No. of Recommendations: 2
Jim,
Thanks for verifying, it's what I expected. Please don't take any of my comments/questions as criticism, it's just that your results are so beautiful that I want to understand every detail. In the same vein, how about data used in this one, how was the 1.5 picked?
And last step: take that WMA and multiply by 1.50. Why 1.50? I looked for the multiple that gave the best fit to price data. Also, are results e.g. the forward return bucket results, fairly robust to changing the WMA16? That's sort of an open-ended question, to be more specific in a worst case of sophistication, hopefully just a SMA still shows a significant trend?
A prior question to which Jim had replied is:
Also, do you think that it would help to go back to, say, 1995, just for perspective?Side comment:
As many have noticed, a plot of price to book (logged), or even just share price (logged), versus time shows a distinct change in slope at the end of 1998. The GenRe acquisition was, perhaps coincindentaly, at end of 1998.
@ Value Trend
Several year ago I looked at price versus sales, book value, operating income and net income for the companies in the S&P 500 from 1962 to 2017. Berkshire Hathaway's price versus book value had the highest r^2, 0.994, of any price versus metric of the S&P 500 companies.Wow, that analysis sounds extremely interesting! You don't happen to have any of the results of that lying around that you could post?
@Bluehorseshoe
I think I will try to keep updating periodically even if it is only for my own benefit :)We'd all benefit if you would so generously continue to share.
because historical options pricing is nearly impossible to findActually, it's not so hard anymore to find historical options data at a semi-reasonable price. But I've found that it's hard to find
good historical options data. Maybe data from ivolatility.com or other quite high priced shops are better. Data from, e.g.
https://historicaloptiondata.com, and even data from the CBOE "datashop" (just google it), is NBBO data i.e. "National Best Bid Offer Data". You can read some anecdotes about NBBO data in Flash Boys (I re-read it recently, and found the book pretty disturbing). I've seen weird stuff going in historical NBBO data, e.g. Bids greater than Ask, also NBBO values that are silly (and changing) values even when volume and open interest are zero. The last is understandable if one realizes the data is NBBO i.e. national best bid/offer and not actual transaction data. Bid/Offer data is whatever the market makers (now algorithms) decide to post, and they seem to post some weird stuff some not insignificant part of the time, well, assuming that the historical data is an accurate record which may or may not be a good assumption. In a backtest of e.g. SPX, one can start off with some quite reasonable option, say something not too far ITM or OTM and with good volume and open interest, and then follow it in the historical record, and find that it enters a period of zero volume. And NBBO values can get whacky. What do you do in a backtest? Perhaps say that if volume goes to zero then you hold (because if you don't, at what whacky historical Volume=0 data value do you close the trade?). But, if you do that, what if Volume=0 right through to expiration (which isn't just a one-off thing), how do you close out the back test? Perhaps in response to this, CBOE uses something called VWAP data (Volume Weighted Average Price data) for some of their indices that track option strategies, see
https://www.cboe.com/us/indices/benchmark_indices/ My understanding, which could be wrong, is that VWAP data is actual transaction data that's volume weighted over some time period, e.g. the half hour before noon. They use this apparently odd choice of time period to avoid whacky stuff that occurs at say the half hour before close. But when I asked CBOE if their VWAP data, which you can buy at greater cost from them than their NBBO data, would allow me to reproduce some of their option strategy backtests (
https://www.cboe.com/us/indices/benchmark_indices/) the answer was emphatically "No!". The reasons were never made clear to me, and resolving the issue of whether and where I can get
good historical options data will have to wait until I summon the energy to get back into the very gory details of such data. Meanwhile, I'm not sure I trust any of my option backtests, and by extension, currently am suspicious of much of what has been published in peer-reviewed business school academic papers on option backtests.
