fbpx

What is Standard Deviation? – Data Science for Finance

Standard deviation or t-value variance and a few other terms are all different ways of describing the spread of the data set or the very variable you’re looking at so if you have on the y-axis sales and the x-axis retargeting spend and when you look at how the retarding spend is influencing the various data points on sales if you have the data points very spread out it means you have a high standard deviation if they’re all concentrated along the line which is the multiplier line or the slope line then you have a very low standard deviation so the dots the data points are very centered around your description of that variable then you have a low standard deviation it’s okay to have a high standard deviation so if you look at a certain marketing category it has a very high multiplier or slope which is great we get $1 in seventy dollars out it then have it still has a very high standard deviation and that can be fine if you have a very long time period to run campaigns on because it’s okay that the fluctuates so you run a campaign it was very high it was fifty-three dollars back per dollar and then the next one is two dollars back per dollar that could be fine if you have a very long time here to do it because the average will be seventeen dollars back per dollar expense so a standard deviation can be okay but as a contrast if you are looking at your  beginning of q4 CEO and the CFO come they come into the room of the management team and they say guys all hands on deck we’re not rich reaching our habitat if we’re not reaching our Abita our interest rate will kick up because we have to hide their debt ratio so a debt ratio with the kind of agreements we have with the bank is going to hit our interest rate and we cannot go below the editor threshold and this is an urgent situation I want a management extra management team meeting in a week normally the marketing thing goes back at best they come back with cost cuts but when knowing both the multiplier per type of campaign the time lag and the standard deviation they can also come back with a very concrete answer and in this case the standard deviation matters a lot so if you only have four to go you have very little time you can’t with a highest standard deviation gamble you need to go for things which have a decent multiplier or slope but with the low standard deviation because you need to have high precision you cannot have two dollars back per dollar spent you need it to be predictable so then the standard deviation is very very critical okay again if you have it much longer if you have two years to go for you can go for the highest multiplier because it’s okay especially if you have high liquidity in your bank account it’s okay to do that it fluctuates like this in performance but as long as the average is very high with a short time here the average can be lower but with the height predictability so a low standard deviation