The Cost Monster: Slippage, Spread, and the Death of Edges
- Oct 3, 2025
- 2 min read
Small frictions, repeated often, become destiny.
In a spreadsheet, every trade fills at the price you see. In real markets, there is a **line**—the bid/ask spread, the order queue, the trader ahead of you who grabbed the last fill. The cost of that line is small on a single trade but enormous in aggregate. A system that “makes 0.1% per trade” with no costs often makes **nothing** once you pay the toll.
There are two big costs. **Spread** is the gap between buying and selling—what you pay for instant execution. **Slippage** is the difference between your intended price and your actual fill, influenced by volatility, order size, and liquidity. Around **news**, spreads can widen dramatically and prices can jump; at **session opens**, the book reshuffles; in **thin hours**, a small order can move the price. If your edge depends on very tight targets or hyper-frequent entries, these costs are predators that eat first.
Design with the predator in mind. Prefer **fewer, higher-quality trades** over frantic churning. Enter when the microstructure is friendly: mid-session liquidity, calmer spreads, less jumpy order books—unless your edge *requires* the chaos and you have evidence it survives there. Use **order type policy** deliberately: market orders buy certainty but pay; limit orders save cost but risk missing the move; stop-limits try to control both, but can be skipped in fast markets. Don’t pay spread twice for the same idea—**batch** related signals.
Most importantly, **model costs honestly** in back-tests. Use variable spreads and slippage that grow with volatility. Inject order rejections and partial fills. Run **stress scenarios** with elevated costs and delayed entries. Publish **gross vs. net** results and slippage histograms so you can see the tax you’re truly paying. If the edge vanishes when the market charges rent, it wasn’t an edge; it was an **accounting error**. Fix the accounting before you trade.
Do you need help creating your own bot? Contact us.
Comments