Model Sequences for Optimizing location and duration
As you may already know there are 3 distinct AI model sequences specific to their duration and target, but for this specific strategy we will be discussing on 2 of the 9 total models present: the High Forecast for the front week (Forecast Period: W1), and the High Forecast for the back week (Forecast Period: W2). Let’s say you have a series of long positions given by:
- 100 shares of AAPL
- 100 shares of TSLA
- 100 shares of AMZN
Hypothetical!
Of course this is a hypothetical, please use some sort of optimization process for the number of shares according to your risk tolerance and profit target preference which meets the necessary minimum thresholds of the bid/ask spread, and total cost per trade.
At each 15 minute interval for the RTH session inference at this time (in the future we will take this to 5 minute inferencing, and then 1 minute inference). To Inference is AI terminology for using the model structure to “predict” given a series of current features at that exact moment, using a dataframe or a series of historical values. Our lookback period is approximately 2,000 bars which in this case is almost 4 months of feature engineering which is impossible for any human to do in a hundred years, let alone within seconds.
Let’s get back to the strategy though!
The first 9:45 AM EST inference will set the High Forecast for the week leading up to Friday at 4:00 PM. This is what the model believes the price maximum from the point of inference to exactly Friday at 4:00 PM will be.
By placing your call in that vicinity (preferably above this point), this will be the optimal position because our models do not believe it will exceed this point until expiration. The worst thing for any option trader when hedging a long position is for them to be sitting ITM or “In The Money” when they do not want their stock positions to be taken, and they must close their positions incur costs and potentially roll their position which will incur costs such as the spread and fees, double in this case.
Given your current portfolio you have 3 positions of 100 shares which is great for an option seller because 1 contract is equivalent to 100 shares. The High Forecast for AAPL, MSFT, QQQ, hypothetically is 220, 450, and 500, respectively. In this case what would be optimal in my view would be to sell 1 call option at each of these High Forecasts at the beginning of the week to generate income.
To add sophistication, you can close any of these short call positions as sort of an arbitrage which can reduce risk and increase your delta through certain times, then placing the hedge back when you see it being optimal.
Of course, for full disclosure, in an ideal world the models will always be perfectly accurate, but as you know being a trader, anomalous things happen. This is your duty as a trader to actively adjust positions to suit your given portfolio and situation.
