Backtesting assistance required

[mariposa]

I have devised a strategy, and generated historical portfolios. I now want to backtest it, and I have started learning R but it is taking a while as I am learning R from scratch. I am looking for someone who can help me with this specific project, and I am prepared to pay an hourly or project rate. Familiarity with backtesting options strategies is required. Contact me at:
http://tinymailto.com/d5d4

 

[Dave Robinson]

A couple of questions:

1. How long are you holding your positions?
2. Do you have historical data?
3. If you have historical data, what time frame data do you have (end of day/EOD, 15 minute, etc)?

I've found that cleaning options data can be quite time consuming (lots of duplicates and bad prints). The time required for this step can vary quite a bit by vehicle and time frame of data. For example, I found that 20% of the 15-minute SPX data from one vendor required scrubbing, while ~2% of the EOD RUT data from another vendor required scrubbing.

my two cents.

 

[mariposa]

1. It is a low frequency strategy, i build one portfolio per week and hold the positions for 4 weeks then close out. (So at any time i will have four portfolios of positions)
2. I plan to buy historical data from historicaloptiondata.com
3. It is end of day data, which is ok for a LF strategy. Ideally i would enhance the backtest with intra-day data, but do not want to incur the additional costs right now until i get an idea how it looks using EOD data. If you can recommend a reasonably priced, clean intra daydata set I would appreciate it.

Thanks.

 

[Dave Robinson]

For indexes (RUT, SPX, NDX), and their corresponding ETFs (IWM, SPY, QQQ), I've used the following vendors for end of day option data:

1. iVolatility - http://www.ivolatility.com/
2. Historical Option Data - https://www.historicaloptiondata.com
3. LiveVol - http://www.livevol.com/

For these three vendors, I had to do the least scrubbing with iVolatility. I had to do the most work on the options data from LiveVol. Historical Option Data was somewhere in between these two in terms of cleaning.

What underlyings are you using?

 

[mariposa]

For indexes (RUT, SPX, NDX), and their corresponding ETFs (IWM, SPY, QQQ), I've used the following vendors for end of day option data:

1. iVolatility - http://www.ivolatility.com/
2. Historical Option Data - https://www.historicaloptiondata.com
3. LiveVol - http://www.livevol.com/

For these three vendors, I had to do the least scrubbing with iVolatility. I had to do the most work on the options data from LiveVol. Historical Option Data was somewhere in between these two in terms of cleaning.

What underlyings are you using?

SPX options and options on SP500 constituents.

 

[Dave Robinson]

I store my historical data in MySQL. To give you an idea of memory requirements for options data:

• iVolatility EOD SPY (Jan-2005 to Aug-2013) - 1.2 GB, 4,277,380 rows
• iVolatility EOD SPX (Jan-2006 to Dec-2013) - 1.0 GB, 3,588,858 rows
• LiveVol 15min SPX (Jan-2008 to Dec-2012) - 24.5 GB, 68,036,658 rows

The ETF HYG might be similar in data size, possibly representative, for the amount of data/rows for an SPX constituent:

• Hist. Opt. Data EOD HYG (Apr-2007 to Sep-2013) - 130 MB, 431,249 rows

Also, since there are more expirations now, there is more option data each month today than there was for each month five years ago. So, the storage requirements going forward will not be an average of my metrics above, but larger.

Hope this helps.