When it comes to earnings, what is the common method used to measure the analysts dispersion. Should you look at the dispersion relatively or vs other companies as a whole. Thanks
On my phone your thread title showed up as "How do you measure the dispersion of anal..." in the thread list. I know, totally juvenile, but made me laugh nonetheless.
Sig I'm pretty sure you would have experience with this strategy. Thoughts and opinions? There's a correlation with dispersion vs straddle price. But the analysts earnings forecast change all the time. I put a trade on IRBT because the: analyst high - analyst low / analyst mean was extremely high and the vol level of the straddle was relatively low vs actual move. 2 days later the estimates tightened so much, straddle priced dropped 20%......
@Secret Santa, this strategy might not be your forte but maybe you can provide some insight. Using the Bloomberg security monitor I scan for stocks whose 1 day implied earnings move is much lower than the average 1 day implied earnings move. I then check to see how disperse the analysts estimates are using the formula (analyst high -low)/mean. If it is greater than 20 % I buy the straddle. I turn the position neutral every 100 deltas per 10 lot (using my specified vol, not implied). If the ambient vol is greater than my forecast vol I will buy a calendar instead. ( trades are usually done 15 days until earnings). So far this strategy is my worst performer. I have been doing this specific strategy and my ROC annualized is -14%... I can provide all data for these specific trades if need be.
I will also add that the 1 day implied vol is using the standard formula with 2 expiration dates. If front month is less than 3 days until expiration the back month is used
I don't have experience with it but it does sound interesting. You may be able to code a backtest on this with Quantopian, I think they have earnings forecasts data you can access although maybe not by individual like you would need.
Just for the fact that stocks with higher analyst's coverage will have lower dispersion compared to lower coverage by analysts. Assuming you are doing systemically over a wide spectrum of stocks, in my opinion it should be taken into account. Anyway, I am off to long vacation in few minutes. Good luck and it is interesting project and look forward to see the results over a course of period
Thinking about this some more, I wonder if it might also be valuable to figure out somehow if the dispersion is caused by "stale" forecasts, i.e. your outliers might just be analysts who haven't updated their numbers yet? Not sure of a great way to do this, just throwing the idea out.