Slippage in Futures Trading: Where It Comes From and How to Cut It

A scalper backtests a system to +1.5 ticks of edge per trade, takes it live with identical entries, and loses money. The signals never changed. The fills did. On MES that works out to $1.875 of edge against $2.50 of round-turn slippage, a losing trade before the first bad signal, and it never shows up on a chart, because charts display traded prices, not the prices you received. Most traders have never measured their own slippage even once. What follows is where slippage comes from, how to measure it in ticks, and the arithmetic that decides whether an edge survives contact with the order book.

What slippage actually is

Slippage is the difference between the price you intended to trade at and the price your order actually filled at. The intended price can be the quote on screen when a signal fired, the level of a stop or limit, or the price you clicked; the actual price is whatever the exchange printed on your fill confirmation.

Measure it two ways. Per fill, tracking entry slippage and exit slippage separately, because they behave differently. And per round turn, entry plus exit combined, because the round-turn number is the one that actually hits P&L. A half tick on the way in and a tick and a half on the way out is two ticks off every trade you take, whether or not you ever notice it.

Quote it as a cost, so positive ticks mean a worse fill. Favorable slippage, where the market improves your price, does exist, but it is rare on marketable orders and no plan should assume it. A trader who models slippage as symmetric noise is modeling a market that does not exist.

The four sources of slippage

Every slipped fill traces back to one of four mechanisms, and on bad days they compound.

Crossing the spread. A market order buys the ask and sells the bid. Even with zero latency and infinite depth, a market-order round turn pays at least one spread; on a market that is one tick wide, that is one tick per round turn as a structural floor. No amount of infrastructure spend removes it.

Latency. Between the moment your signal fires (or your finger clicks) and the moment your order reaches the matching engine, the book keeps moving. The retail path from home PC to broker to exchange typically takes tens to hundreds of milliseconds depending on the setup, and the faster the market is moving, the more that gap costs you.

Thin depth. If your size exceeds what is resting at the best price, the remainder walks the book to worse levels, and your average fill price degrades with size relative to visible depth. If you have never watched your clip against the ladder, order flow and the DOM covers what the book is actually telling you.

Volatility spikes around scheduled news. Market makers pull quotes seconds before releases like CPI, FOMC and NFP. Spreads widen and depth vanishes just as price moves fastest, so slippage that is normally sub-tick can become many ticks inside that window. How many depends on the release and the product; treat any published figure as one scenario, not a constant.

How to measure your slippage

The protocol is almost embarrassingly simple, which may be why so few traders run it.

For every trade, log two prices: the intended price and the actual fill. Do it for a minimum of two weeks of live or sim-live trading. Then compute the average slippage per side, entries and exits separately, and the average per round turn, denominated in ticks. Ticks are comparable across days; dollars are not once you change size or product.

The split between entries and exits is the most useful sub-stat in the log. Expect exits, and stops in particular, to run worse than entries. The reason is structural and gets its own section below.

Retail traders obsess over entries. Professionals obsess over exits and fills. Logging intended-versus-filled is the cheapest edge audit in existence: it costs nothing, takes two weeks, and requires no new strategy. Many "failed" live systems were never broken; they were thin edges eaten by one or two ticks of unmeasured friction. And plenty of marketed backtests are only profitable because they assume fills at the price on the screen, which is a simulation privilege, not a market feature.

One caveat on the sample. Two weeks is a floor, not a guarantee, and a quiet fortnight can miss a high-volatility regime entirely, so slippage measured in calm conditions understates the bad-day case. Keep logging after you go live.

The edge math: when slippage eats the system

Now the arithmetic that decides whether a system is live-viable. MES, the Micro E-mini S&P 500, trades in 0.25-point ticks worth $1.25 each; that is CME contract spec and exact. The edge and slippage figures below are illustrative scenario numbers, but every arithmetic step is exact.

Take a scalper whose backtest shows +1.5 ticks of gross edge per trade: 1.5 x $1.25 = $1.875 per trade. The two-week slippage log comes back with entries averaging 0.75 ticks worse than intended and exits averaging 1.25 ticks worse. That is 0.75 + 1.25 = 2 ticks per round turn, or 2 x $1.25 = $2.50 per trade. Net edge: $1.875 - $2.50 = -$0.625 per trade.

The $500 monthly gap is 200 x $2.50, and it is entirely fills. Not one signal degraded. The system did not stop working; it never worked at the prices it actually traded. Commissions and exchange fees come on top, vary by broker, platform and membership, and only make the live column worse.

That table forces the decision rule: subtract measured round-turn slippage from backtested edge per trade and judge the system on the net number, never the gross. A strategy that is only profitable at intended prices is not a strategy. It is a spreadsheet.

How to cut slippage

Five levers reduce it, listed roughly in order of how available they are to a retail-sized trader.

1. Use limit orders where the strategy allows. A resting limit does not cross the spread and cannot slip. The cost is non-fill risk: you miss the trades that would have run without you.
2. Trade the most liquid contract and hours. Front month, highest volume, and for equity index futures the regular hours around the US session. Overnight and thin hours mean wider spreads and shallower books.
3. Keep size inside the visible depth at the best price. If the top of book routinely shows less than your clip, you are guaranteed to walk the book on every entry.
4. Skip roughly the two minutes around high-impact scheduled news. "Roughly" is deliberate; the danger window varies by release and product. If you trade funded accounts, your firm may force the issue anyway, so check prop firm news rules and blackout windows before the calendar does it for you.
5. Shorten the physical path. A server or VPS near your broker's gateway, a wired connection, and direct market access instead of chained platforms all shrink the latency component. The practical build is covered in the VPS setup guide for futures and copy trading.

Each lever has a cost. Limit orders are bad advice for momentum and breakout systems, because the trades that run hardest are exactly the ones your limit misses; for those strategies, some market-order slippage is simply the price of participation. Infrastructure spend attacks only the latency component and does nothing about spread cost or thin depth, so for a low-frequency swing trader it is mostly wasted money; slippage matters in proportion to trade frequency and edge thinness. And avoiding news windows means skipping some of the biggest moves of the month, a fair trade for most systems and a terrible one for a news-driven strategy.

Why stops slip worse than entries

Exit slippage running worse than entry slippage is not bad luck. It is structure.

A stop order converts into a market order at the precise moment price is trading through your level, which is also when one side of the book is being consumed and everyone holding a similar stop is trying to exit the same way. The book is thinnest right when your order needs it most, and you are competing with correlated flow for whatever liquidity remains. Entries are different in kind: you choose the moment for an entry, while the market chooses the moment for your stop.

Slippage is not random noise that averages out. Its worst fills cluster on your worst trades, so it fattens the left tail instead of shaving the mean.

This is why the measurement protocol insists on splitting entries from exits. A blended average hides the fact that the expensive fills concentrate on losing trades, which means slippage damages your worst outcomes far more than your typical ones. Risk models built on average slippage will be wrong precisely on the days they matter.

Slippage in copy trading

Copy trading adds a slippage layer with its own structure, and it matters because funded traders increasingly run one strategy across several accounts.

The follower always fills after the master. Whatever price delta exists between the leader's fill and the follower's fill is pure slippage: a direct, measurable, per-trade cost with no offsetting benefit. It also scales with the number of accounts and the total copied size, because more follower orders hitting the book after the master consume more depth, so later fills get progressively worse. That direction is structural; how big it is per account depends on product, size and conditions. The mechanics of running many accounts in sync are covered in multi-account copy trading: sync and slippage risk.

A fast server-side copier narrows the delta but never removes it. Thor's server-side copy runs at about 17 ms, and that number needs both halves stated: it is a measured figure under specific conditions, not a per-fill guarantee, and even at 17 ms the follower is still second in line. Speed buys a smaller delta, not a free one. A copier, however fast, is the right tool for scaling a robust edge across accounts; it is not a rescue for a marginal one, because a strategy that barely survives the master's own slippage will die on follower fills.

The rule that falls out: a copied strategy must be profitable on follower fills, not master fills. The backtest overstates the master, and the master overstates the followers.

The full checklist before declaring any system live-viable:

1. Log intended price versus actual fill for every trade for two weeks.
2. Average it per side and per round turn, in ticks, splitting entries from exits.
3. Subtract measured round-turn slippage, plus commissions, from backtested edge per trade.
4. If the net edge is positive with room to spare, go live. If it is negative or marginal, fix the fills first (limit orders, liquid hours, smaller size, no news window, shorter path) and re-measure before touching the signal logic.
5. If copying to funded accounts, repeat the measurement on follower fills. The master's numbers do not count.

The two-week log costs nothing and answers the only question that matters: not whether the system finds good trades, but whether it keeps its edge at the prices you actually get.

Frequently asked questions