Same price on screen, different fill at two brokers

[Fairman]

Same price on screen, different fill at two brokers

Put the same strategy on two brokers and the equity curves will differ, even with identical entries. People assume it is the spread. It is rarely the spread alone.

Here is a scenario. A buy signal fires at 10:15:02. Broker A shows 1.0841 and fills at 1.0841 in 90 milliseconds. Broker B shows 1.0841, takes 240 milliseconds, and fills at 1.0843. Both saw the same market. The second one saw it a quarter of a second later.

Over a thousand trades, two extra tenths of a pip per fill is a serious amount of money. It is also easy to blame on the strategy instead.

Several ingredients go into this:

- Distance between the trading server and the broker's matching engine
- The number of hops in the bridge
- How many liquidity providers contribute to the price, and how they are aggregated
- Whether the broker applies a delay filter on certain order types

The comparison that matters is fill price versus the price at signal time, not fill price versus screen price. The first measures total cost. The second only measures what the platform admits.

Has the same strategy ever been run on two live accounts at once, and how big was the gap in fills?

[Shadow Trader]

A few brokers publish order books or position ratios showing how many clients are long and short, and where pending orders cluster. They're popular on forums, and the usual pattern is contrarian: when most clients are long, price tends to fall.

Before building anything on that, consider the sample. It covers only clients of that broker, which isn't the whole market, and those clients are not representative of the institutional flow that moves price. The data updates every few minutes rather than in real time, and the clusters of stops can reveal where retail participants are vulnerable.

There's some value in knowing where stops pile up, since price is drawn to liquidity as covered in earlier posts. As a standalone signal, though, the ratios work inconsistently and in many periods they go the wrong way for weeks.

For a scalper, the information can help avoid placing entries in the most crowded zones. Seeing a large cluster of orders at a round figure suggests the level will be contested, and price action there may be violent.

Checking the same data from two brokers can show how different the "crowd" looks depending on the firm.

A related caution is that these data sets are often presented as "percent long" figures that swing between 30 and 70 percent. The extremes get the headlines, but most of the time the number hovers in the middle, saying nothing. Only the rare readings above 80 or below 20 deserve any attention, and even then the lead time is long and uncertain.

Does positioning data add anything to a scalping approach, or is it just curiosity?