Unusual Whales: Developing a Trading Strategy, Part 2

welcome to my second video in my series of videos about unusual wells if you haven't seen the first video i'll put a card on the screen for you right now last time i tested whether there were significant differences between the mean averages of the winning and losing alerts today i want to see if we can use those variables to find winning trades and increase our win rate beyond the unusual whales benchmark first let's define that benchmark in my first video i used data from december 2020 to march 2021 but for this video i'm going to expand the data set to include may 2021 as with my prior video i'm going to be showing a variety of wind definitions but i'm going to be focusing on the 25 wind definition for the rest of this video here's the performance of unusual whales without any filters applied as you can see there were a total of 9765 call alerts and 8 419 put alerts in our time period for the 25 win definition the win rate for calls is 59.8 percent and the win rate for puts it's 55.7 percent this will be our benchmark to beat as we do a deep dive into our variables i want to see if any of the variables have certain ranges where the win rate is greater than the benchmark for each of the 12 variables i'm going to sort the list by that variable and then i'm going to group the alerts into deciles using puts as an example we have 8419 put alerts in the time period so each decile will have 842 alerts except for the 10th decile which will only have 841. next i will calculate the win rate of that decile so we can start doing comparisons here are the results and i highly encourage you to open the spreadsheet linked in the description to follow along as you can see all 12 variables are split into deciles with decile 1 being the decile with the smallest values of that variable and decile 10 being the decile with the largest values of that variable the next column shows the number of alerts in that decile then we get to the min and max columns which show the smallest and largest value in that decile respectively the thing to note is that the values are inclusive and there is some overlap between the deciles this is a bit of a weakness because alerts with identical variable values sometimes get arbitrarily split into different deciles but i wanted to keep the decile sizes roughly the same the next columns are the wind definitions and the win rates for that definition next we have the average high return for all alerts in that decile and finally i have calculated the average amount of days it took for winners in the 25 win definition to reach their high return and the standard deviation of that average as an example this is how i would interpret the first row of the results for puts for the ask variable decile 1 had 842 alerts the minimum value of the decile was around 2 cents and the maximum value was 27 cents if we set our win definition to 25 percent then 55.7 percent of the alerts in this decile were winners the average high return of all alerts in this decile was 122.3 percent of the winning alerts that met the 25 win definition it took an average of around seven and a half days to reach their high with a standard deviation of around seven days [Music] here are some of my general observations lower buy amounts have a higher win rate this is counterintuitive as higher buy amounts are supposed to mean more conviction in the trade there are so many whale hunting strategies out there right now that tell you to focus on the highest volumes and the largest buy amount but that might not be the best course of action with an exception for the first decile of ask and bid smaller ask and bid values seem to have a higher win rate perhaps this effect is similar to the small cap effect in stocks other than buy amount ask and bid i wouldn't really feel comfortable applying a larger is better or smaller as better rule to any of the other variables for most of the others there seems to be a sweet spot or multiple sweet spots that have a higher win rate many of the whale hunting strategies that i see out there rely on a simple larger is better or smaller is better rule for some of their variables but judging from the data i think it would pay to adopt a more nuanced approach to developing your strategy [Music] now that we have the win rate by decile for all of our variables it's time to develop our strategy obviously this is a personal choice but i'll just share some ideas using calls as an example we could go back to the first video where we ranked our variables by significance and develop a strategy around the most reliably significant variable for calls it was by amount which was significant for every single win definition from the win rate spreadsheet it looks like decile 1 had the highest win rate of 65.9 percent so we can filter our alerts for buy amounts between thousand two hundred eighty four and twenty nine thousand four hundred and eighty if we're more active trader we could expand our filter to decile two which had a similar win rate by increasing the upper bound to fifty 53 650. we could also just choose a variable based on personal preference for example i always have tons of anxiety when buying weeklies so maybe i'll develop a strategy around the expires in variable i can see that decile 6 has the highest return so i will filter my alerts for options with between 30 to 39 days to expiration that way none of my alerts will be for weeklies so far i have been focused on the 25 win definition which is my personal preference but maybe you want to go for 100 return on your trades to maximize your win rate it might be better for you to go for decile 3 which is between 8 to 14 days to expiration we could also develop a strategy based around restrictions on our portfolio let's say you have a small portfolio but you still want to diversify your options trades we would want to use the ask variable to keep our individual traits small but in this case we can see that decile 1 the lowest asked performed worse than benchmark it would probably be best if we focused on decile 2 which has the highest win rate but still keeps our trades between 25 and 43 dollars for me personally i don't have any strong preferences or restrictions so i'm just going to sort the deciles from highest to lowest win rate and i'm going to go with gamma which is a significant variable and i'm going to filter for alerts between 0.07 and 0.09 gamma the expected win rate would be around sixty six point five percent which is six point seven percent higher than the unusual whales benchmark similarly for puts i will sort by highest win rate and choose expires in between 23 and 32 days with a win rate of 65.2 percent which is 9.5 percent higher than benchmark as you can see we are well on our way to choosing filters that will beat the unusual whales benchmark and i sincerely hope that this analysis and the win rate spreadsheet will help you in developing your own strategy [Music] in addition to my win rate spreadsheet shared in the description i'm also going to make available my consolidated data file it consolidates all months from december 2020 to may 2021 into one file and adds columns for volume open interest ratio expires in days to high and adjusted high return for a full explanation of how i prepare my data please see the first video in this series i'm also going to share the same file with winner columns that indicate whether that particular alert met the win definition the columns are labeled true if it does and false if it doesn't this file is split into one for calls and one for puts to make the size a bit more manageable finally i'm going to share a github repository of all the code that i used for this analysis but before i share it let me be clear that none of what i did requires any coding whatsoever the only reason i did the coding was as i was adding more and more months of data google sheets was starting to get slower and slower if you run into the same challenge you're welcome to use my code everything is done in damfo js which is a javascript library that's heavily inspired by the pandas library anyone familiar with the pandas api will easily be able to use dampo js so far i have focused on using a single variable but obviously we're not limited to just one up next i want to analyze whether different combinations of variables can produce even higher win rates and i'll show you my actual results over the last two months [Music] thank you to all the people who liked or commented on my last video and thank you to the people who emailed me from my site this is such a niche topic and i really wasn't expecting this many people to be interested but thank you everyone for the support and i really do hope these videos and these posts are helpful to you in some way you

Channel: Bear With Me

Views: 5,162

Rating: 4.9715304 out of 5

Keywords: unusual whales, unusualwhales, options, trading strategy

Id: wX4AC0X-hQs

Length: 10min 21sec (621 seconds)

Published: Sun Jul 04 2021