Straddling news with pending orders: why both sides can fill badly

[Shadow Trader]

The classic news trick is to place a buy stop above the market and a sell stop below it a few minutes before the release. Whichever way price spikes, one order catches it.

On paper it's neat. In practice, several things interfere.

First, the spread widens right before and during the release. A buy stop triggers on the ask. If the spread blows out to 8 pips, the ask can touch the buy stop level while the bid, the price on the chart, has barely moved. The order fills without any real breakout.

Second, slippage. A stop order becomes a market order once triggered, and the fill comes at the next available price. In a fast news move, that can be 5, 10 or 20 pips from the stop level.

Third, whipsaw. Price can spike up, trigger the buy stop, reverse sharply and trigger the sell stop too. Both orders fill, both at bad prices, and both end up as losses.

Fourth, brokers may reject or delay orders near the release, or refuse to place pending orders within a few minutes of a scheduled event. Some have rules that restrict trading during news altogether.

None of this means the idea can't work. It means the expected outcome is very different from the clean diagram in a tutorial. Anyone testing it needs real fills from real news events, not a backtest on one-minute bars.

There are variations that handle some of the weaknesses. One is to wait a few seconds after the release and place a single order in the direction that has shown itself, accepting a worse entry in exchange for lower slippage and no double fills. Another is to use a very wide stop and a small size, treating the trade as a quick lottery ticket with a limited cost.

To evaluate any of them, the data needed is the fill price versus the intended price for each news trade, plus the spread at the moment of the fill. A log of twenty events usually reveals whether the average result is positive after costs.

A reasonable test is to run the approach with the minimum size for a month of events before considering anything more.

Has a news straddle ever worked well in live conditions, and what did the slippage look like?

[Amanhasnoname]

I ran a straddle on the US retail sales numbers for a couple of months with tiny size, mostly out of curiosity. The results were almost exactly as you describe. Out of maybe a dozen events, three produced a decent winner, five or six had one side fill with several pips of slippage and then go nowhere, and twice both sides triggered within about twenty seconds of each other. Those double fills wiped out everything the good trades had made.

The part that surprised me most was the spread trigger. On two occasions my buy stop filled before the number was even printed, purely because the ask jumped out while the bid stayed put. I could see it afterwards on the tick chart with both lines turned on.

What I do now, on the rare occasions I touch news at all, is your first variation. I wait five to ten seconds, look at which way the first real move went, and enter once with a market order and a wide stop. The entry is worse, but it's one trade instead of two.