Support Board
Date/Time: Tue, 26 Nov 2024 03:38:25 +0000
Numbers bar study bid - ask split incorrect behaviour
View Count: 2184
| [2014-03-13 08:04:59] |
| Pawel_Szpak - Posts: 48 |
|
Hi, Please have a look at the attached youtube clip to highlight exactly what I mean. Basically, I'm looking at a numbers bars chart along with a DOM. The issue is that when a price is bid (or offered) and that bid (or offer) gets cleared and that price now goes offered (or bid) without ticking up or down, the numbers bar study (and presumably everything else relying on this bid/ask split) keeps accumulating trades on the bid (or ask) side until the market actually makes a tick up or down. So in the case of the video below. Note how 142.87 has approx 176 contract sitting on the bid. This gets hit (164 contracts do the trick) and 142.87 is now trading offered. Any offers now lifted at 142.87 show up on the bid side on the numbers bar study instead of the offer side. This is surely incorrect. https://www.youtube.com/watch?v=_GSLGIOC6hQ This happens all the time when a price changes from bid to offer or vice versa without actually ticking up or down from that price. I've just quickly made a video this morning, but sometimes this affects quite a large number of contracts. Assuming the same bid/ask split then carries through to cumulative delta/other bid/ask studies - that data there must be incorrect too. Please can you advise what is going on here. |
| [2014-03-14 00:50:01] |
| |
|
The reason is because Sierra Chart uses up ticks and down ticks to establish bid trade volume and ask trade volume for the EUREX: Technical Discussion: EUREX Bid Volume and Ask Volume This produces a more accurate Cumulative Delta Bars study. The data feed from the EUREX is unusual compared to other data feeds and requires special handling. There is another algorithm we want to try and we will try to get to it as soon as we can. Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing Date Time Of Last Edit: 2014-03-14 00:50:27
|
| [2014-03-19 08:10:21] |
| |
|
So you know, we will be testing the new algorithm for EUREX bid volume and ask volume, tomorrow. Hopefully there will be an improvement . Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing |
| [2014-03-26 09:12:01] |
| Pawel_Szpak - Posts: 48 |
|
How did that testing of the new algorithm go? Any idea when it will be implemented? How will the new algorithm work? I'm not sure how you can say using up/down ticks produces a more accurate cumulative delta, when it's clearly mislabelling some bid/asks using the up tick/down tick rule, or are you specifically referring to "more accurate VS IQfeed" as per the claim in the link you posted. Also, I would say perhaps review a week's worth of cumulative delta on FGBL on IQfeed vs your feed. The cumulative delta's certainly vary a LOT more on the 2 approaches than is apparent. ie, at the moment, I'm seeing a cum delta of ~-9k on your feed, vs apparently ~+6k on iqfeed on FGBLM4 for today's session.(had to ask someone else, don't have screenshots of Iqfeed - 5m chart though). That's quite a difference. I'm slightly confused why this is difficult to code. Looking at the DOM in the video I had posted, Sierra can obviously tell whether a price is trading bid or offered. Why does that not just end up allocating the bids/offers like that? Is there some kind of latency issue because the DOM is coming live from OEC and the underlying tick data is a different feed via Barchart/Eurex? However, if it's latency, it can't be too large if I can see differences with the naked eye and make a video of it. |
| [2014-03-26 09:55:47] |
| |
|
The new algorithm did not give a reasonable result for the Cumulative Delta. Up/down ticks definitely produces a more reasonable/ accurate Cumulative Delta rather than just comparing to the bid or ask price. We have confirmed this. It is abundantly clear to us, that with the EUREX data feed, you cannot possibly get an accurate bid volume or ask volume with IQ Feed any data feed. Only an estimation. So doing a comparison to IQ Feed, to us is not valid. With what we know it is completely meaningless. It is not accurate. We are not saying it is very inaccurate, but certainly it is not 100% accurate. To what degree it is not known. The problem is there are trades which occur between the most recent reported bid and ask, frequently. Or above or below the bid and ask This is why with the EUREX the only stable way to do this is with up ticks and down ticks. If you are using the Sierra Chart Data Feed, the depth always comes from the Sierra Chart data feed. Not any other source. If we were using another source for depth, there would be less accuracy. It would not even work right. Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing Date Time Of Last Edit: 2014-03-26 09:56:37
|
| [2014-03-26 10:17:26] |
| |
|
The reason we say the result with IQ Feed is completely meaningless, is because it cannot be safely relied upon. Not that it is completely inaccurate, but because it cannot be relied upon, quoting a particular value with IQ Feed and comparing it to what you get with the Sierra Chart data feed, is meaningless because it proves nothing. This is the point we are trying to make. There is no valid point of reference because it is not possible to achieve accurate bid volume or ask volume with EUREX. We are still thinking about this but in the end, we only have come to the conclusion that we have to really understand what the objective here really is. Only users can answer that for themselves. And what is the most stable way of doing the calculation? And the most stable way is to use the price feed itself and that is to look at up ticks and down ticks. One area we could possibly change is when a trade occurs at the same price as before. Possibly if it matches exactly the current quoted bid or ask, we could use those to count the trade as either a bid or ask trade. That might be a reasonable algorithm but still that is imperfect. Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing Date Time Of Last Edit: 2014-03-28 07:49:56
|
| [2014-03-26 11:27:57] |
| User34947 - Posts: 42 |
|
One area we could possibly change is when a trade occurs at the same price as before. Possibly if it matches exactly the current quoted bid or ask, we could use those to count the trade as either a bid or ask trade. That might be a reasonable algorithm but still that is imperfect.
This would be a big improvement because it would solve the problem when one level of bid or ask is scooped up (which is the main difference between the upticks/downticks split and the bid/ask split.Simply as it is, I think that traders who use number bars or DOM scalpers would like to see a trade counted as either bid or ask trade when they see that such trade matches the first level of bid/ask that we see in our DOM. Unfortunately for us I think that we have difficulties in "seeing" trading between the most recented reported bid and ask or out of them. In most of the cases, this is only my opinion, it seems clear for what I see on the DOM where the trade occurred but of course it seems that is not so easy. Personally I am not so interested in the bid/ask split or in cumulative delta but I use your platform and your Eurex feed and I think they are both excellent, and it is really a pity if some traders are not considering your products only for this issue, but it is an important one, so I hope that you will find a solution, not the perfect one which does not exist, but as you say the one that corresponds more accurately to the needs of the users. Date Time Of Last Edit: 2014-03-30 12:24:47
|
| [2014-03-27 22:31:28] |
| |
|
We will be working on this today.
Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing Date Time Of Last Edit: 2014-03-27 23:00:06
|
| [2014-03-28 07:51:57] |
| |
|
We have deployed a change to the Sierra Chart EUREX data feed where when a trade occurs at the same price as before and exactly matches the latest quoted Bid, then it is considered a Bid trade. If it exactly matches the quoted Ask, then it is considered an Ask trade. Otherwise, logic is as before, where if the market trades higher or trades at the same price and previously traded higher, then it is considered an Ask trade. If the market trades lower were trades at the same price and previously traded lower, then it is considered a Bid trade. We have been monitoring it this morning and the algorithm works well and has been deployed across all servers. At this point, we think we have succeeded with providing a very good Bid Trade Volume and Ask Trade Volume estimation with EUREX that is sensible. Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing Date Time Of Last Edit: 2014-03-28 07:55:56
|
| [2014-03-30 04:01:23] |
| |
|
As we said the new Bid volume and Ask volume algorithm has been applied on 2014-3-28 for the EUREX. When you look at the Cumulative Delta Bars for 2014-3-28 for the DAX, using any data feed whether it is IQ Feed or the Sierra Chart data feed, the Cumulative Delta Bars goes negative even though the market was up for the day. None of this makes any sense. which only proves our point, that you cannot accurately do Bid volume or ask volume for the EUREX. It seems to be questionable from our perspective to do bid volume and ask volume analysis on the EUREX. This is why we adopted the uptick and downtick method as an alternative. Which we are still using with the modification as stated, but nevertheless, it still appears to give a nonsensical result from time to time unless there is a very good reason why the Cumulative Delta would go in the opposite direction as the price of the market. Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing Date Time Of Last Edit: 2014-03-30 04:02:07
|
| [2014-03-30 09:10:08] |
| Futurestrader77 - Posts: 126 |
|
Is it my understanding that you cannot rely upon the cumulative delta on thin markets like the DAX. For example you find the same discrepancies in the Euro. I have attached two pictures, one showing the cumulative delta (not reset every day) going down while the Euro is going up and cumulative delta going sideways while the Euro is going down. In the other picture you see the cumulative delta (reset every day) going down while the Euro is going sideways. I personally do not use the cumulative delta but the DOM, therefore for me it is important that I can rely upon the information I get. What Pawel describes in his post should not happen with your data feed because he is not talking about trades occurring between, above or below the current bid and ask. What I want to say, you should not intend to get the same results with the cumulative delta like IQFeed but make sure that the problem Pawl describes will be resolved. Your data feed is really good and cheap and I personally am very satisfied. Date Time Of Last Edit: 2014-03-30 09:22:25
|
 Euro 1.PNG / V - Attached On 2014-03-30 09:09:09 UTC - Size: 83.61 KB - 462 views Euro 2.PNG / V - Attached On 2014-03-30 09:09:18 UTC - Size: 84.59 KB - 472 views |
| [2014-03-30 11:07:59] |
| Pawel_Szpak - Posts: 48 |
|
Just to clarify, my main issue was always that looking at the numbers bar study & dom the trades were not accumulating on the side of the aggressor - if an offer gets lifted that trade should count on the offer side of the numbers bar study and vice versa for bids getting hit. Anything else is just incorrect. While I'll accept your argument that it is impossible to make it 100% correct, any improvements (such as what you have suggested and done above) are welcome. The fact that cumulative delta is affected is just an extra point as its data would (to my understanding) be derived from the same source as the numbers bars data. Now I don't agree that cumulative delta going down while the market is up on the day is outright non-sensical. It just appears that people were happy to sit on the bid and step up occasionally and soak up a lot of "aggressive" sellers hitting those bids. Certainly there is no reason for the market and delta to always be going in the same direction. |
| [2014-03-30 11:49:46] |
| User34947 - Posts: 42 |
|
I think that it is clear to everyone who uses cumulative delta that delta and price do not have to go always in the same direction, as Pawel_Szpak correctly explained and as FuturesTrader77 showed with a Euro Chart where bid/ask split is coming directly from CME (am I correct?), so there is absolutely no point in stating the correctedness or not of the bid/ask split looking at such things. On last Friday I looked for a short time to FGBL and it seems to me that your last improvement works really well and I would like that other users could comment on that. Date Time Of Last Edit: 2014-03-30 12:11:38
|
To post a message in this thread, you need to log in with your Sierra Chart account: