I Tested Tick-Level Scalping on Gold. The Spread Ate Everything.

I measured tick-level micro patterns on XAUUSD in MetaTrader 5. Most edges were smaller than the spread, and the only survivor needed about 50 ms execution or faster.

Many traders dream of a fast scalper on Gold: small targets, many trades, steady profit. Before building one, I wanted to know if the edge is real. So I measured it.

What I tested

I wrote a research-only Expert Advisor for MetaTrader 5. It did not trade. It only watched every tick on XAUUSD and recorded what price did in the next seconds after common "micro patterns":

• several ticks in the same direction,
• a sudden fast move (a "spike"),
• a pause after a fast move.

For each pattern I measured the average move that followed, in points.

Result 1: the edges exist, but they are tiny

Some patterns really did continue more often than not. On paper, they looked like winners.

Then I compared them with the cost of one trade. The typical spread on my broker was about 10–11 points on Gold. Almost every micro edge I found was smaller than the spread.

In simple words: you pay more to enter than the pattern gives back. A strategy can win most of its trades before costs and still lose money after costs.

Result 2: speed decides the rest

The only idea that survived costs was continuation after a strong spike. But it only worked when the order reached the server very fast.

When I added realistic execution delays in the tester, the result changed quickly:

• around 50 ms or less: still positive,
• slower than that: the edge disappeared.

My own ping to the broker was about 52 ms, right at the limit. A normal home PC or a far-away VPS is simply too slow for this type of trading.

What I learned

1. Always test with real ticks. "Open prices only" or "1 minute OHLC" can hide what happens inside the candle.
2. Always include the real spread and a delay. In the Strategy Tester, set the execution delay to your real ping. A strategy that only works with 0 ms delay does not work.
3. Be suspicious of very high win rates on tiny targets. Check the average win versus the spread.
4. Fast scalping is a speed business. Without very low latency, it is better to trade on higher timeframes where the spread is a small part of the move.

Because of this test, I did not release a tick scalper. My Gold tools work on normal candles, where a typical move is many times larger than the spread.

Try it yourself

You can repeat this kind of test without writing code:

1. Open the Strategy Tester.
2. Choose "Every tick based on real ticks".
3. Set "Delays" to your real ping (you can see it in the bottom right corner of the terminal).
4. Run the same EA with 0 ms delay and with your real delay, and compare.

If the result changes a lot, the strategy depends on speed you probably do not have.

I share more notes like this and my free MT5 tools in my channel: https://www.mql5.com/en/channels/msbgoldpro

Trading involves risk. Backtest results do not guarantee future results.

Kemal Mustafa Ozkan

MSB Pro ALGO