Original subtitles

However, wrong and welcoming this new with you in this video, we're going to create a trading strategy

using machine learning prediction.

So we have already done our prediction, so we needed to create a trading strategy and this trading

strategy will be very simple.

When we have a positive reach on prediction, we're going to take a bad contract.

So we are going to bits to the increase of stock when we have a negative return prediction.

We're going to take a set contract and then predict the decrease of the stock.

So the really important thing here is that we want the sign of the prediction.

So one or minus one.

And that's really the value.

So to have the same?

We are going to use the same function from Mumbai, and in this function, we put the prediction to

have just the sign of the prediction.

So I will pluck you the result here to a better comprehension.

Then we needed to compute the return of this strategy.

So we need to use the return of the assets, multiply by the position, but here we need to.

Poots also as shift white, because it is exactly the same thing has for the moving average because

if we take, for example, a day in the market open at eight a.m. and close at eight p.m. If you do

your prediction at eight p.m., you cannot compute the return of your strategy by the return from eight

a.m. to eight p.m. of the same day because you do your prediction after this variation.

So it is predict the past by the future because

you will not have all these data when you do a correct prediction.

So you do.

You need Zoe to put a shift to make a prediction at 8:00 p.m. and computes the URL of the strategy by

multiplying this position, this signal by the region of tomorrow.

So then we are going to pluck the cumulative return of our algorithm to see if.

This strategy is profitable on that.

And we need to take only the test, it's because here in the train set, it is logic that we have good

results because the algorithm train its coefficient on this period.

So here we have very bad results, but.

Is not really important that we have bad results, because in the next chapter, we're going to see

some

customization of all approach and we're going to have very good results here.

The main point is to understand.

All the process to create a machine learning algorithm, to create a trading strategy, because if you

don't understand all the process, you cannot understand the next chapter.

And in the next chapter, we need to have some specific algorithm to increase the profitability of our

strategy.

So we need to understand what we have done in this chapter and the process that we have used to create.

That trend sets the test set, etc. Because in the next chapter, we are going to go deeper into the

algorithmic trading thing and the future of engineering.