IC Markets cTrader DOM for order flow scalps: first impressions

[LondonScalper]

Three weeks with the depth of market window on my IC Markets cTrader account, mostly on EURUSD and gold between 07:00 and 10:30 London. These are first impressions, so treat them as notes from someone still learning the tool rather than verdicts.

What I expected: a view of where size was sitting so I could lean on it for entries.

What I actually got: a much clearer picture of how much liquidity there is on each price level at any one moment, and a sense of how quickly it changes. On EURUSD during the first ten minutes after 08:00 the visible volume at the top levels refreshes so fast that "reading" it in any traditional futures sense was hopeless for me.

Where it did help was in the quieter stretch from 09:15 onward. I started noticing that when the volume shown on the offer side thinned out over two or three refreshes, price tended to lift a few tenths soon after. Is that a real pattern or my brain doing what brains do? I have 41 tagged observations so far, which is not enough to say.

Practical bits. The DOM ticket lets you click a price level to place a limit, which is quick but risky with a wandering mouse. I disabled one-click for the first week. The Level II on gold showed noticeably wider gaps between levels around 13:30, which matched what my fills told me anyway.

I'll keep logging until I have two hundred observations before I let it change a single rule.

[Shadow Trader]

Treating the first ten minutes after 08:00 as unreadable matches what I saw on EURUSD too. The refresh rate is so high that what you're looking at is mostly market makers updating quotes, not orders from real participants.

The 09:15 observation is interesting, but I'd be careful with how it's measured. If the offer thins over three refreshes, the spread is often already moving, so the "lift" may just be the mid price moving with the quote. I'd log the mid at the time of the observation and 30 seconds later, rather than the last trade price.

Another thing I'd add to the log is whether the bid side changed at the same time. A thin offer with an unchanged bid is a different situation from both sides thinning.

Turning off one click at first was sensible. I've seen someone place a limit order a full figure away by clicking the wrong row.

Two hundred observations before changing a rule is the right number. Are you recording the ones where the offer thinned and price didn't lift, or only the ones you noticed?

[Fairman]

I started noticing that when the volume shown on the offer side thinned out over two or three refreshes, price tended to lift a few tenths soon after.

Worth knowing what that DOM actually is, because it changes how to read the observation. On a retail cTrader account the depth shows the quotes your broker receives from its liquidity providers, aggregated, not resting orders from other traders as on a futures exchange. The sizes are what each LP is willing to fill at that price at that instant. So when the offer side thins over a few refreshes, the likely explanation is LPs pulling or skewing their offers because their own models or flows already point up, and the lift that follows is them repricing. That can still be a usable short-term tell, but it's an LP reaction rather than someone absorbing size. It also means the effect may differ by broker, since each has a different LP mix, and it will likely vanish in fast markets when everyone reprices at once.

[Shadow Trader]

Fairman's note that retail cTrader depth is LP quotes, not resting trader size, is the frame I'd keep. Once I accepted that, the tell I trust changed.

What I now watch for after 09:15 is a stack that sits unchanged for five or more refreshes on one side while the other side keeps updating. That sticky stack has been a better short term lean than a thin side that refreshes every tick.

The tell I ignore is a big size that appears for one refresh and vanishes. That looks like spoofing on a futures DOM, but on an LP feed it is usually just a quote flicker and has no edge in my log.

I still will not let DOM change an entry rule until I hit two hundred tagged cases. So far the sticky stack is at 64 observations with a mild lift or drop in the next 20 seconds about six times in ten.

LondonScalper, are you tagging how long the size sat, or only that it thinned? The duration feels like the part that separates noise from a usable lean.