Discover more from Time Series Terminal’s Newsletter
For medium-term asset managers: AAPL may not be helped by NVDA's steam, above the tussles of currencies
We will retrace predictions for Apple
AAPL, and see that it was not helped by Nvidia
NVDA’s steam so far, above all.
To go back in history, add an
asof parameter to the URL, for example,
to search for
AAPL, 6-month ahead horizon right now,
the same search, as of 2023-08-03 end-of-day,
Don’t concatenate “2023-08-03” into “20230803”. It will be treated as the year “20230803”, which is 20 million years away.
Using the 6-month forecast horizon, below are a few dates along the timeline, with the net majority of model votes for the upcoming day, and comments:
2023-03-15 76% of models predict up
2023-04-07 78% of models predict up
NVDA joins as a top causally predictive factor
2023-04-28 77% of models predict up
2023-05-31 72% of models predict up.
NVDA jumped at end of May
2023-06-30. 66% of models predict up
2023-07-31. 67% of models predict up
2023-08-02. 66% of models predict up
USDJPY joins as a top causally predictive factor
2023-08-31. 61% of models predict up
2023-09-08 61% of models predict up
According to tsterm,
NVDA joined as causally predictive factor for
AAPL from the start of April,
USDJPY only from the start of August. From April to August, the predicted “bullishness” for the upcoming day for
AAPL shaved from 78% to 66%, then from August onward, dropped from 66% to 61%.
From the end of May to now,
AAPL was roughly flat. Over this period, if one uses the net majority of model votes as synthetic trading position, with a dwindling position, the person would have achieved flat return but lower volatility, thus higher Sharpe ratio.
It may be the status of being a mega cap stock that has affected
AAPL in such a way. If we retrace the prediction history for another old semiconductor stock Texas Instruments Inc
TXN, the net prediction was always that “above 90% of models predict up” for the upcoming day. Over the last decade,
TXN performed better than the
DJIA index, but less than the mega tech stocks.
Does that mean one shall shift investments onto
NVDA? If we search for
NVDA as prediction target, the net prediction was always “around 25% of models predict up” for the upcoming day. Actually, for the horizon as far as 6 months away, about three quarters of models would predict the conditional distribution of
NVDA to be infinitely wide, thus unable to cast a vote of either “up” or “down” for any day along the forecast horizon.