[Tasuki]

The paragraphs below come from a post I just made to the Tradestation forum. My question to you, my fellow Traders Lab folks, would be--how common is this sort of glaring data corruption? Does it happen with Ninjatrader? eSignal? If so, or if not, please let me know.

Here's a copy of my post to the Tradestation forum which is at this link:
https://www.tradestation.com/Discuss...Topic_ID=98375
....
Please see attached chart.

The UPPER chart of GLD is the one that's been on my workspace for years, and I've never questioned its accuracy until today, when I compared the daily volume (chart not shown here) with the volume reported on the 5 minute chart. I started mentally adding up the volume bars on the five minute chart and realized that the total did not come even close to the volume shown on the daily chart. So, I called Tradestation, and we figured out that the volume on the five minute chart was way too low. So, I created a new 5 minute chart on the same workspace (which is the LOWER chart of GLD shown here) and the volume on that new chart was correct, according to the Tradestation tech I spoke to.

Question: how can I tell if a chart has become corrupted? For heaven sake, I haven't got time to go through every chart every day and question whether Tradestation has screwed something up and corrupted its data. Isn't there some way for the Tradestation platform to flag a chart that has potentially become corrupted??????? If there isn't any way to do this, then there sure as heck should be.

Tasuki

[Tasuki]

Turns out that my concern was unfounded, thank goodness. It just makes me look like an idiot. Apparently the chart properties got switched somehow so that the volume was measuring tick count instead of trade volume.

Do the following excercise if you have time. Go to CME website and check daily volumes for a number of instruments.

Then compare the volume in TS on daily bars. Quite a number of data feed providers whose data have filtered ticks fail to report volume correctly, e.g. Transact, OEC, etc.

[Tasuki]

Tresor, great idea. Here's an interesting question---does it really matter? What I mean is, if the data providers whose numbers are wrong still provide numbers that are proportional to the true numbers, maybe that's good enough. Think about VSA or Wyckoff analysis---they don't need to know the actual volume, they need the relative volume---today's volume vs the volume two days ago, for example, So (I'm just hypothesizing here) maybe polled or filtered data is good enough, if the filtering is done proportionally. The real problem would be a filtering process that was more aggressive when volume was high or when volume was low. THAT would create a serious problem for the trader, methinks.

Originally Posted by Tasuki »

Here's an interesting question---does it really matter? What I mean is, if the data providers whose numbers are wrong still provide numbers that are proportional to the true numbers, maybe that's good enough.

Let me give you an example. I checked Transact once; their volume used to be wrong by as much as 30% / 130% / 230% in relation to official volume reported by the exchange. You never knew if one day their volume for FDAX will be 30% of the official volume or 130% of the official volume. The more volatile day the more incorrect the data was for a given instrument. Because of my complaint MC cut off Transact as a supported broker.

I also checked OEC forum for developers. OEC IT guys simply do not know how to fix their data feed with regard to volume. And you will find a guy in this forum who keeps on convincing ignorant newbies that OEC data feed is free (true) and very good (very untrue).

And funny thing is: a number of different gurus over the internet are posting volume related strategies and base their screenshots and movies of trades on such incorrect data feeds.

Either their strategies are wrong, either they are just lucky or they only record a minority of their trades (the profitable ones).

Check this site for CME official volume: Volume & Open Interest

[Tasuki]

Originally Posted by Tresor »

You never knew if one day their volume for FDAX will be 30% of the official volume or 130% of the official volume. The more volatile day the more incorrect the data was for a given instrument.
Check this site for CME official volume: Volume & Open Interest

Tresor, now this is interesting, and significant. If they were always off by exactly the same percentage, that might be OK, but if the percentages change, that's not good at all. It's for reasons like this that I've asked TL's founder, Soultrader, if he could create a forum specifically for discussing data reliability issues.

I just checked Tradestation's data (see attached chart) and bless Bess, they got it right.

Tasuki

p.s. thanks for the link. Very useful.

[daedalus]

lol... i've done the exact same thing myself before... tick/volume chart settings are easy to overlook.

But like always, when in doubt - CTRL+R.

Sometimes the data will go corrupt on your machine and you have to manually delete the cache as well and restart the program to have a fresh new download of the data to your machine but that only happened to me on rare occasions.

Cheers!

Originally Posted by Tasuki »

I just checked Tradestation's data (see attached chart) and bless Bess, they got it right.

You may want to do the last test. It's easy and simple and will give you a definite answer about the quality of volume reported by TS.

Make a chart (daily resolution) of any of these Eurex instruments: FESX, FDAX, FSMI, FGBS, FGBM, FGBL. Then go to this website Eurex - Europe's Global Financial Marketplace which displays 15 minute delayed quotes (with volume under ''contracts'') for these instruments.

Or make a chart of 1H resolution and then add hourly volumes in Excel and repeat the above exercise.

If the volume reported by this website matches (with 15 minute tolerance) the volume on your chart then you should be pretty sure that TS is doing their job right.