Hey guys. Curious, where do you guys get your data from for backtesting? I'm aware of https://www.historicaloptiondata.com (though im not sure if they are any good). Are there good subscription services out there similar to CSI Data that provide daily historical options data I can download? I searched the forum. There were 4,000 results...so I figured I'd ask. Thanks!
I'm a subscriber to historicaloptiondata since about two years (EOD updates) and bought their historical data and can tell you that I'm satisfied. Their data isn't perfect so it helps if you're able to clean it up a little, and they don't have corporate actions (only splits and dividends), which sometimes causes weird situations like one I've discussed here, where some stock symbol was aquired and changed name clashing with an older one.. weird shit... but out of almost 10,000 underliers there still are lot to choose from after you discard the ones with errors that can't be corrected. So when you ask "are they any good", I guess it depends both on how fastidious you are (but also mind you're not paying Reuters/Bloomberg feed prices) and how skilled you are in ignoring or making up for incorrect data. Having worked in finance / IT in general, what I can tell you is that there is no perfect data (or if it is, you have to pay a heckton for it). Again I'm saying I'm satisfied with the data from www.historicaloptiondata.com and also the guys running it are helpful and responsive. That's a bit surprising from my view of the world since when I worked in finance (billion dollar private fund), the last thing you'd do was try get Reuters/Bloomberg for support. Now I'm not working in finance no more but my employer still spends some 70% of their income on just buying the data (millions of $$$ that is) and I still cringe if/when I have to send a mail to the provider's support. Just getting an answer may take weeks, let alone an intelligent one (plus solving our problem). So yeah, maybe I have low standards but I'm satisfied with historicaloptiondata.
Well, my experience with historicaloptiondata.com is so-so. They were responsive to sales emails, but ignored my emails about how to handle stock splits. They don't provide option symbol list before/after stock splits, while options are renamed after each split. Those options symbols are then mixed up in their data and there is just no way to match up pre-split and post-split symbols, or even get reliable pricing for some days around the split. Even when you figure out the pre-split symbols, you'll then find pre-split and post-split pricing mixed up for the same symbols. It's a mess and I still have to spend a lot of time to unwind it. But if you don't need to handle splits (or basically don't need full history of stocks through their splits) then the data is quite good. The only problem is that you cannot know the actual pricing of options until you place orders and get fills. Many option prices have wide bid/asks and are skewed in a way that you cannot determine the mid price. For example $0.05 bid and $0.50 ask, while you may not be able to sell it for more than 0.10. The last trade price may be from a week ago. But this is not a problem with historicaloptiondata.com, just with usability of option chain snapshots.
>> For example $0.05 bid and $0.50 ask, while you may not be able to sell it for more than 0.10 Just subscribe to real time actual marketdata then within a limit of 100 symbols for $10. Last time I checked, the guys from historical option data gave you 10,000x more data for less than 10x more $$$.
I do get real-time options data and save it to my own database. I was just providing feedback on historicaloptoionsdata. I bought their stuff because it’s cheap and because it may give me more history than I have. Just not sure I’ll ever use it, so it’s like buying discounted stuff just because it’s cheap. And even real time data isn’t that useful without actually placing orders and testing fills.
For whatever it is worth, I have reviewed a fair number of tear sheets from both hedge funds and CTAs that trade options and there tends to be a material difference between from backtesting and live trading. More so than equities and future. Sometimes from +50% per year to fighting to breakeven. Making assumptions from past Bid/Ask spreads, last sale or end of day value, will not match real trading, or come close, in my experiance. I find real trading with smaller lots a better judge.
You might want to think about creating your own theoretical options data. A while ago I put a lot of work and effort into looking at Black Scholes and various other option pricing models. Given that you can pay a small fortune for each option series you purchase from the CBOE, it occurred to me that you may as well back test on theoretical option prices that you generate yourself. The Jupyter Notebook set our below (and hosted on Jupyter NB Viewer where you will be able to see it more clearly) was the result of my efforts, with which I was fairly pleased. I then realized that I had failed to estimate the "Volatility Smile" whereby options further out of the money or in the money tend to have an implied volatility slightly different from ATM options in the real market place. At some stage I will revert to the notebook and attempt to model the volatility smile but at this stage I thought I may as well release the Notebook to those (few!) who may have interest. I have not looked at this code for quite some time. Caveat Emptor, make of it what you will. If you don't code, or are not familiar with Python and its libraries, "move on and ignore this post" would be my advice!
By the way, my gist has a number of options related Jupyter Notebooks, including ones designed to cope with the extremely difficult task of reading cboe options Data, cleaning it, and making concatenated series of options prices given a choice of expiry and strike. By way of example: