Please hold…

The more I trade my Alpha 2 system the more data issues I become aware of.  The system uses OHLC data in its inputs which can be trouble.  Different sources have significantly different quotes for some days.

I’m taking a short break from trading this system.  I’ll begin posting trades again when I’m more confident in the data being used for testing and live trading. 

Any suggestions for sources of clean OHLC stock data are appreciated.  So far I’ve compared data from PI Trading (constructed from 1-minute), Bloomberg, and Yahoo finance.

About these ads

Posted on February 12, 2013, in Alpha Strategy. Bookmark the permalink. 11 Comments.

  1. I have found by using just the H L C and ignoring the open, my systems work better.

  2. I use IB and find their data to be very reliable.

    I don’t have a subscription to Stockcharts.com, but I understand from those who do that their data is also reliable.

  3. I use premiumdata which provide daily data only and they are reliable.

  4. What types of errors are you seeing? Is it more common on high volatility days (when you need it the most, I would guess), or just random?

    I use a lot of Yahoo historical data, and there are many days with missing data. Very annoying. They are not the same for all stocks, but there seem to be some patterns. It also appears they are improving; maybe due to more competition? I haven’t tried comparing Yahoo to anyone else, since it is just incredibly time consuming just to get one provider’s data set. In any event, I try not to rely too much on any single day’s data in my work.

    • There are errors in the daily high and low values. It’s not just Yahoo Finance, but other sources as well. It seems to be somewhat random from what I see so far. It’s not always that bad but it can be.

      I can give a couple recent examples from the SPY ETF. There’s an error in the low value for 1/17/13 (~28 cents off) and in the high value for 2/4/13 (~69 cents off). Looking at the PI Trading data, there are a number of days in the past with a greater than $1 difference in a high or low value between it and Yahoo Finance/Bloomberg (some differences as high as $3).

      One free source that has correct values for the two above dates is freestockcharts.com. Not sure how accurate it is overall but it’s encouraging to see clean values for these dates on the charts. I don’t think you can download data from here though.

  5. Just to be sure. Are you using the right ticker? For SPY on Bloomberg you have SPY US (aggregate of all exchanges and not tradable), SPY UP (NYSE) and some more…Depending on which ticker you look at you will have some (slightly) different prices. What I would recommend is using the primary exchange which is SPY UP in this case.

    My $0.02

  6. High and Low for yahoo finance is not accurate. It makes backtesting strategies like fading the opening GAP way better than reality. This blog documents it a bit: http://www.quantifiedstrategies.com/the-importance-of-good-data-sets/. I was going to say go with IB but I have personally seen spurious quotes on my chart especially after hours.

  7. Has you model a stop loss? Eventually how much?

  8. I trade professionally and I have coined a phrased. Data is a Bit$h! I don’t care what data vendor you go with, you will have to keep an eye on your data. The data could be clean for months but one day it will bite you hard.

    We use IQFEED and pull the data into our trading spreadsheets using XLQ http://www.qmatix.com/XLQ.htm. I have no affiliation with this company. XLQ is a must have for traders who code up their trading systems into Excel for trading.

    Like you we use OHLC data. One trick that I learned is how to handle dividends on the day they are paid. For some strange reason, the %change calculation will correctly account for the dividend but the price will be wrong on dividend data. It gets cleaned up later but on dividend payment day, it will be wrong.

    We simply take (todays high – Y days close)/Y days close and compare that the %change value. If it is different, then you know a dividend is being paid. You then have to add that difference back to the Open, High, Low, and Close price because the price will not be adjusted for the dividend. I know this seems a little complicated but once you wrap your hands around the data, you will understand what I am talking about. I don’t how I could account for this if I wasn’t using Excel as my trading platform.

    Thanks for sharing your research. It makes me think about new ideas.

    John

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 137 other followers