Unreliable data

[pkfryer]

I've been fiddling with price and volume data-mined from yahoo finance who use reuters for there data. But when comparing the price and volume with advfn they slightly differ mainly on the open and close, but sometimes also on the low and high. But always on the volume!

Whats going on? Surely there should be only 1valid price/volume for a particular day, company and exchange!! Why the difference? Anyone got a clue?

 

[theknifemac]

I've been fiddling with price and volume data-mined from yahoo finance who use reuters for there data. But when comparing the price and volume with advfn they slightly differ mainly on the open and close, but sometimes also on the low and high. But always on the volume!

Whats going on? Surely there should be only 1valid price/volume for a particular day, company and exchange!! Why the difference? Anyone got a clue?

Hiya PK

I have been struggling with a similar problem comparing the UK EOD data from ESignal and MyTrack. I started out with the view that ESignal would be a lot better than MT but have found the reverse. I compared the data with Yahoo and with Bloomberg for PSON for the month of November and found that ESignal and Yahoo pretty much matched but MT and Bloomberg agreed. I was mainly looking at closing prices.

I then checked against the LSE web site and the FT newspaper and MT agreed with both. E.g. for last Thursday (2nd Dec) MT was showing a closing price for PSON of 619 and Esignal was showing 614.66. The FT and LSE website both has 619 for the closing price.

In the UK market, the FTSE100 constituents (and stocks which have been in the FTSE) trade on the SETS electronic platform. This is the electronic order driven market. Closing prices are determined by a closing auction from 4:30 to 4:35. If no price is determined in the auction then a VWAP is taken of the last X minutes, failing that the last automatic trade is used. The UK market is a bit strange in that people can trade off order book (i.e. directly with a marketmaker in size and these trades are reported in SETS but not as automatic trades (can't remember the SETS term for these). So trades are reported as automatic or not. Also worked principal agreements and other block trades can be reported late and often after the close. Some providers may simply be looking for the last trade in the day and using this as the close rather than the official close reported by the LSE from the auction.

Not sure if this is what is happening with my example. On the volume front, you can look at on order book (automatic trades) and off order book volume and the sum of both. This can account for differences in volume which are often seen in the UK. Also a lot will depend on how vendors keep track of volume through the day and how they handle cancellations and corrections - do they remove them from the running total ?

For non SETS stocks, differences can arise from whether people use the last trade price or the mid price at the open / close etc. I am less au fait with this than SETS.

Hope this helps

Stew

 

[pkfryer]

Thanks, thats a great help.

So your saying that more sources agree: LSE,FT, MT and bloomberg than agree with yahoo and eSignal. I would have thought for a data feed dedicated company like eSignal they could have been trusted. advfn doesn't agree with yahoo either, so probably would join the other group. I wonder if there are several destinct groups.

There can't be that many price data sources, most of these companies must be getting there data from a big third party. Probably bloomberg or reuters. Maybe, its these two biggies that cause the difference with a slightly different set up on how they call the price and volume. Maybe there really isn't a black and white about the close and volume for the day and I should develop systems that are robust enough to handle a certain give in price levels.

The difference wasn't dramatic, less than a percent, it just concerned me that there would be any difference at all.

What you said could also explain why it is always volume and mostly close that differ. But I wonder why the open price would vary sometimes too.

What did you do in the end? Did you check out who provides your brokers data feed and stick with that?

One other thing that concerns me is that my broker is dedicated to CFDs and has out of hours trading on a lot of stock. These out of hours trading sessions could have dramatic impact on any systems I develop and could really be a pain in the ****.

I think its irritating that most brokers dont allow us to download there data for our own information so we know exactly what prices effect us. Do you know of any brokers that do this?

cheers

 

[theknifemac]

Don't know any brokers who let you download their prices. I think some might be too embarrassed if you highlighted differences with the "real" prices.

I am going with MyTrack for now as they agree with Bloomberg which I trust. Shame I don't have access to a Reuters terminal as I am sure I could work out where ESignal are getting their data from.

For CFDs I would prefer to use a broker who let you trade directly in the market so their prices were live market prices, rather than quotes from the broker. This would make backtesting a whole lot easier and less likely for you to suffer from broker manipulation etc. I think ETrade do this but then you pay a comm rather than other CFDs / spreadbets where it is all included in the price and spread.

HTH

Stew

 

[roguetrader]

I'm sure if you ask eSignal where their data comes from they would tell you, can't see why it would be a secret. Course I could be wrong

 

[theknifemac]

I'm sure if you ask eSignal where their data comes from they would tell you, can't see why it would be a secret. Course I could be wrong

I was meaning more what they are looking at for the closing price, whether they are picking up the last trade of the day rather than the trade marked as the official close... But will ask them on Monday.

Stew