Read a Slippage Log

[Fairman]

Read a Slippage Log

Ten stop-loss fills, with slippage in pips measured against the stop price, where negative means a worse fill: minus 0.5, minus 1.2, plus 0.3, minus 0.8, minus 2.5, minus 0.4, plus 0.1, minus 1.0, minus 0.6, minus 3.0.

Sum: minus 0.5 minus 1.2 is minus 1.7. Plus 0.3 is minus 1.4. Minus 0.8 is minus 2.2. Minus 2.5 is minus 4.7. Minus 0.4 is minus 5.1. Plus 0.1 is minus 5.0. Minus 1.0 is minus 6.0. Minus 0.6 is minus 6.6. Minus 3.0 is minus 9.6.

Average: minus 0.96 pips per stop. Favorable fills: two of ten. Worst two: minus 2.5 and minus 3.0.

Is this acceptable? Some adverse slippage on stops is normal, since stops become market orders. Two favorable fills out of ten suggests the process is not entirely one-sided, which is a healthy sign. If none of the fills had been favorable, it would raise concern.

Look at the worst cases. If those two occurred around news releases, they reflect market conditions. If they happened in quiet markets, they point to execution quality.

Cost: at 0.50 lots, a pip is worth 5 dollars, so average slippage costs 4.80 dollars per stop. On a stop that risks 100 dollars, that is nearly 5 percent extra loss per losing trade.

Include slippage in your expectancy: if you have 20 losing trades a month, the extra loss is about 96 dollars.

Exercise: log slippage on your next twenty stop exits, and flag which occurred near news.

[FXS HongKong Scalper]

Good exercise. The negative skew is typical: two of ten favourable, and the bad ones are larger. Losing close to a pip on each stop matters a lot when the stop itself is only 5 or 6 pips.

It would be useful to add the time of each fill. Often the worst slippage clusters around news or rollover, and avoiding those windows reduces it a lot.

[Fairman]

The running sum is correct, and the average of minus 0.96 pips per stop is a useful number to have. It means each losing trade costs about 1 pip more than planned. On a 25 pip stop, that's about 0.04R per loss. At a 50 percent loss rate, it's about 0.02R per trade overall. That might sound small, but on a strategy with a net edge of 0.10R, it's a fifth of the profit. Including it in backtests, as discussed in the cost stress thread, gives a more realistic estimate of what the strategy will actually produce live.

[Amanhasnoname]

I kept a slippage log on a live account for about two months and it showed me something the average hid. My typical stop fill was about 0.6 pips worse, but nearly all of my large misses came within a couple of minutes of a scheduled release or right at the daily rollover, when the spread was also several times wider. Quiet London hours were almost clean.

The practical changes were small. I stopped leaving stops on trades through the 21:00 to 23:00 server rollover window on my broker, and I closed or hedged positions ahead of high impact news instead of trusting the stop. After that the average fell to around 0.3 pips, and the worst fill in the next month was only 1.4 pips.

I also noted the order type and the platform used, because stops placed from the mobile app seemed to fill slightly later in my sample, though I only had thirty trades there and would not claim more than that.

[Fairman]

The point about favourable fills is a good check of execution fairness. If slippage were purely random, with orders filled at the best available price, some fills would be better than the stop price and some worse, depending on how price moved in the milliseconds around the trigger. Two favourable fills out of ten suggests symmetric execution is at least possible, though the average is still negative, as expected for stops that trigger in moving markets. A log with zero favourable fills over many trades would suggest asymmetric slippage, which some regulators have taken action against brokers for.

[Fairman]

The worst two fills, at minus 2.5 and minus 3.0, account for more than half of the total slippage. That's typical: most fills are close to the stop price, and a few outliers drive the average. If those two occurred around news, avoiding trading near releases would cut the average sharply. If they occurred in quiet markets, it's worth raising with the broker. Recording the time of each fill alongside the slippage makes this analysis possible. Many platforms show the requested and executed price in the trade history, which gives the slippage directly.

[Fairman]

The monthly cost of about 96 dollars for 20 losing trades is a concrete figure. Over a year, that's over 1,100 dollars, which could be a significant share of a small account's profit. Comparing the slippage across brokers, or across account types at the same broker, would show whether the cost is unavoidable or could be reduced. Has anyone here logged slippage on their stops over a few months? I'd be interested in what average people found and whether it differed noticeably between news periods and normal trading hours.