Skip to main content
All posts
· 7 min read·Smartbull research

How we model and budget for slippage

Real fill prices are not mid-prices. We measure your actual slippage, recalibrate weekly, and refuse to ship strategies that don't survive 10 bps.

The single biggest curve-fit trap

Most published crypto strategies look great on paper because the backtest fills at the mid-price or the open. Live markets fill you at the opposing book — wider in low-liquidity hours, wider on volatile candles, wider on the small-cap symbols where the visible edge is biggest.

That's why every Phase 37 sleeve must survive the 0/5/10/20 bps stress test before it ships. If a strategy needs zero-slippage to be profitable, we drop it.

How we measure your real slippage

Every filled order writes <code>slippage_bps</code> into <code>bot_orders</code>: the signed difference between the fill price and the mark price at the moment of decision. We aggregate 30 days of those fills into a notional-weighted mean, cut by exchange / hour / size bucket / regime.

The aggregator runs nightly and feeds the maker-first execution layer (Phase 92). On low-fill-rate buckets, we step the limit price 0.25 ticks inside the book and adapt the wait time. On high-fill-rate buckets, we save bps on every trade.

What you can do

Two levers move your real slippage more than anything else:

  • Pick a Tier-1 exchange with deep books in the pairs you trade. The same strategy can cost 3 bps on Binance and 18 bps on a long-tail venue.
  • Don't over-allocate the small caps. Sizing 5% of your account into a $10M-liquidity altcoin will eat the edge alive. The orchestrator caps single-position size for this reason.