Why average spread beats typical spread marketing

[LondonScalper]

Every broker publishes a spread figure, and the word in front of it matters more than the number. "From 0.0", "typical", "minimum", "average": these mean different things, and only one of them is any use to me.

"From" or "minimum" is the best moment in the day. It might happen for a few seconds at 09:30 on a Wednesday. It's useless as a planning number.

"Typical" is vaguer than it sounds. Typical over what period, measured how? I've seen typical figures that looked as if they were taken during the most liquid hour of the most liquid day.

Average, if it's calculated across the whole trading day or properly by hour, at least includes the bad moments. But even a broker's published average can include hours I never trade.

What I care about is my own average, and specifically the average at the times I actually press buttons. My log of 2,300 EURUSD tickets this year shows a mean spread at entry of 0.17 pips during 07:00 to 11:00 London. Over my whole sample, including the occasional US data trade and a few early-morning ones, the mean rises to 0.26. The median barely changes, but the mean does, because the expensive tickets drag it up.

That gap between my median and my mean is where the real cost hides. Marketing tends to show you something closer to the median, at best. Your account pays the mean.

[Shadow Trader]

The median versus mean point is the one I wish someone had shown me years ago. I put my own tickets into a histogram to see it, and the attached chart is an illustration with synthetic numbers shaped like my log: median 0.16, mean 0.21, with about 5% of tickets out in the red tail.

img_13055.png (49.63 KiB) Viewed 25 times

That tail is almost entirely 13:30 data releases and a few early trades before 07:00. Take them out and my mean and median are nearly the same.

So the useful question for me isn't which broker has the best average. It's whether I should be taking those 5% of tickets at all. When I checked, they also had worse outcomes after costs, so I'm paying a higher spread for weaker trades.

One edge case worth flagging: some platforms log spread at the moment of the click, others at the fill. On a fast move those can be quite different, and a log based on click spread will flatter you.

Did you check whether your 0.26 trades were also your weaker setups, or are the US data trades genuinely worth the extra cost?

[Shadow Trader]

One more thought on this: spread is only half the entry cost, and the published figures never include the other half. Commission, converted to pips, is easy. Slippage relative to the price you saw is the one nobody publishes.

On my log, average slippage on market entries at 08:00 London is about 0.05 pips against me. At 13:30 on data days it's closer to 0.4. So the real cost gap between my good and bad hours is wider than the spread histogram shows.

I now log a single column called total entry cost: spread at fill plus half the commission plus slippage, all in pips. It changed which hours I trade more than any spread table did.

The other surprise was Fridays after 16:00 London. Spread there looked normal on average, but the slip was consistently worse, presumably because liquidity was thinning while the quotes still looked tight.

Do you track slippage separately from spread, or does your 0.17 already include it?