment: Store the FX rate used at the trade timestamp, not an end‑of‑day average.

If you also trade Canadian crypto ETFs

• Maintain separate datasets for ETF OHLCV, distributions, and splits; don’t mix with spot coin data.
• Model tracking differences between ETF market price and underlying reference price if you pair trade.

Model what you actually pay and what you can actually fill

• Fees and funding: Apply venue‑specific maker/taker fees and perp funding cashflows to every trade in your backtest.
• Slippage and impact: Use bid/ask quotes or depth snapshots to simulate realistic fills, not mid‑price fantasy.
• Latency: For bots, bake in order latency and cancellation rules to avoid impossible executions.

Avoid common biases

• Survivorship bias: Keep delisted coins in the universe with their true, sometimes ugly, histories.
• Look‑ahead: Build indicators only from information available at the time; no peeking at later candles.
• Data snooping: Reserve out‑of‑sample periods and perform walk‑forward validation.

Stress and regime testing

• Replay high‑volatility windows and liquidity droughts to test stop‑loss behavior and position sizing.
• Model stablecoin de‑pegs or exchange outages using scenario overrides in your dataset.
• Monitor tail risk with drawdown distributions, Expected Shortfall, and exposure “heat” at the portfolio level.

for bar in candles:
  signals = model.update(bar)
  target   = riskManager.size(signals, volatility=ATR, base="CAD")
  orders   = executionEngine.route(target, bookDepth, fees, latency)
  fills    = broker.simulate(orders, bidAsk, venueRules)
  ledger.update(fills, fundingRates, fxRatesCAD)
  metrics.record(equityCurve(), drawdown(), turnover(), feeLoad())

Pre‑trade checks

• Price sanity checks (no stale quotes), notional limits, and max leverage per instrument.
• Validate that stop‑loss and take‑profit orders obey tick size and minimum notional rules for your venue.
• Confirm CAD P&L impact before sending orders to avoid unplanned currency exposure.

Post‑trade reconciliation

• Match fills against expected prices within tolerance bands; investigate slippage spikes.
• Reconcile positions, fees, funding, and FX cashflows daily; store immutable end‑of‑day snapshots.
• Tag anomalies with incident notes (e.g., venue maintenance, API throttling) in your dataset.

Operational notes for mobile and API trading

• Use read‑only API keys for dashboards; restrict trade permissions to specific bots and symbols.
• Require 2FA for manual overrides; log every manual intervention in your journal.
• Separate production and research environments so backtests don’t accidentally hit live accounts.

• Include all fees in your ledger: Trading, withdrawal, conversion, and network fees add up; ignoring them inflates returns.
• Tag market regimes: Volatility states, funding regimes, and liquidity tiers help explain when a strategy works.
• Capture “not traded” decisions: Logging when and why you skipped signals reduces hindsight bias.
• CAD‑centric risk: Set position limits and max loss in CAD even if the instrument trades in USD or USDT.
• Immutable archives: Keep read‑only monthly snapshots for audit trails and peace of mind.