Today I had a problem with "ABNB". There is a price difference between WealthData and Q-Data. WD's closing price is 139.42 and Q-Data has a closing price of 143.25. This is a difference of 2.7%. Various sources show a price close to 139.
Well, this cost me money today. This is important because this price is used for the strategy screener signals (please correct me if I am wrong).
Can you explain such a big difference and do you think there are more symbols with such a big difference?
Can you tell me more about Q-data. So far I have used it confidently for backtesting, and also the signals in the strategy screener tool. However, $4 difference is really too much for such a stock, or do you disagree?
Well, this cost me money today. This is important because this price is used for the strategy screener signals (please correct me if I am wrong).
Can you explain such a big difference and do you think there are more symbols with such a big difference?
Can you tell me more about Q-data. So far I have used it confidently for backtesting, and also the signals in the strategy screener tool. However, $4 difference is really too much for such a stock, or do you disagree?
Rename
Generally WealthData is more reliable, our team scrutinizes those symbols and corrects prices even many other vendors miss. QData is a more automated provider which consumes from several sources.
Well, if WealthData had also been consumed, this would not have happened.
Don't you use some kind of verification tool that gives alerts when there are significant price differences? I mean, there will always be some minor differences between different sources, but more than 2.5% should trigger some kind of alert and correction before being inserted into a trading tool like the strategy screener. (Backtesting issues are of course also involved at this point but secondary).
Don't you use some kind of verification tool that gives alerts when there are significant price differences? I mean, there will always be some minor differences between different sources, but more than 2.5% should trigger some kind of alert and correction before being inserted into a trading tool like the strategy screener. (Backtesting issues are of course also involved at this point but secondary).
Good point, we'll work to beef up QData with your recommended alerts.
I have thought about the issue again and would like to raise two points. These could be taken into account in the improvement.
1. signals from the strategy window could/should also be verified. One could add a button "Verify signals". This is useful for the users who submit the signals to the Quotes tool. (Edit: Although the verifcation for the signals would be independent on the data provider i could easily understand if there would be a restriction to strategies using Q-Data / Wealth Data)
2. if one uses the Strategy Screener, it can generate many more signals for the universe than a strategy can for its data set. If one could filter a list (data set) in the screener by the symbols they want, that would also be useful to get these "verified" signals.
Why the prices should be verified is self-explanatory. Why the signals are generated via the strategy or why the list should be filtered in the screener I can briefly explain.
Although the Quotes tool can process a virtually unlimited number of signals or many signals can be split across multiple Quotes windows, there are limitations that have not been considered so far.
One limitation, for example, is that there are brokers who only allow a limited number of simultaneous requests. If there are more requests in the quote window than allowed, the functionality is disrupted.
This is the reason why you want to get the signals from the strategy window and process them in the quote tool. You get the signals for a defined data set. At the same time you would like to have a verifcation of prices too.
Conversely, if you use the screener for a large universe, you might just get too many signals. Preparing the results for the quote tool by comparing them with a list of symbols (data set) would implicitly circumvent this kind of limitation. With the filter you can influence which and how many symbols are transferred to a Quotes Window. The best thing about it would be that the prices are verified.
Have a nice weekend.
1. signals from the strategy window could/should also be verified. One could add a button "Verify signals". This is useful for the users who submit the signals to the Quotes tool. (Edit: Although the verifcation for the signals would be independent on the data provider i could easily understand if there would be a restriction to strategies using Q-Data / Wealth Data)
2. if one uses the Strategy Screener, it can generate many more signals for the universe than a strategy can for its data set. If one could filter a list (data set) in the screener by the symbols they want, that would also be useful to get these "verified" signals.
Why the prices should be verified is self-explanatory. Why the signals are generated via the strategy or why the list should be filtered in the screener I can briefly explain.
Although the Quotes tool can process a virtually unlimited number of signals or many signals can be split across multiple Quotes windows, there are limitations that have not been considered so far.
One limitation, for example, is that there are brokers who only allow a limited number of simultaneous requests. If there are more requests in the quote window than allowed, the functionality is disrupted.
This is the reason why you want to get the signals from the strategy window and process them in the quote tool. You get the signals for a defined data set. At the same time you would like to have a verifcation of prices too.
Conversely, if you use the screener for a large universe, you might just get too many signals. Preparing the results for the quote tool by comparing them with a list of symbols (data set) would implicitly circumvent this kind of limitation. With the filter you can influence which and how many symbols are transferred to a Quotes Window. The best thing about it would be that the prices are verified.
Have a nice weekend.
All Data Providers, including Wealth-Data, are prone to errors, but we can correct Wealth-Data errors after discovery. Furthermore, we expect Wealth-Data to be different from other EOD Providers. Normally the settled close should be identical - but it's even possible that the current day's settled close be different if a special dividend (or some other adjustment for corporate actions) is applied.
Knowing this insight, how do you know which data provider has the "correct" data?
What exactly are you verifying with "Verify Signals, and how do you verify them"?
Knowing this insight, how do you know which data provider has the "correct" data?
What exactly are you verifying with "Verify Signals, and how do you verify them"?
Hi Cone.
1. Identify a bad price.
The first and most important step in checking a "bad" price is to identify it. This can be done by comparing two prices from different sources. If there is a difference with a certain amount, you will get your alert.
1. choose two sources and compare the price
2. choose an acceptance level for the difference
3. generate an alarm signal in case of a big difference
In the example of ABNB (Q-Data) compared to ABNB Wealthdata, there would have been a difference of at least 2.5%. This must be detected and can be detected At this point it does not matter which price is false, but both prices cannot be true.
Now the simplest handling would be for the user to choose their own price.
What can the user expect?
Up to this point, there is no magic involved. Before looking for a solution to the correction part, it must be clear what can be done. In #2 it is clearly stated that the requirement is not to get all data from different sources identical. This will never be the case.
I think it is a fair expectation for the product owner and the user that Wealthdata and Q-Data can be used as a reference. This data belongs to the product. The process described for the alerts can of course also be used to prepare Wealthdata and Q-Data behind the scenes.
From this point on, a third data source can be compared with these reference data.
The level of acceptance or alarm can be set in the trading settings.
Technical solution
I think Welathdata and Q-Data should be prepared behind the scenes as described in the alarm part. Choose a data source of your choice to see the big differences. If this is done, the people who use WD and QD for generating signals will get the best you can offer. (According to a comparison with an independent source).
If you want to go a step further, you can compare signals with WD and QD.
Let's say my strategy generates signals based on IB data. A button in the strategy window like "Verify price with WD/QD" would be great to identify problems or before sending them to the quotes window.
Trading relevance
Since we are not talking about backtesting but trading, we are talking about real money at the same time. This improvement will save the community money.
I only hit one symbol yesterday. I wanted to buy the dip with N%, which ended up being 2.5% above because of the wrong price. The dangerous thing is that we don't know how often this happens without realising it.
Now, if you compare WD/QD with a third party data provider of your choice, it is very, very unlikely that you will encounter data problems with differences in this order of magnitude when you trade. You will know beforehand, and that is the most important part of this process.
1. Identify a bad price.
The first and most important step in checking a "bad" price is to identify it. This can be done by comparing two prices from different sources. If there is a difference with a certain amount, you will get your alert.
1. choose two sources and compare the price
2. choose an acceptance level for the difference
3. generate an alarm signal in case of a big difference
In the example of ABNB (Q-Data) compared to ABNB Wealthdata, there would have been a difference of at least 2.5%. This must be detected and can be detected At this point it does not matter which price is false, but both prices cannot be true.
Now the simplest handling would be for the user to choose their own price.
What can the user expect?
Up to this point, there is no magic involved. Before looking for a solution to the correction part, it must be clear what can be done. In #2 it is clearly stated that the requirement is not to get all data from different sources identical. This will never be the case.
I think it is a fair expectation for the product owner and the user that Wealthdata and Q-Data can be used as a reference. This data belongs to the product. The process described for the alerts can of course also be used to prepare Wealthdata and Q-Data behind the scenes.
From this point on, a third data source can be compared with these reference data.
The level of acceptance or alarm can be set in the trading settings.
Technical solution
I think Welathdata and Q-Data should be prepared behind the scenes as described in the alarm part. Choose a data source of your choice to see the big differences. If this is done, the people who use WD and QD for generating signals will get the best you can offer. (According to a comparison with an independent source).
If you want to go a step further, you can compare signals with WD and QD.
Let's say my strategy generates signals based on IB data. A button in the strategy window like "Verify price with WD/QD" would be great to identify problems or before sending them to the quotes window.
Trading relevance
Since we are not talking about backtesting but trading, we are talking about real money at the same time. This improvement will save the community money.
I only hit one symbol yesterday. I wanted to buy the dip with N%, which ended up being 2.5% above because of the wrong price. The dangerous thing is that we don't know how often this happens without realising it.
Now, if you compare WD/QD with a third party data provider of your choice, it is very, very unlikely that you will encounter data problems with differences in this order of magnitude when you trade. You will know beforehand, and that is the most important part of this process.
Did you have a look at this other extension, from DrKoch?
https://www.wealth-lab.com/extension/detail/finantic.BestData
It's one of the very few that I didn't yet buy (I mostly trade futures), but it looks like it would address most of your data provider concerns, by combining them and finding the best quotes, using he principle of redundancy.
https://www.wealth-lab.com/extension/detail/finantic.BestData
It's one of the very few that I didn't yet buy (I mostly trade futures), but it looks like it would address most of your data provider concerns, by combining them and finding the best quotes, using he principle of redundancy.
IMHO, I believe we should not make the product even more complicated with "Verify"-like options and creating the unnecessary choice for the user. Errors in the data must be fixed at the data feed side, following an automatic alert to the admins (i.e. the "verification tool that gives alerts" that @MIH mentioned in Post #2).
@alkimist's pointer is very good in this context, too. It's the best that can be done to the multivendor data, and it's already done.
@alkimist's pointer is very good in this context, too. It's the best that can be done to the multivendor data, and it's already done.
I have already tried it. It is pretty good. I think it helps more in the context of backtesting with good quality. If I were to create a data set and use it to create signals, the function to create an alarm would still be missing.
Q-Data in particular is so important because it is used with different tools. One can use it explicitly for backtesting and in strategies. But most importantly, it is used for the strategy screener.
In my opinion, Wealthdata and Q-Data play a particularly important role in the product. For this reason, it is precisely this data that should take the described process into account. It should be the reference data to minimise "gross" errors in trading.
Q-Data in particular is so important because it is used with different tools. One can use it explicitly for backtesting and in strategies. But most importantly, it is used for the strategy screener.
In my opinion, Wealthdata and Q-Data play a particularly important role in the product. For this reason, it is precisely this data that should take the described process into account. It should be the reference data to minimise "gross" errors in trading.
@Eugene
I agree. The most important part is the " Alerts" function.
Just keep in mind that the quality of the signals for different tools should be equal.
I mean using Q-Data in the strategy context needs to be of the same quality as in the screener.
The question is if you do it behind the scenes or on the user side. I described the solution to this point already. There is no need to change the software for the user. You can do the verifaction for WD+QD behind the scenes. The user simply gets better quality data for WD+QD.
@Eugene+alkimist
Using combined data created with Dr Koch's tool or an unlinked data set does not help with the problem. There are a few reasons, but one is simply that this data is not used for the strategy screener.
I agree. The most important part is the " Alerts" function.
Just keep in mind that the quality of the signals for different tools should be equal.
I mean using Q-Data in the strategy context needs to be of the same quality as in the screener.
The question is if you do it behind the scenes or on the user side. I described the solution to this point already. There is no need to change the software for the user. You can do the verifaction for WD+QD behind the scenes. The user simply gets better quality data for WD+QD.
@Eugene+alkimist
Using combined data created with Dr Koch's tool or an unlinked data set does not help with the problem. There are a few reasons, but one is simply that this data is not used for the strategy screener.
We already put in substantial effort in maintaining the limited symbol selection in WealthData. The dilemma here is that if we choose to invest similar effort into QData, which is now fairly automated, we wouldn't be able to offer it for free. After all, we don't have unlimited dev resources. However, we are going to spend some effort to beef up QData and will correct the current error(s) over the weekend.
Your Response
Post
Edit Post
Login is required