What “scalping” really means in crypto
Scalping targets many tiny price moves using frequent entries and exits on liquid markets. In practice, that means seconds-to-minutes holding times and a sharp focus on execution costs and slippage. While headlines often blur “scalping” with institutional high-frequency trading, the institutional edge typically comes from ultra-low-latency infrastructure and colocation that most individuals can’t access. For example, Deribit states its matching engine is hosted at Equinix LD4 and offers colocation/cross-connect for low-latency access, illustrating the structural speed gap between pros and most retail setups.
The structural limit for retail: speed and location
Exchanges and some providers advertise low-latency APIs, colocation and special feeds for high-throughput clients. Deribit’s LD4 colocation and multicast market data, and Binance’s low-latency API services for futures, highlight that meaningful latency advantages are available but explicitly targeted at institutional/VIP users. Individuals connecting over the public internet will generally be slower and should build strategies that don’t require microsecond reactions.
Cost math first: your breakeven edge per trade
Your strategy must clear fees, spread, slippage, and—if you trade perpetuals—funding. Maker/taker schedules from large venues show why taker-heavy scalping can be difficult:
- Binance lists spot fees around 0.1% base for makers and takers (discounts and promos can apply).
- Coinbase Advanced uses a published maker/taker schedule (taker fees generally higher at lower volumes).
- Kraken Pro runs a volume-tiered maker/taker schedule and even rebates on some pairs under incentive programs.
On top of fees, slippage and the bid-ask spread are real trading costs. Tight spreads and high liquidity reduce these costs; volatile or thin markets raise them.
Rule of thumb: if your typical scalp aims for +0.10% but you’re paying 0.10% in fees and giving up 0.02–0.05% to spread/slippage, the expected edge is likely negative. Use arrival-price slippage benchmarks to measure execution against when your signal triggered, not the eventual fill.
Funding rates matter on perpetuals
Perpetual futures add a periodic funding payment between longs and shorts to keep prices near spot. Funding mechanics differ by venue and can be positive or negative; it’s often computed from a premium vs. index price. Deribit and Binance Academy both explain the formulas and caps (e.g., Deribit details premium dampening and rate caps). Your scalp’s expectancy must include expected funding over your holding window.
CEX vs DEX: slippage vs MEV
On centralized exchanges (CEXs), you worry most about fees, spreads, and matching-engine microstructure. On decentralized exchanges (DEXs), you also face gas, mempool timing, and MEV, where sandwich attacks can worsen execution by inserting transactions before and after yours. Ethereum.org and Uniswap docs explain slippage and price impact, and recent academic/industry analyses document sandwich dynamics. For small, frequent trades, MEV risk and gas make DEX scalping especially challenging unless you use private order-flow routes.
Retail-friendly scalping methods
1) Maker-side spread capture on liquid pairs
Place post-only limit orders inside or at the best bid/ask on high-volume pairs. This seeks to collect spread without crossing it, sometimes with maker fee incentives or rebates on select markets. It lives or dies by queue position and cancel/replace discipline.
2) Liquidity windows and session overlaps
Trade only during peak liquidity (e.g., when US and EU hours overlap or when a venue is busiest) to minimize slippage and widen your buffer versus costs. Higher liquidity and tighter spreads improve odds for short-hold strategies.
3) Perpetual “funding-aware” micro-swings
Short holding windows around funding snapshots can distort flows. Be careful: funding can flip and caps/venue rules vary. Build the funding component into expectancy rather than treating it as “free yield.”
4) Eventless momentum bursts only if your edge survives fees
Short momentum bursts can look attractive, but taker-side entries must overcome the full cost stack. Many retail strategies underperform after realistic fee/slippage assumptions. Validate with rigorous transaction-cost analysis.
Tools and data flow you actually need
- Real-time market data via WebSocket: Maintain a local order book; both Kraken and Binance provide guidance and rate-limit rules, and Coinbase offers Advanced Trade WebSocket feeds.
- Order-entry API with post-only/TIF controls: Use post-only to avoid taker fees when you intend to make. Check your venue docs for supported order flags.
- Unified client libraries: CCXT provides a popular unified interface across many exchanges for prototyping/backtests, but always confirm exchange-specific rules.
- Mind the limits: WebSocket and connection limits are enforced; breaching them can disconnect or ban an IP. Plan subscriptions and throttling accordingly.
Position sizing and risk controls
Fixed-fractional position sizing and Kelly-based approaches are widely discussed; many practitioners run fractional Kelly to reduce drawdown risk. For short-hold, high-turn strategies, small sizing, hard daily loss limits, and strict stop placement are essential so one outlier fill or liquidation doesn’t ruin your day.
If you trade leverage, understand liquidation mechanics and the risk of auto-deleveraging (ADL) during extreme moves; venues document how insurance funds and ADL work.
Backtesting without fooling yourself
Many “profitable” scalps vanish once you add fees, spread, slippage, and latency. Statistical literature warns about backtest overfitting and selection bias; tools like the Deflated Sharpe Ratio and White’s Reality Check exist to reduce false positives from strategy mining. Use robust walk-forwards and stress tests.
Security basics for API trading
Treat API keys like cash. Use IP allowlists/whitelists where available, keep withdrawals disabled on trading keys, and limit permissions. Binance, Coinbase and Kraken document key creation and recommend IP allowlisting.
A practical 10-step workflow
- Choose a liquid pair with tight spreads and a fee structure you can beat. Review your venue’s fee page.
- Set up API/WebSocket data and maintain a local order book with throttling that respects rate limits.
- Define an edge that does not rely on beating colocation-level latency.
- Decide entry style: maker (post-only) for spread capture or taker for momentum.
- Encode risk: fixed-fractional (e.g., 0.25–0.5% of equity per trade) or fractional-Kelly; cap daily loss.
- Paper trade or sim with full costs: fees, spread, slippage, and funding if using perps.
- Live test small; log every trade and compute arrival-price slippage and effective spread paid/earned.
- Iterate on cancel/replace logic; monitor queue position and fill ratios.
- For DEXs, consider private routing to lower MEV exposure; otherwise expect worse execution.
- Review weekly with backtest-overfitting checks before scaling.
FAQs
Is scalping legal?
On regulated and permitted venues, scalping is generally allowed, but you must comply with exchange rules and local laws. Always read your venue’s rulebook and terms.
How much capital do I need?
Enough to absorb fees, spreads, and a statistically meaningful sample of trades. Low volume tiers often pay higher taker fees, which can erase small edges; many find maker-side approaches more viable at small scale.
What’s the best timeframe?
The “best” is the one where your measured edge still clears costs. For many individuals, strategies that operate on seconds-to-minutes but don’t require microsecond speed are more realistic given infrastructure constraints.
