Thinkorswim errors in historical data

E

Eliot Hosewater

I've been downloading historical options data from TOS thinkBack. First of all it's tedious, you have to download one day at a time. Then, when running my backtest I was getting some strange results. It turns out there are quite a few errors in the data.

For example in the SPY:

A few days in (IIRC) January 2008 where some of the near expirations are just plain missing. I think they were non-trading days, but my simple minded program doesn't check for those.

From 2005 to present there were more than 1200 entries with positive theta and some of those were like +97 etc.

Some errors in the high,low on the underlying, like +1120 (I think the max high for SPY was 157). Related were some where the last price was 0 so the net change for the day was way off.

I think there a couple more that I found that I can't recall at this point.
 
These are two of my posts from a previous thread:

Thinkorswim’s thinkback option data is a complete joke. There are so many days missing, data duplicated over multiple days by mistake, and just a lot of random strikes missing from random days it makes it completely useless for anyone serious about back testing options. If you are just doing a few symbols you can purchase the actual EOD data (not thinkorswim’s embarrassingly corrupt data) for just a few hundred dollars.

Look at SPX for 9-23-03, 7-5-05 through 7-13-05, and 12-27-05 the data is missing, they don’t have it. Look at the top of the screen it says “No Data” and they just fill in the previous day’s data so if you aren’t paying attention you will use fake data and not realize it.
And on random days like 6-24-04 they are missing a whole bunch of the strikes.

->so you are correct that they have many errors. They know about it and don’t care, basically they just took a corrupt file and uploaded it or corrupted it in the upload, but either way they basically just tell you that they are providing it for free so don’t complain that it has errors. You can deal with all the errors or pay a little bit to someone like the CBOE for decent EOD data.
 
Wish I had seen your earlier posts.

That means thinkBack itself is pretty useless since it is based on the bad data.

Edit: They DO fill in from previous days on non-trading days, which I was aware of and my program depends on.

It's "free" (except for the hours spent downloading by hand), so I guess you get what you pay for.
 
Think or swim has a lot of features, but since their data is so bad, they're mostly useless. It's really designed for new traders to get their feet wet with all the features.

You'll really need to look at something like E signal.
 
Quote from Eliot Hosewater:

That means thinkBack itself is pretty useless since it is based on the bad data. They DO fill in from previous days on non-trading days, which I was aware of and my program depends on.

If they're providing bad data then you have to trade badly in order to get good results

It's "free" (except for the hours spent downloading by hand), so I guess you get what you pay for.

You pay peanuts, you get monkeys and a dumbass face like this --> :D
More...
 
If they're providing bad data then you have to trade badly in order to get good results

But all the data isn't bad, so you have to know when to trade badly. --> :p

One nice thing is you can get the greeks with TOS, which I use to estimate if I would have gotten stopped out, or gapped out. As long as most of the numbers are good then it's not so bad. In fact, the errors mostly caused my tests to get stopped out when they otherwise wouldn't have been.
 
I think you can call E signal and they'll tell you all about it. You may be able to get what you want with just the on demand, which is a lot cheaper than the full package. Every thing except the exchange fees will be refunded if you cancel before the trial ends. They're probably your best bet.
 
Quote from Eliot Hosewater:

>> If they're providing bad data then you have to trade badly in order to get good results<<

But all the data isn't bad, so you have to know when to trade badly. --> :p

One nice thing is you can get the greeks with TOS, which I use to estimate if I would have gotten stopped out, or gapped out. As long as most of the numbers are good then it's not so bad. In fact, the errors mostly caused my tests to get stopped out when they otherwise wouldn't have been.
More...
My 2 cents is that if the data isn't reliable then the backtesting isn't. If the data is not so bad then it's really not so good.
 
You must log in or register to reply here.
Back
Top