We use SPY as a proxy for the index. Unless otherwise specified all positions indicated below are assumed to have taken place at the close of the specified day.
A long positions is identified when the closing price is above the simple moving average. No position (staying out of the market or investing in safe heaven assets) is assumed when the moving average is above the closing price. Notice, that the weekly and monthly positions can only change at the end of the corresponding period.
The Period column defines the period for which this position is in-place. For a daily moving average, this is the position hold during the day. It should have been established at the close of the previous trading day, which could have been achieved putting a limited order based on the Trigger Price. For a weekly average, this is date of the first trading day of that week.
The Trigger Price column is the closing price which divides the Long and the None positions. A close at the Trigger Price or higher requires a long position to be established at the end of the period. The percentage quantifies the change in price that must occur over this period for the Trigger Price to be reached. For a weekly moving average, the Trigger Price is meaningful only on Fridays, or at the last closing day of the week if Friday happens to be a holiday.
Using the Trigger Price one can ensure that he is in the correct position as of the close of the day.
The position based on oscillators. To define the positions we use interpretations found on the internet and in the literature, typically backed by own experience from backtesting.
The RSI indicator of length two is a good predictor for short term oversold or overbought conditions. A greater than 90 value indicates an overbought condition and a value less than 10 indicates an oversold condition. No position is assumed in-between.
Larry Connors’s Strategies
We have implemented four strategies from High Probability ETF Trading. These strategies are quite similar in behaviour. They trade infrequently, but have a very high percentage of successful trades on the SPY, as well as on other EFTs. For more details – refer to the book.
ARMA models with various numbers of parameters are fitted over approximately two years of historic data. The best fit is determined based on the AIC statistic. This model is then used to produce a single day prediction. For more details look in the tutorial’s section.