There are 3 main "signatures" of order flow that raise TRADE INTENSITY if looking at a higher end data flow (Bloomberg or similar) on a sub second basis; 1) Capitulation TI (Losers LEAVING the market - "market order" driven order flow). 2) Commercials Covering TI (Winners LEAVING the market as the available LOSERS supply is getting gobbled up - "market order driven order flow). 3) Commercials Initiating TI (Commercials initiating new entries with algorithmic automated entry mechanisms to JOIN the market - "market order" driven order flow). There is enough difference between each of these three types of order flow TI events to statistically categorize their "signatures" (so you can know what TI signature is dominant at any time for determining trade entry probabilities). Obviously, everyone digging into this information would LOVE to know the exact moment in time when Commercials are initiating new directional trade. Until you can find the three separate "signatures" of order flow TI events then you will be stuck getting way too many signals to know which ones are valid (or of highest probability). Trying to pull this order flow "signature" identification off with a TS feed and DLL's seems not at the infrastructure level needed to properly utilize the full context of available information. My guess is that UB is doing ok (slightly profitable) with his developed TI methods, but is no where near where he would like to be on a ROI basis.....just my $.02 BTW, I do think it was GREAT for UB to show the methods he has been working on to get more traders looking at the "possibilities" in trading. Also, these TI based signals are the type which fully automated trading can take full advantage of, and imo is the best way to trade them. Having properly developed "BOTS" looking for the Commercials initiated TI events in 30 to 40 different markets 24 hours a day is the ultimate endgame.
Hey AMT, thanks for participating in my thread and thank you for sharing your ideas on this. The current limitation for me is Ninjatrader timestamps to 1 second. So going subsecond is a lot more difficult and would work realtime only. But I'm going to continue investigating this. When I first found UB's posts I didn't really know how to do such a thing, especially with buying & selling volume. So I kind of forgot about it. Many months later I saw a video on you tube and that's where I got the idea that intensity could be measured by the speed of the trades on the T&S. I had had T&S and would watch for flooding. And this indicator let me take my eyes off the T&S. Then someone mentioned UB and I reread all his posts again and I see I'm not that far from what he's doing. The biggest challenge is all the signals. That's why I'm not trading based on these signals alone. In fact I don't watch this chart while trading. I just hear the audio alert and it's one more input into my decision making process. If price is more likely to reverse after a spike but doesn't reverse after all spikes, then if I get a setup from X,Y,Z and I"m timing my entry and I get a TI spike then that's one more thing in my favor. I hope that made sense.
Makes total sense and I agree it is very good information to pay attention too, just like order book delta or "market order" driven order flow bias (Cumulative Delta). Great work BTW!
I forgot to add that UB is doing sub-second. On mine I've found the best results from using ticks instead of volume and 5 seconds. My hope is that by sharing the code for this others will be able to contribute, either in ideas (like you've done) or in code. And it'll be a learning experience for everyone.
Yes, sub second is where the "signatures" within the overall TI event can be found. Most all significant reversals during the trade day, for most futures instruments, do have one or several "signature" TI events as the cause. This is very good information to research and your blog is a great idea.....looks sharp!