Maker orders on entries and exits
Fill quality of maker-first orders on entries and on exits, from two days of Binance Demo Trading fills.
- Maker fills on exits
- 0%
- Exit slippage
- 9–22 bps
- Maker fee saving
- ≈ 2 bps
Findings
- Binance Demo fills landed within ±5 bps of the market price in the same minute.
- On entries, maker-first orders filled 74% as maker, at +0.7 bps versus the decision price.
- On exits, no order filled as maker. Trims and closes finished 9–22 bps worse than the decision price, against a fee saving of about 2 bps.
Method
On Binance futures a maker order, which rests in the book, pays about 2 bps less than a taker order, which crosses the spread. With turnover of about 45× equity a year the saving accumulates, so the bot was configured to post a maker order first, wait 45 seconds and send any unfilled remainder at market. Every fill on Binance Demo Trading from 27 to 29 September 2026 was then checked against Binance's maker/taker flag and against the market price in the same minute.
Results
- Fill prices. Demo fills landed within ±5 bps of the live market price in the same minute.
- Entries. One new position was opened in the window. It filled 74% as maker, at +0.7 bps from the decision price.
- Exits. No trim or close filled as maker, and all of them finished 9–22 bps worse than the decision price.
Implications
The strategy exits when price is moving against the position. A passive sell order then rests above a falling market and is not filled. After 45 seconds the remainder is sold at market, below the price at which the wait began, so the exit pays the taker fee anyway and adds the adverse move during the wait. Entries are less time-sensitive. The strategy buys in the direction of an established move, and a resting entry order is often filled by ordinary price noise.
The execution rule was changed. Entries and adds keep maker-first; trims, closes and stops are now sent to market directly. Each fill is still logged with its maker share and its slippage against the decision price, and both appear in the weekly report.
Limitations
The sample is two days and a small number of fills, with a single entry. Every exit in the sample finished worse than the decision price, but the size of the effect is not yet precisely estimated. A backtest that charges a flat cost per unit traded cannot detect an effect of this kind, which is one reason the strategy runs on a demo venue before real capital.