# 7 – M2L2 07 Handling Outliers In Signal Returns V4

After you’ve identified anomalous skew in your signal returns and you think you know the cause, what do you do? The best strategy for dealing with these types of outliers varies on a case-by-case basis. If you know it’s just bad data from the data vendor, you can try to fix it by replacing the affected data with correct values from a different vendor. If you can’t do this, you might try to determine if your research result will be greatly affected if you replace the value with any reasonable value. For example, the previous day’s values. If this isn’t going to work, you might try to figure out whether removing this data row, all together, will have any significant effect on your signal research results. You are the one in the best position to make this judgment call. After all, you know the signal the best. If you think the extreme values are due to legit market events, there are a few different things you should think about. You might try to find out whether there are many similar occurrences in other stocks. If there are and if you can identify a common cause for all the extreme values, you could consider isolating this group of stocks out of the research stock universe. You should think about what effect this might have on your signal overall. Here, we’re showing several thinly traded small market cap Asian stocks experiencing large sudden unpredictable price fluctuations. Another type of stock that experiences dramatic unpredictable price fluctuations is the stock of single drug biotechnology companies. These type of companies evaluation hinges upon the success of various milestones: phase one,two, and three drug efficacy trials and FDA approval. With only one product, the company is worth a lot if it is successful and very little if it fails. However, the outcome of these binary events is very difficult to predict and to handle. So, quants frequently exclude these stocks from their trading universes. As one example, the price of the stock of Sage Theraputics, a company that makes a postpartum depression drug, jumped from $93 to$167 in a single day last year when the company announced positive test results. You should also think about whether events you think are legitimate are due to special market events like corporate earnings or central bank announcement’s. Realistically, if you eventually use this signal to build a trading strategy, you might want to think about whether you can avoid these events by, for example, pausing the strategy in advance of them. If it’s possible to do this, then you can probably remove these rows and proceed with the research. If you can’t pause the strategy ahead of these events, then you’re going to need to think about how you will be able to avoid losing money when executing this strategy. As you can see, a lot of this is common sense. You’ll always need to think about how you can prevent the quality of your research results being degraded by outliers and how you can avoid losing money when trading your final strategy.