8

I bought quantquote.com historical data of AAPL on second basis.

To comapre I also got activetick.com

For activetick I used the historical trading API.

If you look at around 15:13:53 you see that ActiveTick reports Quotes where QuantCode claims non exist in specific second.

Does QuantQuote only consider quotes form certain exchanges?

ActiveTick (quotes) Datetime,Price,Size,Exchange,Cond1,Cond2,Cond3,Cond4 2014-07-21 15:13:40.134,94.045,1100,Q,37,0,14,0 2014-07-21 15:13:41.003,94.047,3,D,0,0,0,37 2014-07-21 15:13:43.429,94.05,100,D,0,0,0,0 2014-07-21 15:13:43.472,94.045,900,D,0,0,0,0 2014-07-21 15:13:47.612,94.0401,50,D,0,0,0,37 2014-07-21 15:13:47.869,94.04,12,D,0,0,0,37 2014-07-21 15:13:48.508,94.0401,21,D,0,0,0,37 2014-07-21 15:13:48.648,94.0449,159,D,0,0,0,0 2014-07-21 15:13:49.291,94.0497,100,D,0,0,0,0 2014-07-21 15:13:50.350,94.047,105,D,0,0,0,0 2014-07-21 15:13:51.449,94.05,100,Y,0,0,0,0 2014-07-21 15:13:51.450,94.05,95,J,0,0,0,37 2014-07-21 15:13:51.450,94.05,100,J,0,0,0,0 2014-07-21 15:13:51.450,94.045,100,Z,0,0,0,0 2014-07-21 15:13:51.450,94.045,500,Z,0,0,0,0 2014-07-21 15:13:51.450,94.05,100,Z,0,0,0,0 2014-07-21 15:13:51.623,94.05,400,D,0,0,0,0 2014-07-21 15:13:51.623,94.05,200,D,0,0,0,0 2014-07-21 15:13:51.623,94.05,200,D,0,0,0,0 2014-07-21 15:13:52.683,94.045,200,Z,0,0,14,0 2014-07-21 15:13:54.319,94.0499,50,D,0,0,0,37 2014-07-21 15:13:55.278,94.045,200,Z,0,0,0,0 2014-07-21 15:13:58.186,94.05,20,Z,0,0,0,37 2014-07-21 15:13:58.212,94.05,100,D,0,0,0,0 2014-07-21 15:13:58.346,94.0499,20,D,0,0,0,37

QuantQuote (seconds) datetime_unix, datetime,time,usd_open,usd_high,usd_low,usd_close,volume 1405969978 2014-07-21 15:12:58 33178000 94.63 94.6394.63 94.63 55 1405969988 2014-07-21 15:13:08 33188000 94.62 94.6294.62 94.621000 1405969993 2014-07-21 15:13:13 33193000 94.62 94.6294.62 94.62 300 1405969994 2014-07-21 15:13:14 33194000 94.62 94.6294.62 94.62 200 1405969998 2014-07-21 15:13:18 33198000 94.62 94.6294.62 94.621919 1405970000 2014-07-21 15:13:20 33200000 94.62 94.6294.62 94.62 100 1405970006 2014-07-21 15:13:26 33206000 94.63 94.6394.63 94.63 25 1405970010 2014-07-21 15:13:30 33210000 94.62 94.6294.62 94.622000 1405970016 2014-07-21 15:13:36 33216000 94.62 94.6294.62 94.622400 1405970017 2014-07-21 15:13:37 33217000 94.62 94.6294.61 94.61 100 1405970023 2014-07-21 15:13:43 33223000 94.63 94.6494.62 94.642100 1405970029 2014-07-21 15:13:49 33229000 94.63 94.6394.63 94.63 100 1405970030 2014-07-21 15:13:50 33230000 94.64 94.6494.63 94.631900 1405970033 2014-07-21 15:13:53 33233000 94.63 94.6394.63 94.63 100 1405970035 2014-07-21 15:13:55 33235000 94.63 94.6394.63 94.63 200 1405970037 2014-07-21 15:13:57 33237000 94.63 94.6394.63 94.634323 1405970039 2014-07-21 15:13:59 33239000 94.63 94.6394.63 94.63 400 1405970048 2014-07-21 15:14:08 33248000 94.63 94.6394.63 94.631000 1405970049 2014-07-21 15:14:09 33249000 94.63 94.6394.63 94.63 200 1405970061 2014-07-21 15:14:21 33261000 94.63 94.6394.63 94.63 200 1405970062 2014-07-21 15:14:22 33262000 94.62 94.6294.62 94.62 100 1405970072 2014-07-21 15:14:32 33272000 94.63 94.6394.63 94.63 100 1405970073 2014-07-21 15:14:33 33273000 94.63 94.6394.63 94.633176 1405970106 2014-07-21 15:15:06 33306000 94.64 94.6494.64 94.64 265 1405970112 2014-07-21 15:15:12 33312000 94.65 94.6594.65 94.65 100 1405970120 2014-07-21 15:15:20 33320000 94.65 94.6594.65 94.65 200 1405970133 2014-07-21 15:15:33 33333000 94.66 94.6694.66 94.663339 1405970134 2014-07-21 15:15:34 33334000 94.66 94.6694.66 94.66 400 1405970135 2014-07-21 15:15:35 33335000 94.66 94.6694.66 94.66 200 1405970142 2014-07-21 15:15:42 33342000 94.66 94.6694.66 94.66 296 1405970146 2014-07-21 15:15:46 33346000 94.66 94.6794.66 94.67 400 EDIT: ActiveTick includes the following exchanges in the data: exchange value ExchangeAMEX A ExchangeNasdaqOmxBx B ExchangeNationalStockExchange C ExchangeFinraAdf D ExchangeCQS E ExchangeForex F ExchangeInternationalSecuritiesExchange I ExchangeEdgaExchange J ExchangeEdgxExchange K ExchangeChicagoStockExchange M ExchangeNyseEuronext N ExchangeNyseArcaExchange P ExchangeNasdaqOmx Q ExchangeCTS S ExchangeCTANasdaqOMX T ExchangeOTCBB U ExchangeNNOTC u ExchangeChicagoBoardOptionsExchange W ExchangeNasdaqOmxPhlx X ExchangeBatsYExchange Y ExchangeBatsExchange Z ExchangeCanadaToronto T ExchangeCanadaVenture V ExchangeComposite

user670186
  • 189
  • 1
  • 5
  • It appears you are not the only one having issues with QuantQuote data. http://michaels-musings.com/sources-intraday-historical-stock-data.html#comment-786 From another comment (#comment-690), QuantQuote has two sets of data: a clean set and a raw set. Do you know which set you received? This is a disgrace. A prestigious school like Caltech being used to exploit people with fake, commerical reviews. I bet Caltech doesn't even know this is going on. This really needs to be reported to their ethics counsel. Obvious conflict of interest. –  Jul 14 '15 at 01:40
  • 1
    See my answer below (and accept it since it is correct). Nothing wrong with the QuantQuote dataset, user670186 simply failed to do the time conversion properly. Typically newbie coding mistake. – user788171 Sep 25 '15 at 06:28

8 Answers8

9

Let me guess, you fell for one of the fake Quantquote reviews and decided to purchase their buggy data?

The reason for the missing quotes is Quantquote data is more of a snap-shot of market activity. It will not record every quote the way TickData or CQG does. ActiveTick is not as expansive as TickData but is more comprehensive than Quantquote.

Maybe this will help on their capture process: https://www.bigmiketrading.com/421509-post14.html

I wouldn't recommend QuantQuote for your purposes. They only provide the historical/live consolidated feed and IQFeed is better priced for those features.

It's not a seasoned group. I happen to know one of the main persons (a 25 year old physics major out of Caltech) behind QuantQuote through the CERN network and he is currently focused on crowdfunding a secure email product rather than QQ. From what I know of their credentials, they don't have a strong development team. Also, If I'm not wrong, QuantQuote is associated with Sam Barnett's (Sam Barnett SBB Research Group - Business Insider) group - I don't know for sure but I'm guessing that they are just selling a capture of SBB Research's live data.

In a nutshell, Garbage In Garbage Out.

Community
  • 1
DonaldRC
  • 91
  • 1
  • 1
    thanks for that interesting comment. Actually read on the Caltech website that QuantQuote was supposed to be one of the best quality data sources. I was not aware, that they are actually also affiliated with it. So it seems one cannot even trust .edu sources! http://quant.caltech.edu/historical-stock-data.html – user670186 Jul 12 '15 at 01:12
  • So could I just take the ActiveTick data to build my model and be on the save side? (would take some effort collecting the historic data, but should be fine) – user670186 Jul 12 '15 at 01:13
  • The answer depends on the precision your simulation requires. One solution is to run your simulation on both data sets and note any outliers, If you have actual real-time trades, you can compare the simulated results with the actual trades. The other method is to compare the second data with the tick / time and sales data for the session. Again, it all depends on the precision your test requires. Also, you may be able to salvage the data if you can work on a higher frame of time (1 sec to x seconds, or even sec to min). The higher frame of time should filter / smooth the data inconsistencies. – DonaldRC Jul 14 '15 at 16:31
8

This is an old post, but I thought I would offer the following facts:

1) QQ claims to be sited in the Empire State Building, Suite 2100. (https://quantquote.com/contact.php) That is false. They do not lease or sublease that address.

2) The VM message on their sales line is a standard carrier-provided one such as a private individual would have. It does not mention the company name, the name of an employee, the hours of operation, or anything else you would expect from a corporation.

I believe that QQ is a fraud.

Here's how I got the information about their address:

*) I intended to buy QQ's data, so I contacted them asking for the specific custom package I wanted, and also asking why their Volume column has decimals.

*) I got no response to phone or email.

*) I called again today and noticed how unprofessional their sales line's VM was. That made me curious. That was when I noticed they were in the ESB.

*) Having gotten suspicious, I looked up their address (the Empire State Building, Suite 2100). I then called the Leasing Office for the ESB (http://www.empirestaterealtytrust.com/properties/office/empire-state-building1) and went down the list calling each agent and asking if they would confirm that a company named "QuantQuote Research" rented that suite. Some of the agents weren't in or said that they couldn't give out information about clients. I finally found one whose secretary looked up the information for me and said that no, someone else rents that suite. She even called over to ask security at the ESB proper to please check that the tenant did not have any subleases outstanding. They did not.

I am not giving out the name of the person who gave me this information because I don't want to potentially get her in trouble, but you can definitely do the same as I did to confirm it.

This is not definitive proof, but I find it suspicious enough that I'm not going to do business with QQ.

David K. Storrs
  • 81
  • 1
  • 1
6

I have used QuantQuote minute data previously and found it to be by far some of the best data I have used. It's interesting to observe that DonaldRC has just a single post and is pushing ActiveTick strongly.

Anyways, I came across this post while researching market data for my next project as I wanted to see if there's any new info about QuantQuote. After seeing this, I decided to email QuantQuote support to get their input into it. I got a comprehensive response which I think is worth sharing here.

Dear REDACTED,    

Thank you for contacting QuantQuote. We looked into the post that you linked 
us to and it is also quite mysterious to us. First of all, QuantQuote does
not provide data with Unix timestamps, so it is possible the data is not
actually ours.    

It is also possible that it is user error as in our database, the prices for
that time interval do not correspond to what that post displays. This is 
what we have from our minute resolution dataset (derived from the second 
dataset) which is quite different than the prices posted.     

20140721,1511,94.055,94.06,94.04,94.05,71806,1,0,0,123          
20140721,1512,94.045,94.05,94.02,94.02,49993,1,0,0,123
20140721,1513,94.022,94.05,94.022,94.0447,48435,1,0,0,123
20140721,1514,94.045,94.05,94.04,94.05,31260,1,0,0,123
20140721,1515,94.05,94.08,94.04,94.08,48934,1,0,0,123     

Our best guess is that this is in fact user error. In fact, if you look
earlier on that same day, we find the following prices which match what was
posted:     

20140721,911,94.63,94.65,94.63,94.65,2100,1,0,0,123
20140721,912,94.66,94.66,94.62,94.63,1720,1,0,0,123
20140721,913,94.62,94.64,94.61,94.63,17167,1,0,0,123
20140721,914,94.63,94.63,94.62,94.63,4776,1,0,0,123    

It seems the user accidentally introduced a 6 hour offset when processing
the data. Although our datasets are geared towards a sophisticated audience
(largely university and hedge fund researchers), sometimes we do get less
sophisticated customers who improperly process the data and posts like this   
appear. However, among experienced industry professionals, we continue to
have an excellent reputation. We regret that this customer did not reach out 
to us as we could have helped him easily fix his error.

Best Regards,  
QuantQuote Support

So, it seems user670186 did not write his code properly and instead of fixing his bug, he went ahead trashed the reputation of a good data provider.

user788171
  • 161
  • 1
  • 3
4

In some of their documents, QuantQuote say that they adjust the prices for dividends to provide price continuity. However, the formula they use is wrong. They use

$$\frac{price - dividend}{price}$$ as a factor to adjust prices backward when it should be $$\frac{price}{price + dividend}$$

Over time this adds to a meaningful difference.

I also would like to add that from my interactions with them they are unprofessional at best and scammers at worst. I purchased data that I received in the wrong time resolution and they have never answered any of my calls or email to get the data I purchased. I would stay away from them and find a better source.

JejeBelfort
  • 1,219
  • 11
  • 24
rxex
  • 151
  • 1
4

I want to share my recent experience with QuantQuote. Long story short: just don't buy anything from them.

I put in my purchase and shared my credit card. I ordered 10 years of SPY minute data. A few days later I received last months data of random stocks. After realizing that I received the completely incorrect data, I emailed and called just about everyone in the company... no response.

A week later, I realize that this company is fraud. I repurchased the data from a separate source. Do not make this mistake, please save your time and money by buying from a different source.

Liam Shalon
  • 41
  • 1
3

You won't get an exact answer for this I'm sure, all one can say is welcome to the world of finance and bad data.

All jokes aside there could be a number of reasons (one vendor missed a message, network lag, not getting quotes from a certain exchange exc).

pyCthon
  • 2,111
  • 2
  • 18
  • 39
  • 1
    well also it could be a timezone issue! Quantoquote says their timestamp is in Eastern. For ActiveTick I havent heard back from their support yet, but I also assumed when querying their API its in Eastern timezone. It could also be the case, that my server application submits the timezone of my country (Switzerland) and then I get other results...but they dont document that anywhere. – user670186 Jul 11 '15 at 12:55
  • 1
    also, these data are so completely off, it would be a joke to build a model on that!!!! – user670186 Jul 11 '15 at 12:55
  • 1
    @user670186 an yes that's a good one too!! Time zones! Also if the vendor is in Chicago and taking quotes from New York need too account for that as well! I'm sure there are more possible explanations we both missed as well. – pyCthon Jul 11 '15 at 15:20
  • i will post more here as soon as I have answered from both of the companies. ActiveTick Reports quotes from different exchanges and I also dont know which exchanges quantquode includes in the data. --- not sure how other people solve these problems, but insights would be greatly appreciated – user670186 Jul 11 '15 at 16:11
3

As other members have pointed out, the Quantquote data does have dubious origins and undisclosed conflicts of interest. The ownership of the company matches with certain individuals at Caltech, who also ran the Caltech quant finance group which is pretty much the only place that Quantquote is referenced. They also do not seem to be very active in improving on the business, as they are working on Protonmail at the moment, which may explain the various user reports of slow support.

madilyn
  • 5,240
  • 20
  • 39
2

@user788171: I checked the 20140721,911-914 prices above and they also don't match with TickData.com's minute quotes & trades history. TickData is the premier data feed used by the exchanges themselves to archive their history.

The QuantQuote volume appears larger than TickData and the prices offset. Maybe QuantQuote doesn't remove bad ticks expunged by the exchange.

Or, it could be that QuantQuote corrected their data since the original post and the poster just got a bad order. I wouldn't be surprised, it is not the first time they have done a bait and switch with their price data. As other posters have reported, QuantQuote has run multiple forks of their data in the past to conceal errors.

Bonerrr
  • 21
  • 1