order flow analysis

[⭐ aotegaoteg]

I have been backtesting nearly 10 years of historical data across different asset classes using Ninjatrader market replay data. What I'm about to describe may be controversial, but I have to say it: using order flow as a strategy does not work and provides no real edge.

I backtested the following order flow strategies using high-quality data:

• Footprint
• Absorption
• Imbalance
• Delta
• Delta divergence
• Cumulative Volume Delta 
• Volume Profile
• Iceberg
• Order block
• Market depth

I can conclude that 90% of these strategies produced no meaningful results. After digging deeper to confirm this, I discovered changes that occurred in 2017 to the real-time streaming trading data distributed by the CME.

In 2017, the CME significantly altered order flow data through the introduction of the Market by Order MBO data format, which is now used by most retail traders and trading platforms. This new format changed how trading data is streamed to platforms. It's well documented that these changes obstructed the visibility of traditional order flow data compared to pre-MBO feeds.

The current MBO format obscures aggregated orders at price levels and provides only snapshots of executed trades. This masks resting orders, making footprint analysis and similar techniques far more complicated.

So why do many traders still try using order flow strategies for day trading if they're largely useless?

As I noted earlier, 90% of these strategies produced no meaningful results. The only ones I found with some potential edge are those focused on imbalances between bids and asks. Additionally, analyzing delta and cumulative volume delta can help spot traps and divergences in buying versus selling pressure. Other strategies, like absorption and iceberg orders, are simply not visible in the streaming data.

Edited 1 hour ago by aotegaoteg
add more

[⭐ laser1000it]

3 minutes ago, aotegaoteg said:

I have been backtesting nearly 10 years of historical data.....

Totally agree with what you wrote

[⭐ RichardGere]

Thank you for sharing with us your observations. 👍