Yes, I am aware of what Disraeli said about stats. But the beauty on this particular testing protocol is that it has nothing to do with all the other tests that I have seen around in the financial world. This one is coming from what I consider to be the basis of the statistical testing: That means no Candlesticks nor Kondratieff lines or ‘sophisticated’ stuff like that. This is basic stats, and as such it took some time for me to figure out the proper way to test the ever-changing data of markets.
So a bit of context, we took the data for one year for the following Australian shares (AGO, ASB, BHP, BOQ, BPT, CBA, CCL, CFX) those happened to be in our sights so that is what we used. Yes, I guess you could’ve used other shares but we did not.  The data being tested is the difference of the actual value X from the sample average expressed as a percent (%).  By doing it like that we are trying to detect if we are actually going to gain something out of this, meaning that we want the highest percentage of profit that we can get right?
Then we had to decide on the tests, namely what are the situations that we are going to call… Aha!

Test 1,1 – Low price and Low volume
Test 1,2 – Low price and expected volume
Test 1,3 – Low price and high volume
Test 2,1 – Expected price and low volume
Test 2,2 – Expected price and expected volume
Test 2,3 – Expected price and high volume
Test 3,1 – High price and low volume
Test 3,2 – High price and expected volume
Test 3,3 – High price and high volume

The -low, expected and high- buckets are based on whether the last day’s variation falls within the expected variation for that stock for that period of time. I intend to explain the maths behind that calc later, really I do! But for the time being lets just stick with the fact that there is a threshold (or sensibility) for variation and that we then put our last day in one of three buckets – low, mid (expected) or high – Well it is really one of three buckets for price; and one of three buckets for volume. If we were more sophisticated we could use the last hour of trading, or the last minute; but the app at the moment is focusing on the last day as it was a good and readily available measure.

Ok, now that we’ve established the tests and the buckets we can then interpret the results.  You won’t see four of the test in the final hypothesis (2,1 – 2,2 – 2,3 – 1,2) because of two reasons:  1)They had significantly higher values of sample sizes, meaning that they occurred a lot more times which makes sense because 2)They all involve a ‘2’ in the test; which means that they are expected or random occurrences and as such we don’t really want to take or advice to take decisions on randomness. Ok, once it was all a balanced test then we can see that the outcome of the medians test is telling us that there is at least one different median from the sampling.  Which is pretty much all that this basic test will do.  You need to have a bit more thinking on what you are seeing here.

So I can see that definitely test (1,3) is different than (3,3)… we did a one-to-one comparison after but I don’t want to bother with the details.  This means that the point in time when the (1,3 – volume was really low and the price really high) has a different median value than the time when (3,3 both price and volume are very high).  That difference is about 2% from a buy at (1,3) and sell at (3,3). The only change between those results and the advice that we are giving on the app is that there is a better sell position at (3,1) instead of (3,3) because the market could still be going up at (3,3) and we know that it is turning at (3,1) when the volume has dropped and the price has continued to go up building that momentum.

I have struggled a bit with the decision to publish this one. Like I’ve mentioned before, I am often asked to provide proof that what we put out there works, or is working, or is going to work… essentially that it is not ‘snake oil’. Well I can say squarely that 1)Our app is only 4.99 American -so therefore give it a go!; and 2)We have statistical evidence to suggest a difference in medians, among other things. I’ve struggled because I didn’t want to kerfuffle things neither come across as smug. I hope i didn’t do any of those two.