NexusFi: Find Your Edge


Home Menu

 





Is Cumulative Delta useless?


Discussion in Traders Hideout

Updated
      Top Posters
    1. looks_one loantelligence with 10 posts (12 thanks)
    2. looks_two eisi83 with 4 posts (0 thanks)
    3. looks_3 iantg with 2 posts (29 thanks)
    4. looks_4 canoekoh with 2 posts (6 thanks)
      Best Posters
    1. looks_one iantg with 14.5 thanks per post
    2. looks_two canoekoh with 3 thanks per post
    3. looks_3 Anagami with 2.5 thanks per post
    4. looks_4 loantelligence with 1.2 thanks per post
    1. trending_up 26,172 views
    2. thumb_up 73 thanks given
    3. group 33 followers
    1. forum 32 posts
    2. attach_file 13 attachments




 
Search this Thread

Is Cumulative Delta useless?

  #11 (permalink)
 oasisjoe 
Taipei, Taiwan
 
Experience: Intermediate
Platform: Custom
Broker: IB/IQFeed
Trading: HG
Posts: 30 since Jul 2014
Thanks Given: 21
Thanks Received: 25

Cumulative delta, like any other indicators, is a reflection of the past. Yes, I know sell orders outnumbers buy orders in the last hour, but the big seller may already be done with selling and price reversion may be just around the corner. Or alternatively, another big sell order may arrive at an unknown time later with TWAP. In either case, cumulative delta simply won't tell you anything about the future. You need another edge like crystal ball to predict the future first , then you can use cumulative delta as a confirmation tool.

Reply With Quote
Thanked by:

Can you help answer these questions
from other members on NexusFi?
NT7 Indicator Script Troubleshooting - Camarilla Pivots
NinjaTrader
REcommedations for programming help
Sierra Chart
MC PL editor upgrade
MultiCharts
How to apply profiles
Traders Hideout
Pivot Indicator like the old SwingTemp by Big Mike
NinjaTrader
 
  #12 (permalink)
 
mmaker's Avatar
 mmaker 
Toronto Canada
 
Experience: Intermediate
Platform: ninjatrader
Trading: es
Frequency: Several times daily
Duration: Minutes
Posts: 400 since Feb 2011
Thanks Given: 1,178
Thanks Received: 508


iantg View Post
From what I have seen, everyone is looking at the wrong things. Transacted volume, level 2 and even level 1 volume needs more context to be helpful. I have never seen a software company, indicator guru, or retail trader get any of this information correctly classified in terms of a meaningful micro-structure. There are bets you can make regarding volume that are very good, but most of these are in the HF space.

While I can't share any edges with you, I can get you thinking a little bit about this from a different angle.

Every single price level has a fight between the bid and ask right? But if you look deeper there is much much more going on that everyone misses. Every price level starts with one side dragging in resting volume from the DOM, and the other side will start with 0 because it is a newly created price level. So statistically speaking 95% of the time, the ES for example clears the weak side that started with 0 volume. So you can predict with very high certainty which side will win the immediate price level. The obvious issue though is the participation paradox, where you can only participate in this edge by a toxic fill, so this alone will get you nothing more than a 1 tick toxic fill that will 95% put you neutral by flipping to the adjacent strong side price level.

The real price fight, that I have never seen anyone discuss or even notice... (Not saying people haven't found this, but that no one is discussing it here at least) is between the two adjacent strong sides that keep breaking their respective weak sides over and over.

Enclosed is an example of a fully flushed out micro structure of the ES.

If you follow the data in the example, you will see that the Ask has heavy starting volume on one level and wins, then on the next level the bid has heavy starting volume. Each of these keep flipping back and forth with their respective weak sides breaking until something special happens. One of the two weak sides gets a huge surge of limit orders that stack their level up. This in turn temporarily freezes the price level and exposes for the first time the adjacent strong side to the heat (Market orders). Once the market orders hit 25% to 50% of the starting volume, the HF guys bail and you will see the cancels hit in mass on the strong side.

These are the kinds of events that occur on a micro-structure level and really do move the market, but everyone is mostly looking at lagging stuff that has no context, isn't correctly even sequenced, and has little to no predictive value.

I can show this just to give you an alternate perspective, but the bad news is, 99% of people will never be able to participate in these types of edges for obvious reasons.

Best of luck.

Ian

I am using ninja and I have always wanted to identify the volume of pulled orders using OnMarketDepth(MarketDepthEventArgs).

The adds are easy to identify when reading incoming volume. Ninja has two other possibilities "update" and "remove". Did you assume that all the updates are due to transacted volume and all the removes were cancelled volume (pulled orders)?

Thanks in advance.

Visit my NexusFi Trade Journal Reply With Quote
Thanked by:
  #13 (permalink)
 iantg 
charlotte nc
 
Experience: Advanced
Platform: My Own System
Broker: Optimus
Trading: Emini (ES, YM, NQ, ect.)
Posts: 408 since Jan 2015
Thanks Given: 90
Thanks Received: 1,148


Hi,

If you are looking to differentiate adds vs. cancels vs. transactions I can point you in the right direction. The Update, Add, Remove status has nothing to do with cancels vs. adds. This relates to if the price is being added to the DOM because it is newly created or if it is being removed because the price level just cleared.

So with Ninjatrader all the volume updates are aggregates so you have to run your own logic to break this out.

With strictly level 2 that is far far away from the transaction level, this is easy. You will need to assign a variable to hold a snapshot of the volume at whatever given level you are tracking, then when you get a new update, you just calc the delta. Like this.

200 = starting volume
205 = + 5 Added
195 = - 10 Canceled
etc...

And you can just keep summing your add and cancel variables through the life of that price level. As long as all the action lives in level 2 ( The price never hits the best bid / best ask) that is all there is to it.

But for anything that hits level 1, there is added complexity of differentiating transactions from cancels.
Again, with the market by price feed everything is just aggregated so we don't get any specific messaging denoting a cancel vs. a transaction so we have to break this out with some logic.

For level 1 data NT has 3 feeds. Ask, Bid, and Last. The Last is the transaction feed. Within the Last feed you can point to the bid or last with some basic < or > references to the bid or ask price to determine where the transaction hit. Unlike level 2 where we can assume any downward movement is a cancel, with level 1 you have to have a little more logic. So applying the same example as before, but with level 1 data, we would characterize it like this.


200 = starting volume
205 = + 5 Added
195 = - 10 Subtracted (Maybe cancel or transaction but we don't know)
etc...

Then you run your total sum of Subtracted volume - the total sum of Transaction volume, the difference is your canceled volume. And that is it.

If there is enough interest I might share more of my microstructure research on a new thread at some point.

Ian



mmaker View Post
I am using ninja and I have always wanted to identify the volume of pulled orders using OnMarketDepth(MarketDepthEventArgs).

The adds are easy to identify when reading incoming volume. Ninja has two other possibilities "update" and "remove". Did you assume that all the updates are due to transacted volume and all the removes were cancelled volume (pulled orders)?

Thanks in advance.


In the analytical world there is no such thing as art, there is only the science you know and the science you don't know. Characterizing the science you don't know as "art" is a fools game.
Visit my NexusFi Trade Journal Reply With Quote
  #14 (permalink)
 
mmaker's Avatar
 mmaker 
Toronto Canada
 
Experience: Intermediate
Platform: ninjatrader
Trading: es
Frequency: Several times daily
Duration: Minutes
Posts: 400 since Feb 2011
Thanks Given: 1,178
Thanks Received: 508


iantg View Post
Hi,

So applying the same example as before, but with level 1 data, we would characterize it like this.


200 = starting volume
205 = + 5 Added
195 = - 10 Subtracted (Maybe cancel or transaction but we don't know)
etc...

Then you run your total sum of Subtracted volume - the total sum of Transaction volume, the difference is your canceled volume. And that is it.


Ian

That was the part that was giving me grief. Thanks for clearing that up. I have some old code kicking around. maybe i will ressurrect it. Gawd - that might have been and prolly was NT7.

I used to print out a mini dom beside price with the delta for each bid and ask level in separate additional columns.


Visit my NexusFi Trade Journal Reply With Quote
Thanked by:
  #15 (permalink)
lumanauw
Bandung
 
Posts: 1 since Feb 2019
Thanks Given: 1
Thanks Received: 1

Good explenation, how to use this logic for trading forex/gold?


iantg View Post
From what I have seen, everyone is looking at the wrong things. Transacted volume, level 2 and even level 1 volume needs more context to be helpful. I have never seen a software company, indicator guru, or retail trader get any of this information correctly classified in terms of a meaningful micro-structure. There are bets you can make regarding volume that are very good, but most of these are in the HF space.

While I can't share any edges with you, I can get you thinking a little bit about this from a different angle.

Every single price level has a fight between the bid and ask right? But if you look deeper there is much much more going on that everyone misses. Every price level starts with one side dragging in resting volume from the DOM, and the other side will start with 0 because it is a newly created price level. So statistically speaking 95% of the time, the ES for example clears the weak side that started with 0 volume. So you can predict with very high certainty which side will win the immediate price level. The obvious issue though is the participation paradox, where you can only participate in this edge by a toxic fill, so this alone will get you nothing more than a 1 tick toxic fill that will 95% put you neutral by flipping to the adjacent strong side price level.

The real price fight, that I have never seen anyone discuss or even notice... (Not saying people haven't found this, but that no one is discussing it here at least) is between the two adjacent strong sides that keep breaking their respective weak sides over and over.

Enclosed is an example of a fully flushed out micro structure of the ES.

If you follow the data in the example, you will see that the Ask has heavy starting volume on one level and wins, then on the next level the bid has heavy starting volume. Each of these keep flipping back and forth with their respective weak sides breaking until something special happens. One of the two weak sides gets a huge surge of limit orders that stack their level up. This in turn temporarily freezes the price level and exposes for the first time the adjacent strong side to the heat (Market orders). Once the market orders hit 25% to 50% of the starting volume, the HF guys bail and you will see the cancels hit in mass on the strong side.

These are the kinds of events that occur on a micro-structure level and really do move the market, but everyone is mostly looking at lagging stuff that has no context, isn't correctly even sequenced, and has little to no predictive value.

I can show this just to give you an alternate perspective, but the bad news is, 99% of people will never be able to participate in these types of edges for obvious reasons.

Best of luck.

Ian


Reply With Quote
Thanked by:
  #16 (permalink)
LaserChicken
Cheltenham, United Kingdom
 
Posts: 5 since Jul 2019
Thanks Given: 1
Thanks Received: 9

I think cumulative delta is certainly not useless, when used in confluence/support of an existing thesis, so really for confirming what you already think. I don't think you can get much use out of it on its own... just my two cents!

Reply With Quote
Thanked by:
  #17 (permalink)
 
Silvester17's Avatar
 Silvester17 
Columbus, OH
Market Wizard
 
Experience: None
Platform: NT 8, TOS
Trading: ES
Posts: 3,603 since Aug 2009
Thanks Given: 5,139
Thanks Received: 11,527

I think it can be useful by one of the very basic functions:

- does cd confirm new highs and new lows? (of course it can also be applied to swing highs and lows)



Reply With Quote
Thanked by:
  #18 (permalink)
 loantelligence 
Syracuse, NY
 
Experience: Intermediate
Platform: Ninja Trader
Broker: Mirus Futures/Zen-Fire
Trading: NQ
Posts: 218 since Jan 2011
Thanks Given: 31
Thanks Received: 189

I use the Orderflow Cumulative Delta on NT8 to show the direction of the market...one the NQ and MNQ...works great for scalping....in fact I give it more weight than the unirenko chart...see examples...

Attached Thumbnails
Click image for larger version

Name:	2019-08-29_1255Last_trade.png
Views:	894
Size:	130.3 KB
ID:	275147   Click image for larger version

Name:	2019-08-29_112811_19_NQ...showing_OrderFlowCumulativeDelta_and_JO_Colored_Zones_on_the_right.png
Views:	660
Size:	133.2 KB
ID:	275150   Click image for larger version

Name:	2019-08-16_1015Crude_this_AM.png
Views:	666
Size:	50.1 KB
ID:	275153  
Reply With Quote
  #19 (permalink)
 loantelligence 
Syracuse, NY
 
Experience: Intermediate
Platform: Ninja Trader
Broker: Mirus Futures/Zen-Fire
Trading: NQ
Posts: 218 since Jan 2011
Thanks Given: 31
Thanks Received: 189

just another example now using NT8....added ADXHistogram and SMA histogram with the OrderFlowCumulative Delta...also corresponds with the Dynamic SR Color Zones on the Chart...almost a no brainer.....I use the SMA historgram….has to be greater/lesser than 3.5'/-3.5 to stay away from chop....once the ADX turns purple....golden opportunity..

Attached Thumbnails
Click image for larger version

Name:	2019-09-15_2126current_screens.png
Views:	744
Size:	161.3 KB
ID:	275156  
Reply With Quote
Thanked by:
  #20 (permalink)
 eisi83 
Salzburg Austria
 
Experience: None
Platform: NinjaTrader
Trading: ES,YM
Posts: 16 since Sep 2017
Thanks Given: 41
Thanks Received: 3



loantelligence View Post
just another example now using NT8....added ADXHistogram and SMA histogram with the OrderFlowCumulative Delta...also corresponds with the Dynamic SR Color Zones on the Chart...almost a no brainer.....I use the SMA historgram….has to be greater/lesser than 3.5'/-3.5 to stay away from chop....once the ADX turns purple....golden opportunity..

Are these Indis freeware or did u buy/program it?

Reply With Quote




Last Updated on September 12, 2021


© 2024 NexusFi™, s.a., All Rights Reserved.
Av Ricardo J. Alfaro, Century Tower, Panama City, Panama, Ph: +507 833-9432 (Panama and Intl), +1 888-312-3001 (USA and Canada)
All information is for educational use only and is not investment advice. There is a substantial risk of loss in trading commodity futures, stocks, options and foreign exchange products. Past performance is not indicative of future results.
About Us - Contact Us - Site Rules, Acceptable Use, and Terms and Conditions - Privacy Policy - Downloads - Top
no new posts