Leaderboard

Check out RAD.... I put up a few posts ,suggesting IMO,there was an explosion pending....so far,it happened. It could be just the start of a major squeeze. I was talking about buying April $1.50 calls for .20 cents when the stock was around $1.65

In for $1.60. Out for $1.91, a gain of about 18% after commissions

I'm not suggesting anything is sleazy, I'm sure it's a great tool. I'm just trying to understand how it works and how it is applied. The sum total of my knowledge on this product is a brief fast paced webinar. I also understand the purpose of the webinar was to demo features, not be a training session on back testing methodology. So, I would like to understand how it is applied to real trading. I also acknowledge I have more background in other areas of investment and as a newbie I'm open to learning what does and does not apply to the world of options. So, a few options newbie questions: 1. A key principle of back-testing/fitting is the concept of out of sample testing: Pick a point in time, say 2010 and develop your rules/system using only the data that was available a that time (the sample data). Then, walk it forward from 2011 to present to see how it would have performed (the testing data). Not splitting the data in some fashion is almost a guarantee of fantastic results. So, does the software (or how it is generally applied) test in-sample or out-of-sample? In the webinar, I only saw in-sample which prompted the original post. (Systematic Trading by Robert Carver provides a great discussion of over fitting). 2. For a strategy to be considered robust, it generally should work across a markets and time frames. If I am developing a trend following system for futures and it only works for soybeans over the past nine months, I probably don't have much of a strategy. Unless, of course, there is a thesis to go with it: e.g. identification of something that changed in only the soybean market nine months ago. So, I would like to understand how looking at the behavior of an individual stock over a relatively short sample period translates to a robust enduring strategy. How do we know we are not being fooled be randomness? 3. In another post in this thread you stated a win rate as evidence the tool works. Great, but I would like to know more to really know the effectiveness of the tool. What is the win rate for the base strategies without application of the back tester? More importantly, how much did the back tester increase profit for winners and/or decrease the losses on the losers? 4. You also stated in another response that as a practical matter you can't look at "...even 20 occurrences". Why not? How can there be confidence in any back test with a tiny sample size. I would just like to understand this one. Thanks

"For example: if a stock moved after earnings less than the expected move in 8 out of 10 last cycles, there is a 80% chance that it might happen again the next cycle." Way too small of a data pool to have any validity. You would need, at a minimum ....ideally 1000 occurances, but at least 100 to make sense. 8 outcomes = 0 outcomes in statistics.

Just watched the recording of the webinar...the recorded portion seemed to me to show a lot of over fitting. How is looking back to find one (or a few stocks) where a strategy worked (and many did not) going to be of any value in live trading? Maybe I missed it but I did not see a thesis as to why something would work for FB but not AAPL. Find the one person in ten that flipped heads five times and bet on him???

I've been very interested in this post-earnings put spread idea since it was posted a while ago. So now that the margin calculation is fixed, I did a more extensive backtest. I screened for the 100 most liquid stocks 3 years ago (to avoid any survivorship bias), and tested the same trade on all 100 of them over a 3 year period. I sold a 50/20 delta put spread 2 days after earnings and closed 29 days after, using closest 30 DTE (in practice you get something between 30 and 60 DTE). No profit target or stop loss. The results are in the spreadsheet attached, which compares the strategy to buying and holding the same stocks (dividends were not accounted for. The results are really good. I'm tempted to believe there really is a persistent anomaly here, and it's not just a convenient uptrend magnified by options delta). Bear in mind that the average returns at the bottom are 3 year total returns. You'd have to calculate the CAGR to get an apples to apples comparison, and there are multiple ways to approach that, so I left it out. CML Post Earnings put spread backtest.xlsm